Upcoming event
Masterclass on Focal therapy for localised prostate cancer
Back

A Prospective Study of the Association between Physical Activity and Risk of Prostate Cancer Defined by Clinical Features and TMPRSS2:ERG

  • Claire H. Pernar 1,
  • Ericka M. Ebot 1,
  • Andreas Pettersson 1,
  • Rebecca E. Graff 1,
  • Francesca Giunchi 2,
  • Thomas U. Ahearn 1,
  • Amparo G. Gonzalez-Feliciano 1,
  • Sarah C. Markt 1,
  • Kathryn M. Wilson 1,
  • Konrad H. Stopsack 3,
  • Elizaveta Gazeeva 4,
  • Rosina T. Lis 5,
  • Giovanni Parmigiani 6,
  • Eric B. Rimm 1,
  • Stephen P. Finn 1,
  • Edward L. Giovannucci 1,
  • Michelangelo Fiorentino 1,
  • Lorelei A. Mucci 1
1 Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA 2 Pathology Unit, Addarii Institute, S. Orsola-Malpighi Hospital, Bologna, Italy 3 Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, NY, USA 4 Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, MA, USA 5 Center for Molecular Oncologic Pathology, Dana-Farber Cancer Institute, Boston, MA, USA 6 Department of Biostatistics and Computational Biology, Dana-Farber Cancer Institute, Boston, MA, USA

Publication: European Urology, October 2018

© 2018 European Association of Urology. Published by Elsevier B.V. All rights reserved.

Background

Growing evidence shows that clinical and molecular subtypes of prostate cancer (PCa) have specific risk factors. Observational studies suggest that physical activity may lower the risk of aggressive PCa. To our knowledge, the association between physical activity and PCa defined by TMPRSS2:ERG has not been evaluated.

Objective

To prospectively examine the association between physical activity and risk of PCa defined by clinical features and TMPRSS2:ERG.

Design, setting, and participants

We studied 49 160 men aged 40–75 yr in the Health Professionals Follow-up Study from 1986 to 2012. Data was collected at baseline and every 2 yr with >90% follow-up. Total and vigorous physical activity were measured in metabolic equivalent of task (MET)-h/wk.

Outcome measures and statistical analysis

Advanced PCa was defined as stage T3b, T4, N1, or M1 at diagnosis and lethal PCa as distant metastases or death due to disease over follow-up. Presence of TMPRSS2:ERG was estimated by immunohistochemistry of ERG protein expression. Cox proportional hazards models were used to obtain multivariable hazard ratios (HRs) and 95% confidence intervals (CIs) for incidence of subtype-specific PCa.

Results and limitations

During 26 yr of follow-up, 6411 developed PCa overall and 888 developed lethal disease. There were no significant associations between total physical activity and risk of PCa in the overall cohort. In multivariable-adjusted models, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (HR: 0.70, 95% CI: 0.53–0.92) and 25% lower risk of lethal PCa (HR: 0.75, 95% CI: 0.59–0.94) than men in the lowest quintile of vigorous activity. The association was independent of screening history. Vigorous activity was not associated with total PCa in the overall cohort but was inversely associated among highly screened men (top vs bottom quintile, HR: 0.83, 95% CI: 0.70–0.97). Of all cases, 945 were assayed for ERG (48% ERG-positive). Men with higher vigorous activity had a lower risk of ERG-positive PCa (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97). There was no significant association with the risk of ERG-negative disease (p heterogeneity = 0.09).

Conclusions

Our study confirms that vigorous physical activity is associated with lower risk of advanced and lethal PCa and provides novel evidence for a lower risk of TMPRSS2:ERG-positive disease.

Patient summary

The identification of modifiable lifestyle factors for prevention of clinically important prostate cancer (PCa) is needed. In this report, we compared risk of PCa in men with different levels of physical activity. Men with higher vigorous activity had a lower risk of developing advanced and lethal PCa and PCa with the common TMPRSS2:ERG gene fusion.

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967

Given the large burden of prostate cancer (PCa) globally, modifiable lifestyle factors that could lower a man's risk of PCa must be identified. Epidemiologic studies of the relationship between physical activity and risk of PCa overall have been mixed but suggest a moderate inverse association [1]. In some epidemiologic studies, men who engaged in higher levels of physical activity had lower risks of developing advanced and fatal PCa [2, 3, 4, 5]. However, other studies have shown no significant association between physical activity and advanced or fatal disease [6, 7, 8]. Physical activity influences a wide range of biological processes, including hormonal, anti-inflammatory, and insulin pathways [9, 10]. These pathways are implicated in the development of aggressive PCa, suggesting a link between physical activity and clinically relevant disease [11].

