Background
Knowledge of significant prostate (sPCa) locations being missed with magnetic resonance (MR)- and transrectal ultrasound (TRUS)-guided biopsy (Bx) may help to improve these techniques.
Objective
To identify the location of sPCa lesions being missed with MR- and TRUS-Bx.
Design, setting, and participants
In a referral center, 223 consecutive Bx-naive men with elevated prostate specific antigen level and/or abnormal digital rectal examination were included. Histopathologically-proven cancer locations, Gleason score, and tumor length were determined.
Intervention
All patients underwent multi-parametric MRI and 12-core systematic TRUS-Bx. MR-Bx was performed in all patients with suspicion of PCa on multi-parametric MRI (n = 142).
Outcome measurements and statistical analysis
Cancer locations were compared between MR- and TRUS-Bx. Proportions were expressed as percentages, and the corresponding 95% confidence intervals were calculated.
Results and limitations
In total, 191 lesions were found in 108 patients with sPCa. From these lesion 74% (141/191) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions and 61% (86/141) with TRUS-Bx. TRUS-Bx detected more lesions compared with MR-Bx (140 vs 109). However, these lesions were often low risk (39%). Significant lesions missed with MR-Bx most often had involvement of dorsolateral (58%) and apical (37%) segments and missed segments with TRUS-Bx were located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%).
Conclusions
Both techniques have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx with involvement of the anterior part (79%).
Patient summary
Both biopsy techniques miss cancer in specific locations within the prostate. Identification of these lesions may help to improve these techniques.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Early detection of prostate cancer (PCa), with the intent to diagnose this disease in a curable state, currently leads to overdiagnosis which in turn can result in overtreatment [1]. Overdiagnosis is the main disadvantage of the present clinical standard where men with elevated serum prostate-specific antigen (PSA) levels and/or abnormal digital rectal examination undergo random systematic 10- to 12-core prostate transrectal ultrasound (TRUS)-guided biopsy (Bx). Other limitations of TRUS-Bx are underdetection and undergrading of clinically significant prostate cancer (sPCa) [2]. The number of cores and location for systematic sampling is a topic of debate [2], [3], and [4]. In Bx-naïve patients, generally accepted schemes for systematic sampling with TRUS-Bx are limited to the posterior peripheral zone, since most cancers are located in this region [2] and [4]. In the repeat Bx setting, most cancers are located in the anterior apex of the prostate [2], [4], and [5].
Multi-parametric MR imaging (mpMRI) can be used to reduce overdiagnosis and subsequent overtreatment [6]. mpMRI is able to detect sPCa (44–87%) in Bx-naïve mem and men with prior negative Bx. Furthermore, the high negative predictive value (63–98%) of mpMRI can be used to rule out sPCa [7], [8], and [9].
Although mpMRI is a promising technique, sPCa may be missed [10]. The interpretation of mpMRI can sometimes be difficult [11] and [12]. Current research has not focused on the location of cancers missed with mpMRI followed by MR-Bx. Knowledge of lesions being missed with MR- and TRUS-Bx is of clinical importance for diagnosis and treatment.
The purpose of this study is to identify the location of sPCa lesions being missed with MR-Bx and systematic TRUS-Bx in Bx-naïve patients at risk for PCa.
This study was a retrospective analysis of stored MR-images of Bx-naïve consecutive patients with an elevated PSA level and/or abnormal digital rectal examination. All participants were enrolled by referral from urologists from July 2012 through January 2013 and were included in a previous prospectively study. However, no detailed data were published on the location of PCa [13].
All patients underwent mpMRI performed at 3T (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany) according to the European Society of Urogenital Radiology prostate MR-guidelines [14]. mpMRI scans were scored independently by three readers (1 yr, 1 yr, and 19 yr of experience, respectively) using the Prostate Imaging Reporting and Data System version 1.0 (PI-RADS): from 1 (low) to 5 (high) according to the likelihood of sPCa being present [14]. Disagreements in PI-RADS were resolved by consensus.
Both MR- and TRUS-Bx were performed in a second visit (Fig. 1). The standard diagnostic pathway consisted of systematic TRUS-Bx only for men who had a PI-RADS 1 or 2 reported on mpMRI. MR-Bx was performed in all patients with a PI-RADS 3–5 lesion on mpMRI, followed by systematic TRUS-Bx as well [15]. Each PI-RADS 3–5 lesions on mpMRI was biopsied using two to three cores. Seminal vesicles were sampled if suspicious for tumor invasion.
Fig. 1
Flowchart describing the diagnostic pathways. All patients underwent both multi-parametric magnetic resonance imaging (MRI) and transrectal ultrasound (TRUS)-guided biopsy. Patients with a Prostate Imaging Reporting and Data System (PI-RADS) ≥3 on multi-parametric MRI received MR image-guided biopsy in addition to TRUS biopsy.
During the MR-Bx session an axial diffusion weighted image (DWI) was acquired to relocate the cancer suspicious lesions (CSL) described on the diagnostic mpMRI. After needle insertion with MR-Bx two confirmation scans were acquired to verify the Bx location. TRUS-Bx was performed within 30 min after MR-Bx by a urologist blinded to the mpMRI and MR-Bx procedures. Any lesions seen on TRUS were targeted using the core for the relevant prostate zone.
All Bx specimens underwent evaluation by a urogenital-histopathologist blinded to the Bx strategy. Low-risk PCa was defined as either low-volume Gleason score (GS) 3 + 3 or 3 + 4 [9]. sPCa (= intermediate- and/or high-risk PCa) on MR- and TRUS-Bx was defined as a GS ≥4 + 3 or GS 3 + 4 (> 4 mm, or >1 positive core), or a GS 3 + 3 with a core length >6 mm. Furthermore, a GS 3 + 3 was defined as sPCa when more than one or two cores were positive on MR- and TRUS-Bx, respectively [13]. Thus, for lesions with a GS 3 + 3 the number of cores was modality specific.
Results were described on a segment, lesion, and patient level.
Fig. 2
Distribution maps of cancer positive segments in patients with significant prostate cancer on magnetic resonance image-guided biopsy (MR-Bx; left) and transrectal ultrasound-guided biopsy (TRUS-Bx; right). The anterior portion of the prostate was defined by a hypothetical line drawn 17 mm anterior from the posterior prostatic surface irrespective of prostate size, which represents the core length commonly used during TRUS-Bx [17].
SV = seminal vesicles.
The highest GS was determined on a patient and lesion level by combining histopathology from MR- and TRUS-Bx.
One-way analysis of variance was performed to determine significant differences in age, PSA, or prostate volume. The proportions of cancer-positive segments were expressed as percentages, and the corresponding 95% confidence intervals (CIs) were calculated. The Fisher's exact test was performed to determine significant differences between groups. The Wilcoxon-signed rank test was used to determine significant differences between the MCCL and TCCL found with MR- and TRUS-guided Bx. Since patients served as their own control we used McNemar's test to detect a significant difference between the association of MR- and TRUS-Bx and outcome. A significant difference was considered when the p value was <0.05. The STARD (Standards for Reporting of Diagnostic Accuracy Studies) guidelines were followed to report the results. All analyses were performed with SPSS software (Version 20.0. Armonk, NY: IBM Corp.).
Patient characteristics and histopathological outcome of the 223 included patients are shown in Table 1 and Table 2 and reported previously [13].
Table 1
Patient characteristics of all included patients in whom cancer was not detected, and patients in which prostate cancer (PCa) was detected with magnetic resonance image-guided biopsy (MR-Bx) only (Group 1), transrectal ultrasound-guided biopsy (TRUS-Bx) only (Group 2), and both modalities (Group 3). The p value of one-way analysis of variance is given for the comparison between Groups 1–3
| All patients | Significant PCa detected with: | |||||
|---|---|---|---|---|---|---|
| Included patients | Negative on MR- and TRUS-Bx | MR-Bx only (Group 1) | TRUS-Bx only (Group 2) | Both modalities positive (Group 3) |
p value | |
| No. of patients (%) | 223 (100) | 81 (36.3) | 29 (13.0) | 15 (6.7) | 64 (28.7) | NA |
| Mean age, yr (SD) | 62.4 (7.4) | 59.6 (7.2) | 63.5 (7.0) | 62.7 (5.8) | 65.5 (6.7) | 0.2 |
| Mean PSA, ng/ml (SD) | 5.9 (3.1) | 5.1 (2.2) | 6.7 (4.5) | 4.9 (1.0) | 7.1 (3.5) | 0.1 |
| Mean prostate volume, cc (SD) | 46.8 (22.6) | 56.2 (24.4) | 40.0 (17.1) | 40.0 (10.1) | 39.0 (19.9) | 1.0 |
PSA = prostate specific antigen; SD = standard deviation.