The integration of PCa characteristics related to not only clinical course [2] but also molecular features [12, 13, 14] in epidemiologic studies may be the key to understanding the relationship between physical activity and PCa risk. The TMPRSS2:ERG gene fusion is the most common PCa molecular subtype [12]. Found in 40–50% of primary PCas, TMPRSS2:ERG results in androgen-regulated expression of the oncogene ERG [15]. Androgens, cellular stress, and insulin-like growth factor (IGF) signaling may have a role in the development and progression of fusion-positive cancers [16, 17, 18]. Although TMPRSS2:ERG does not independently predict biochemical recurrence or lethal disease [19], the fusion may interact with risk factors to influence PCa prognosis [14, 18]. Furthermore, some PCa risk factors, such as low tomato sauce intake and taller height, are more strongly associated with the development of fusion-positive versus fusion-negative PCa [20, 21, 22]. In this study, we hypothesized that the influence of physical activity on hormonal and anti-inflammatory pathways protects against the development of fusion-positive PCa.

The objective of this study was to examine the associations between long-term, pre-diagnostic physical activity among men and risk of developing PCa defined by clinical features (stage, grade, and lethality) and molecular (TMPRSS2:ERG) subtype.

2.1. Study population

The Health Professionals Follow-up Study (HPFS) is an ongoing prospective cohort initiated in 1986 among 51 529 US male health professionals aged 40–75 yr at baseline. Participants completed biennial questionnaires beginning at baseline to ascertain lifestyle, health-related factors, and disease outcomes. Usual diet was assessed every 4 yr using a validated food frequency questionnaire. Follow-up exceeded 90% in each cycle. The study population for this analysis consisted of 49 160 men. We excluded men who reported cancers except non-melanoma skin cancer prior to baseline (n = 2087) and those with missing date of birth (n = 32) or baseline physical activity (n = 250). The study was approved by the Human Subjects Research Committee at the Harvard T.H. Chan School of Public Health.

2.2. Assessment of physical activity

Physical activity was assessed through biennial, validated questionnaires [23] beginning at baseline. Participants selected a category for the average total time/wk engaged in specific activities during the past year: walking or hiking outdoors, jogging, running, bicycling, lap swimming, tennis, squash or racquetball, and calisthenics or rowing. Participants also reported their usual walking pace and the number of flights of stairs climbed daily. Additional specific activities were included on the questionnaire in subsequent cycles: heavy outdoor work from 1988, weightlifting from 1990, moderate outdoor work from 2004, and lower intensity exercise and other aerobic exercise from 2010. Participants indicated the intensity of activity (low, medium, high) for swimming, biking, and tennis from 2010.

To quantify activity intensity, each activity was assigned a metabolic equivalent of task (MET) value based on a compendium of physical activities [24]. A unit of MET is equal to the amount of oxygen uptake required to sit at rest, approximately 3.5 ml/kg/min. A MET-hour is the metabolic equivalent of sitting at rest for 1 h. A measure of MET-h/wk was derived for each activity by multiplying the activity-specific MET value by the participant-reported average number of h/wk. Total activity was defined as the sum of MET-h/wk for each activity. Vigorous activity included activities with a MET value ≥6: jogging, running, bicycling, lap swimming, tennis, squash/racquetball, calisthenics/rowing, and stair climbing.

2.3. Ascertainment of PCa outcomes

Incident PCa was captured by self-report and confirmed through medical records and pathology reports. Information on clinical and treatment history and disease progression was collected through medical records as well as biennial disease-specific questionnaires for development of metastases. Deaths were ascertained through repeated mailings, telephone calls to non-respondents, and searches of the National Death Index. An endpoint committee of physicians confirmed PCa-specific death.

We classified clinical subgroups of PCa as follows: (1) localized PCa: stage T1 or T2 and N0, M0 at diagnosis; (2) advanced PCa: stage T3b, T4, N1, or M1 at diagnosis; and (3) lethal PCa: distant metastases or PCa death over follow-up. PCa cases were also defined as high-grade (Gleason 8–10 and 4 + 3) or low-grade (Gleason 2–6 and 3 + 4). Stage T1a cases (n = 286) were excluded since these cases are incidentally diagnosed and prone to detection bias.

Tumor tissue microarrays were constructed using archival formalin-fixed paraffin-embedded prostate tumor tissue from radical prostatectomy (RP) or transurethral resection of the prostate (TURP) as previously described [19]. Presence or absence of the TMPRSS2:ERG fusion was assessed using a validated immunohistochemistry assay for ERG protein expression. This study included 910 RP and 35 TURP specimens assayed for ERG, diagnosed from 1986 to 2009.

2.4. Statistical analysis

Each participant contributed person-time from date of return of the baseline questionnaire to date of PCa diagnosis, death, or end of follow-up (January 31, 2012). For ERG-defined PCa outcomes, follow-up ended on December 31, 2009 because this was the last year a case assayed for ERG was diagnosed. Cox proportional hazards regression was used to estimate age-adjusted and multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) between physical activity (by quintile) and incidence of total, lethal, advanced, localized, high-grade, and low-grade PCa. An extension of Cox proportional hazards regression was used to model the associations between physical activity and PCa incidence according to ERG subtype [25, 26]. For ERG-specific PCa outcomes, additional analyses of the 910 RP cases assayed for ERG were performed, applying inverse probability weights to account for clinical factors at diagnosis among cases [21]. Tests for heterogeneity of these HRs across quintiles were performed using a likelihood ratio test [26].

To examine long-term activity, we used cumulative average physical activity updated every 2 yr from baseline in 1986 to the time of PCa diagnosis, death, or end of follow-up. The cumulative average physical activity was categorized into quintiles based on the distribution in each questionnaire cycle.

Age- and multivariable-adjusted models included age in months and calendar time. Only multivariable models are presented because the results were similar to age-adjusted models. Multivariable models were additionally adjusted for race, family history of PCa, diabetes, body mass index (BMI), height, smoking status, multivitamin use, and dietary factors. All variables except for race, family history, and BMI at age 21 yr were updated over follow-up. All models of vigorous activity were additionally adjusted for nonvigorous activity. Nonvigorous activity was allowed to vary by ERG subtype in models for ERG-defined PCa.

To account for potential detection bias, we adjusted for having had a prostate-specific antigen (PSA) test, lagged by one cycle to better capture screening rather than diagnostic PSA tests, and PSA testing intensity over time (defined as reporting a PSA test in half or more questionnaire cycles since 1994). To address potential residual confounding by PSA testing, we conducted the analysis among a highly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996 [27].

These analyses were conducted using SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA). All statistical tests were two-sided with an α level of 0.05 to determine statistical significance.

Table 1 shows the age-adjusted characteristics of the 49 160 men at baseline in 1986, according to quintiles of total and vigorous physical activity. The median amount of total activity was 10.2 MET-h/wk and vigorous was 2.2 MET-h/wk. Men in higher quintiles of physical activity tended to be younger, have lower BMI, were more likely to be nonsmokers and use multivitamins, and reported more PSA testing between 1994 and 2010 than men in lower quintiles.