Table 2
Histopathological biopsy outcome on a per patient level for MR- and TRUS-Bx for the total cohort of 223 men. Green shading indicates patients where magnetic resonance image-guided biopsy (MR-Bx) upgraded the Gleason score in relation to transrectal ultrasound-guided biopsy (TRUS-Bx). Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 High volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 81 | 23 | 8 | 3 | 4 | 5 | 124 | |
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 1 | 1 | 5 | ||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 3 | 2 | 2 | 1 | 2 | 10 | ||
| Gleason 3 + 4 high volume | 8 | 4 | 3 | 2 | 16 | 7 | 40 | ||
| Gleason ≥4 + 3 | 5 | 2 | 4 | 32 | 43 | ||||
| Totals | 97 | 31 | 16 | 7 | 27 | 45 | 223 | ||
PCa = prostate cancer.
Combined histopathology from MR- and TRUS-Bx revealed PCa in 64% of the patients (142/223, 95% CI: 57–70%) and 48% (108/223, 95% CI: 42–55%) of the cases showed sPCa. MR- and TRUS-Bx detected sPCa in 42% (93/223) and 35% (79/223) of the patients, respectively, and this was statistically significant different (p < 0.0001). Low-risk cancer was significantly (p < 0.0001) more often detected with TRUS-Bx (21%, 47/223) compared with MR-Bx (3%, 6/223). In patients diagnosed with sPCa on MR- and TRUS-Bx upgrading of the GS was seen in 32% (35/108) with MR-Bx and in 22% (24/108) with TRUS-Bx, this was statistically significant different (p < 0.001).
In 15 patients sPCa was detected with TRUS-Bx only (three PI-RADS 3, seven PI-RADS 4–5, and in five patients no CSLs were seen on mpMRI). Retrospectively, in six out of 10 patients with a PI-RADS ≥3 the MR-Bx needle was not in the correct position and in one patient the images of the MR-Bx were not stored and thus not available for re-evaluation.
Anterior tumor involvement was seen in 41% (44/108, 95% CI: 31–50%) of the patients with sPCa.
In total 191 lesions were found in 108 patients with sPCa (Table 3). From these lesions 74% (141/191, 95% CI: 68–80%) were defined as sPCa on either MR- or TRUS-Bx. MR-Bx detected 74% (105/141) of these lesions, and this was 61% (86/141) with TRUS-Bx (13% difference, 95% CI: 8–19%). In 37% (64/175, 95% CI: 30–44%) of the CSLs seen on mpMRI no PCa was found with MR-Bx. Reasons for failure of MRI to identify cancer lesions are shown in Table 4.
Table 3
Highest Gleason score for all lesions (n = 191) detected with magnetic resonance image-guided biopsy (MR-Bx) and transrectal ultrasound-guided biopsy (TRUS-Bx) in patients with significant cancer (n = 108). Green shading indicates patients where MR-Bx upgraded the Gleason score in relation to TRUS-Bx. Analogous blue shading indicates patients where TRUS-Bx upgraded the Gleason score in relation to MR-Bx. A dark color represents a larger discrepancy between the two modalities
| TRUS-Bx | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| No cancer | Low-risk PCa | Significant PCa | |||||||
| Gleason 3 + 3 low volume | Gleason 3 + 4 low volume | Gleason 3 + 3 high volume | Gleason 3 + 4 high volume | Gleason ≥4 + 3 | Totals | ||||
| MR-Bx | No cancer | 36 | 12 | 5 | 14 | 15 | 82 | ||
| Low-risk PCa | Gleason 3 + 3 low volume | 1 | 1 | 1 | 3 | ||||
| Gleason 3 + 4 low volume | 1 | 1 | |||||||
| Significant PCa | Gleason 3 + 3 high volume | 7 | 1 | 1 | 4 | 13 | |||
| Gleason 3 + 4 High volume | 24 | 3 | 1 | 1 | 9 | 9 | 47 | ||
| Gleason ≥4 + 3 | 18 | 2 | 25 | 45 | |||||
| Totals | 51 | 40 | 14 | 7 | 29 | 50 | 191 | ||
PCa = prostate cancer.