Table 1Age-adjusted characteristics by quintile of total and vigorous physical activity (MET-h/wk) among men in the Health Professionals Follow-up Study at baseline in 1986 unless otherwise specified
Total activity quintileVigorous activity quintile
CharacteristicsQ1Q3Q5Q1Q3Q5
Participants, n980196469935888897349989
Age, mean (SD), yr a54.9 (9.7)54.5 (9.7)53.3 (9.7)57.2 (9.7)54.6 (9.8)51.1 (8.8)
Total activity, mean (SD), MET-h/wk0.8 (0.7)10.2 (2.2)53.7 (36.4)7.6 (12.3)9.0 (10.9)46.8 (40.1)
Vigorous activity, mean (SD), MET-h/wk0.2 (0.4)5.2 (4.3)33.1 (38.4)0.0 (0.0)2.4 (1.3)38.7 (36.0)
PSA screening history
Had PSA test in 1994, %343940353742
No. of biennial questionnaires with PSA test, 1994–20105.25.65.65.25.65.8
PSA test on at least half of all questionnaires, 1994–2010, %626968626870
Family history of prostate cancer, %111212111212
Diabetes, %4.32.82.34.23.32.1
Caucasian, %959696959695
Current smokers, %159.46.9159.54.9
Multivitamin use, %384446374449
Height, mean (SD), inches70.0 (2.9)70.1 (2.8)70.2 (2.9)70.1 (3.0)70.1 (2.9)70.1 (2.8)
BMI at age 21 yr, mean (SD), kg/m222.9 (3.2)23.0 (3.0)23.2 (2.9)23.0 (3.3)23.0 (3.0)23.1 (2.7)
BMI, mean (SD), kg/m226.2 (3.8)25.5 (3.3)24.8 (3.0)26.2 (3.7)25.7 (3.4)24.7 (2.8)
Dietary & nutrient intakes, mean (SD)
 Total calories, kcal/d1936 (621)1969 (609)2053 (635)1954 (626)2002 (627)2011 (613)
 Calcium, mg/d861 (427)900 (419)926 (434)862 (429)899 (417)946 (449)
 α-linolenic acid, g/d1.1 (0.4)1.1 (0.4)1.0 (0.3)1.1 (0.4)1.1 (0.4)1.0 (0.3)
 Supplemental vitamin E, mg/d31.5 (79.1)38.4 (84.2)44.8 (91.7)32.2 (80.5)37.7 (83.1)48.8 (94.6)
 Tomato sauce, servings/wk0.9 (1.3)0.9 (1.1)1.0 (1.3)0.9 (1.2)1.0 (1.2)1.0 (1.2)
 Alcohol, g/d10.8 (16.3)11.1 (14.9)12.3 (15.5)11.2 (16.8)10.7 (14.6)11.4 (14.4)
 Coffee, cups/d2.0 (1.9)1.9 (1.8)1.8 (1.7)2.0 (1.9)1.9 (1.8)1.8 (1.7)
View Table in HTML

BMI = body mass index; PSA = prostate-specific antigen.

aVariable not adjusted for age.

Between 1986 and 2012, 6411 men were diagnosed with incident PCa (Supplementary Table 1) including 603 with advanced and 888 with lethal disease. Of 945 cases assayed for ERG, 449 (48%) had ERG-positive disease.

Table 2, Table 3 show results from multivariable-adjusted models for the associations of total activity and vigorous activity with the risk of PCa defined by clinical features in the total cohort and in the highly screened subcohort. There was no association between total activity and risk of PCa overall or of any clinical subgroup in either cohort. In contrast, men in the highest quintile of vigorous activity had a significant 30% lower risk of advanced PCa (top vs bottom quintile, HR: 0.70; 95% CI: 0.53–0.92; p trend = 0.04) and a 25% lower risk of lethal PCa (top vs bottom quintile, HR: 0.75; 95% CI: 0.59–0.94; p trend = 0.04) than men in the lowest quintile in the total cohort. Additionally, there was a borderline significant 16% lower risk of high-grade PCa in the highest than in the lowest quintile of vigorous activity (HR: 0.84; 95% CI: 0.70–1.01). Vigorous activity was not significantly associated with the risk of overall, localized, or low-grade PCa in multivariable-adjusted models in the total cohort. After restricting to highly screened men, however, vigorous activity was associated with a 16–18% lower risk of total, localized, and low-grade PCa.

Table 2Hazard ratios and 95% confidence intervals for the association of total physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professional Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Total activity quintile, HR (95% CI)Total activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases127012751252316361354
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.98 (0.91–1.07)0.81 (Ref)0.96 (0.82–1.12)0.92 (0.79–1.09)0.2
Lethal prostate cancer
 No. incident cases190174166253128
 Multivariable c1 (Ref)0.95 (0.77–1.18)0.95 (0.76–1.18)0.51 (Ref)1.04 (0.60–1.82)1.12 (0.63–1.98)0.9
Advanced prostate cancer
 No. incident cases123107117171520
 Multivariable c1 (Ref)0.91 (0.70–1.18)1.00 (0.77–1.30)0.91 (Ref)0.74 (0.36–1.53)0.89 (0.44–1.79)0.7
Localized prostate cancer
 No. incident cases882971911253297285
 Multivariable c1 (Ref)1.06 (0.97–1.17)0.99 (0.90–1.09)0.81 (Ref)0.97 (0.82–1.16)0.92 (0.77–1.10)0.2
High-grade prostate cancer
 No. incident cases251269280627277
 Multivariable c1 (Ref)1.09 (0.92–1.30)1.14 (0.96–1.37)0.21 (Ref)1.08 (0.76–1.53)1.15 (0.80–1.63)0.8
Low-grade prostate cancer
 No. incident cases728809778207252237
 Multivariable c1 (Ref)1.05 (0.95–1.16)1.00 (0.90–1.11)11 (Ref)0.99 (0.82–1.20)0.92 (0.75–1.11)0.2
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.
Table 3Hazard ratios and 95% confidence intervals for the association of vigorous physical activity (MET-h/wk) and risk of prostate cancer overall and by clinical subgroup a in the total Health Professionals Follow-up Study cohort with follow-up from 1986 to 2012 and in the highly screened subcohort with follow-up from 1996 to 2012
Total cohort, 1986–2012Highly screened subcohort, 1996–2012 b
Vigorous activity quintile, HR (95% CI)Vigorous activity quintile, HR (95% CI)
Q1Q3Q5ptrendQ1Q3Q5ptrend
Total prostate cancer
 No. incident cases140513191129337375327
 Multivariable c1 (Ref)1.00 (0.92–1.08)0.95 (0.88–1.04)0.31 (Ref)0.97 (0.83–1.13)0.83 (0.70–0.97)0.02
Lethal prostate cancer
 No. incident cases248182109373923
 Multivariable c1 (Ref)0.90 (0.74–1.09)0.75 (0.59–0.94)0.041 (Ref)1.01 (0.62–1.63)0.82 (0.47–1.44)0.8
Advanced prostate cancer
 No. incident cases16711680171115
 Multivariable c1 (Ref)0.82 (0.64–1.05)0.70 (0.53–0.92)0.041 (Ref)0.54 (0.24–1.19)0.73 (0.34–1.57)0.4
Localized prostate cancer
 No. incident cases975982864269307263
 Multivariable c1 (Ref)1.04 (0.94–1.13)0.99 (0.89–1.09)0.71 (Ref)0.98 (0.83–1.16)0.82 (0.68–0.98)0.04
High-grade prostate cancer
 No. incident cases341292232828961
 Multivariable c1 (Ref)0.94 (0.80–1.10)0.84 (0.70–1.01)0.31 (Ref)0.99 (0.72–1.35)0.70 (0.49–1.00)0.13
Low-grade prostate cancer
 No. incident cases764787743212242224
 Multivariable c1 (Ref)1.03 (0.93–1.14)1.01 (0.91–1.13)0.81 (Ref)0.97 (0.80–1.18)0.84 (0.69–1.03)0.03
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aLethal prostate cancer: distant metastasis or death due to the disease; advanced prostate cancer: stage T3b, T4, N1, or M1 at diagnosis; localized prostate cancer: stage T1 or T2 and N0, M0 at diagnosis; high-grade: Gleason 8–10 and 4 + 3; low-grade: Gleason 2–6 and 3 + 4.
bHighly screened subcohort of 13 859 men who reported having a PSA test on the 1994 and 1996 questionnaires, with follow-up starting in 1996.
cMultivariable models in the total cohort are adjusted for age (mo), calendar time (mo), non-vigorous activity (quintiles), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 yr (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/wk; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles); multivariable models in highly screened cohort are adjusted for those listed except for height, BMI at age 21 yr, tomato sauce, α-linolenic acid, and having a PSA test in previous cycle.

Table 4 presents associations of total and vigorous activity with risk of ERG-defined PCa. As with clinical features, we did not observe significant associations between total activity and PCa risk of either ERG subtype. However, for vigorous activity, men in the highest quintile had a significant 29% lower risk of ERG-positive PCa than men in the lowest quintile (top vs bottom quintile, HR: 0.71, 95% CI: 0.52–0.97; p trend = 0.04). There was no significant association between vigorous activity and risk of ERG-negative PCa (p heterogeneity = 0.09). After restricting to the highly screened subcohort, the association between vigorous activity and ERG-positive disease persisted, and there was a suggestive inverse association between total activity and ERG-positive disease (Supplementary Table 2). The association between vigorous activity and ERG-positive PCa was similar in magnitude when restricting to RP cases and when using inverse probability weighting to account for potential differences among cases with and without tissue biomarker data (data not shown).

Table 4Hazard ratios and 95% confidence intervals for the association of total and vigorous physical activity (MET-h/wk) quintiles and risk of ERG-positive and ERG-negative prostate cancer in the Health Professionals Follow-up Study with follow-up from 1986 to 2009
ERG-negativeERG-positive
No. of casesMultivariable a HR (95% CI)No. of casesMultivariable a HR (95% CI)pheterogeneity
Total activity quintile0.2c
 Q1971.00 (ref)691.00 (ref)
 Q2830.84 (0.62–1.12)841.19 (0.86–1.64)
 Q31131.12 (0.85–1.48)1021.45 (1.06–1.97)
 Q4970.95 (0.71–1.26)1111.52 (1.12–2.06)
 Q51061.04 (0.78–1.38)831.13 (0.82–1.57)
 ptrend0.60.61d
Vigorous activity quintile b0.5c
 Q1981.00 (ref)1031.00 (ref)
 Q2950.99 (0.74–1.32)890.84 (0.63–1.13)
 Q31061.08 (0.82–1.44)940.89 (0.67–1.19)
 Q4970.98 (0.73–1.30)860.78 (0.58–1.05)
 Q51001.05 (0.78–1.40)770.71 (0.52–0.97)
 Ptrend0.80.040.09d
View Table in HTML

CI = confidence interval; HR = hazard ratio; PSA = prostate-specific antigen.

aMultivariable models adjusted for age (mo), calendar time (mo), height (in; ≤68, >68 to 70, >70 to 72, >72), race (Caucasian or other), family history of prostate cancer in father or brother (yes or no), BMI at age 21 (kg/m2; ≤20, 21 to <23, 23 to <25, ≥25), intensity of prostate-specific antigen (PSA) testing (yes or no), having a PSA test in previous cycle (yes or no), smoking (never, former/quit >10 yr ago, former/quit ≤10 yr ago, or current), diabetes (yes or no), current BMI (kg/m2; ≤22, 23 to <25, 25 to <27.5, ≥27.5), multivitamin use (yes or no), intake total calories (quintiles), calcium (quintiles), tomato sauce (servings/week; <0.25, 0.25 to <1.0, 1.0 to <2.0, ≥2), α-linolenic acid (quintiles), alcohol (quintiles), coffee (cups/d; 0, >0 to 1, >1 to 2, >2 to 3, >3), and vitamin E supplements (quintiles).
bMultivariable models with vigorous activity (quintiles) are additionally adjusted for non-vigorous activity (quintiles).
cBased on a likelihood ratio test with four degrees of freedom using quintiles as the exposure.
dBased on a likelihood ratio test with one degree of freedom using continuous trend variable as the exposure.

Our study found that vigorous physical activity over the long term is associated with a lower risk of clinically meaningful endpoints, including advanced, lethal, and TMPRSS2:ERG-positive PCa. These findings suggest that regularly engaging in higher levels of vigorous activity may be beneficial to men for prevention of clinically important PCa. Furthermore, these results suggest that physical activity may act through pathways related to development of TMPRSS2:ERG and support the hypothesis that this subtype has unique etiological factors.

Author contributions: Claire H. Pernar had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Pernar, Mucci.

Acquisition of data: Mucci, Giunchi, Lis, Finn, Fiorentino.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Pernar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pernar, Ebot, Parmigiani.

Obtaining funding: Mucci.

Administrative, technical, or material support: None.

Supervision: Mucci.

Other: None.

Financial disclosures: Claire H. Pernar certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

Funding/Support and role of the sponsor: This work was supported by the National Institutes of Health (T32 ES 007069 to C.H.P., R25 CA 112355 to R.E.G., T32 CA 09001 to C.H.P. and S.C.M., P50 CA 090381 to L.A.M. and S.C.M., R01 CA 136578 to L.A.M., 4P30CA006516-51 to L.A.M. and G.P.); the Prostate Cancer Foundation Young Investigator Awards to L.A.M., K.M.W., K.H.S. and S.P.F.; the World Cancer Research Fund (2013/1003 to S.P.F. and L.A.M.). The Health Professionals Follow-up Study is supported by U01 CA 167552 from the National Cancer Institute. The TMAs were constructed by the Tissue Microarray Core Facility at the Dana-Farber/Harvard Cancer Center (P30 CA 06516).

Acknowledgments: We would like to thank the participants and staff of the HPFS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. In particular, we would like to acknowledge Elizabeth Frost-Hawes, Siobhan Saint-Surin, Robert Sheahan, Ann Fisher, and Scott Smith.

  1. [1]Liu, Y., Hu, F., Li, D. et al. Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis. Eur Urol. 2011; 60: 1029–1044
  2. [2]Giovannucci, E., Liu, Y., Platz, E.A., Stampfer, M.J., and Willett, W.C. Risk factors for prostate cancer incidence and progression in the health professionals follow-up study. Int J Cancer. 2007; 121: 1571–1578
  3. [3]Patel, A.V., Rodriguez, C., Jacobs, E.J., Solomon, L., Thun, M.J., and Calle, E.E. Recreational physical activity and risk of prostate cancer in a large cohort of U.S. men. Cancer Epidemiol Biomarkers Prev. 2005; 14: 275–279
  4. [4]Johnsen, N.F., Tjonneland, A., Thomsen, B.L. et al. Physical activity and risk of prostate cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Int J Cancer. 2009; 125: 902–908
  5. [5]Orsini, N., Bellocco, R., Bottai, M. et al. A prospective study of lifetime physical activity and prostate cancer incidence and mortality. Br J Cancer. 2009; 101: 1932–1938
  6. [6]Moore, S.C., Peters, T.M., Ahn, J. et al. Physical activity in relation to total, advanced, and fatal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2008; 17: 2458–2466
  7. [7]Littman, A.J., Kristal, A.R., and White, E. Recreational physical activity and prostate cancer risk (United States). Cancer Causes Control. 2006; 17: 831–841
  8. [8]Hrafnkelsdottir, S.M., Torfadottir, J.E., Aspelund, T. et al. Physical activity from early adulthood and risk of prostate cancer: a 24-year follow-up study among Icelandic men. Cancer Prev Res (Phila). 2015; 8: 905–911
  9. [9]Koelwyn, G.J., Quail, D.F., Zhang, X., White, R.M., and Jones, L.W. Exercise-dependent regulation of the tumour microenvironment. Nat Rev Cancer. 2017; 17: 620–632
  10. [10]Thomas, R.J., Kenfield, S.A., and Jimenez, A. Exercise-induced biochemical changes and their potential influence on cancer: a scientific review. Br J Sports Med. 2017; 51: 640–644
  11. [11]Liao, Y., Abel, U., Grobholz, R. et al. Up-regulation of insulin-like growth factor axis components in human primary prostate cancer correlates with tumor grade. Hum Pathol. 2005; 36: 1186–1196
  12. [12]Cancer Genome Atlas Research N. The molecular taxonomy of primary prostate cancer. Cell. 2015; 163: 1011–1025
  13. [13]Setlur, S.R., Mertz, K.D., Hoshida, Y. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst. 2008; 100: 815–825
  14. [14]Ahearn, T.U., Pettersson, A., Ebot, E.M. et al. A prospective investigation of PTEN loss and ERG expression in lethal prostate cancer. J Natl Cancer Inst. 2016; 108: djv346
  15. [15]Tomlins, S.A., Rhodes, D.R., Perner, S. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005; 310: 644–648
  16. [16]Mani, R.S., Amin, M.A., Li, X. et al. Inflammation-induced oxidative stress mediates gene fusion formation in prostate cancer. Cell Rep. 2016; 17: 2620–2631
  17. [17]Haffner, M.C., Aryee, M.J., Toubaji, A. et al. Androgen-induced TOP2B-mediated double-strand breaks and prostate cancer gene rearrangements. Nat Genet. 2010; 42: 668–675
  18. [18]Pettersson, A., Lis, R.T., Meisner, A. et al. Modification of the association between obesity and lethal prostate cancer by TMPRSS2:ERG. J Natl Cancer Inst. 2013; 105: 1881–1890
  19. [19]Pettersson, A., Graff, R.E., Bauer, S.R. et al. The TMPRSS2:ERG rearrangement, ERG expression, and prostate cancer outcomes: a cohort study and meta-analysis. Cancer Epidemiol Biomarkers Prev. 2012; 21: 1497–1509
  20. [20]Graff, R.E., Ahearn, T.U., Pettersson, A. et al. Height, obesity, and the risk of TMPRSS2:ERG-defined prostate cancer. Cancer Epidemiol Biomarkers Prev. 2018; 27: 193–200
  21. [21]Graff, R.E., Pettersson, A., Lis, R.T. et al. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. Am J Clin Nutr. 2016; 103: 851–860
  22. [22]Egbers, L., Luedeke, M., Rinckleb, A. et al. Obesity and prostate cancer risk according to tumor TMPRSS2:ERG gene fusion status. Am J Epidemiol. 2015; 181: 706–713
  23. [23]Chasan-Taber, S., Rimm, E.B., Stampfer, M.J. et al. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996; 7: 81–86
  24. [24]Ainsworth, B.E., Haskell, W.L., Herrmann, S.D. et al. Compendium of physical activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011; 43: 1575–1581
  25. [25]Lunn, M. and McNeil, D. Applying Cox regression to competing risks. Biometrics. 1995; 51: 524–532
  26. [26]Wang, M., Spiegelman, D., Kuchiba, A. et al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016; 35: 782–800
  27. [27]Siddiqui, M.M., Wilson, K.M., Epstein, M.M. et al. Vasectomy and risk of aggressive prostate cancer: a 24-year follow-up study. J Clin Oncol. 2014; 32: 3033–3038
  28. [28]USDHHS. (https://health.gov/paguidelines)2008 Physical Activity Guidelines for Americans. 2008;
  29. [29]Giovannucci, E.L., Liu, Y., Leitzmann, M.F., Stampfer, M.J., and Willett, W.C. A prospective study of physical activity and incident and fatal prostate cancer. Arch Intern Med. 2005; 165: 1005–1010
  30. [30]Mani, R.S. and Chinnaiyan, A.M. Triggers for genomic rearrangements: insights into genomic, cellular and environmental influences. Nat Rev Genet. 2010; 11: 819–829
  31. [31]McTiernan, A. Mechanisms linking physical activity with cancer. Nat Rev Cancer. 2008; 8: 205–211
  32. [32]Van Blarigan, E.L., Gerstenberger, J.P., Kenfield, S.A. et al. Physical activity and prostate tumor vessel morphology: data from the Health Professionals Follow-up Study. Cancer Prev Res (Phila). 2015; 8: 962–967
Dr. Guillaume Ploussard

Physical activity influences several processes, such as hormonal, insulin, anti-inflammatory pathways. Epidemiology suggests that physical activity may lower the risk of aggressive prostate cancer. Nevertheless, studies remain contradictory to definitely link fatal disease and physical activity. Potential correlations between activity and molecular characteristics of prostate cancer have not been thoroughly assessed.

In the present study, the authors used the Health Professionnals Follow-up Study (HPFS) to examine the associations between long-term, pre-diagnostic physical activity and the risk of developing prostate cancer. The HPFS is an ongoing cohort initiated in 1986 among 51 529 American male health professionals aged 40-75 years at baseline. Physical activity was assessed by biennial questionnaires and quantified by the sum of metabolic equivalent of task (MET). A measure of MET-hour per week was derived for each activity. Prostate cancer incidence as well as disease progression were captured self-report ad confirmed through medical records and pathology reports. The TMPRSS2:ERG status was evaluated on tissue microarrays constructed from 910 radical prostatectomy and 35 transurethral resection specimens. To examine long-term activity, cumulative average physical activity was categorized into quintiles from baseline to the time of prostate cancer diagnosis. Several adjustments on potentially confounding factors were performed into multivariable analyses.

Between 1986 and 2012, 6411 men were diagnosed with prostate cancer including 603 advanced and 888 lethal disease. The TMPRSS2:ERG status was positive in 48% of cases. There was no association between total activity and risk of prostate cancer (overall or any clinical subgroup). However, for vigorous activity, men in the highest quintile had a significant 30% lower risk of advanced disease (95% CI: 0.53-0.92) and a 25% lower risk of lethal disease (95% CI 0.59-0.94) than men in the lowest quintile in the total cohort.

Similar findings were reported when assessing the TMPRSS2:ERG status. In the overall population, no clear association was noted. Nevertheless, there was a trend for lower risk of ERG-positive disease in the subgroup of patients with vigorous activity. Potential benefit from vigorous activity was more marked after restricting to the highly screened subcohort.

This prospective cohort analysis suggests that vigorous physical activity is inversely correlated with the risk of advanced, lethal prostate cancer. Interestingly, correlations between physical activity and clinical behavior of prostate cancer are, for the first time, supported by molecular aspects and by the TMPRSS2:ERG status.

In a highly screened cohort, promotion of physical activity could positively influence the risk of any grade and aggressive prostate cancer, and may be beneficial at a population level. As a modifiable risk factor, this could be an important goal for public health intervention. 

Tags: Biomarkers