Table 4
Reasons for failure of MRI to identify cancer lesions
| Situation | Low-risk | Significant PCa | All lesions |
|---|---|---|---|
| Lesions detected with TRUS-Bx in all patients | 93 | 86 | 179 |
| Lesions detected with MR-Bx in all patients | 10 | 105 | 115 |
| negative mpMRI but positive TRUS-Bx (likely reading failure) |
81 | 33 | 114 |
| positive mpMRI but negative MR-Bx and negative TRUS-Bx (likely reading failure) |
56 | ||
| positive mpMRI and positive TRUS-Bx but negative MR-Bx (likely sampling failure). |
5 | 3 | 8 |
PCa = prostate cancer. TRUS-Bx = transrectal ultrasound-guided biopsy. MR-Bx = magnetic resonance image-guided biopsy. mpMRI = multi-parametric MRI.
In 15 patients in which sPCa was detected with TRUS-Bx only, 26 lesions were detected of which 50% (13/26, 95% CI: 31–69%) contained sPCa. In retrospect (with knowledge of the cancer locations from TRUS-Bx) eight lesions could be defined as significant on mpMRI. In three of these sPCa lesions, low-risk cancer was diagnosed with MR-Bx and five sPCa lesions were not seen on the initial detection MRI.
TRUS-Bx detected more lesions (73%, 140/191) compared to MR-Bx (57%, 109/191; 16% difference, 95%CI: 12–21%). However, more lesions were low-risk PCa (28%, 54/191) compared with MR-Bx (2%, 4/191; 26% difference, 95% CI: 20–32%).
The MCCL and TCCL of lesions solitary detected with TRUS-Bx were significantly (p < 0.0001) smaller than MR-Bx biopsy lesions (Table 5). In patients diagnosed with sPCa on MR- and/or TRUS-Bx upgrading of the GS was seen in 41% (58/141) with MR-Bx and in 35% (49/141) with TRUS-Bx (6% difference, 95% CI: 3–11%).
Table 5
Differences between mean maximum cancer core length (MCCL) and total cancer core length (TCCL) in patients with significant cancer. The p-value of the Wilcoxon-signed rank test is provided
| Situation | MR-Bx | TRUS-Bx | p-value (MR-Bx vs TRUS-Bx) |
|---|---|---|---|
| All lesions | 109 | 140 | NA |
| Number of missed lesions | 82 | 51 | NA |
| Number of missed significant lesions | 34 | 49 | NA |
| Mean MCCL (mm) of lesions detected with both modalities (SD): | 10.0 (4.6) | 9.5 (4.7) | 0.64 |
| Mean TCCL (mm) of lesions detected with both modalities (SD): | 19.6 (12.3) | 24.2 (21.1) | 0.17 |
| Mean MCCL (mm) of missed lesions and thus solitary detected with the other modality (SD): | 3.4 (3.4) | 8.6 (4.2) | <0.0001 |
| Mean TCCL (mm) of missed lesions and thus solitary detected with the other modality (SD) | 6.6 (8.6) | 16.9 (10.2) | <0.0001 |
MR-Bx = MR-guided biopsy; TRUS-Bx = TRUS-guided biopsy.
Anterior tumor involvement was seen in 31% (44/140, 95% CI: 24–39%) of the sPCa.
MR- and TRUS-Bx identified 277 and 347 cancer-positive segments respectively in patients with sPCa. The distribution of these segments is shown in Figure 2. With MR- and TRUS-Bx, most cancer-positive segments were located in the midprostate; 53% (147/277, 95% CI: 47–59%) and 37% (128/347, 95% CI: 32–42%), respectively. In total, 39% (109/277, 95% CI: 34–45%) of MR-guided positive segments were located in the anterior part of the prostate.
In lesions with sPCa, MR- and TRUS-Bx missed 97 and 111 segments respectively (Fig. 3). Lesions missed with MR-Bx, which had sPCa detected with TRUS-Bx, appear to be located in the apex (37%, 36/97, 95% CI: 28–47%) and dorsolateral segments (58%, 56/97, 95% CI: 47–67%). Missed sPCa lesions on TRUS-Bx most often had involvement of anterior segments (79%, 88/111, 95% CI: 71–86%). Specifically: anterior midprostate (50%, 56/111, 95% CI: 41–60%) and anterior apex (23%, 25/111, 95% CI: 15–31%).
This paper extrapolates on a previously published study, but focuses on lesions missed with MR- and TRUS-Bx. sPCa lesions missed with MR-Bx most often had involvement of the dorsolateral (58%) and apical (37%) segments. With TRUS-Bx, missed sPCa lesions had involvement of segments located anteriorly (79%), anterior midprostate (50%), and anterior apex (23%). Anterior tumor involvement was seen in 41% of the patients, which is consistent with the literature [18]. Lesions localized in the apical region were difficult to detect with both modalities, which is also concordant with the literature [18], [19], [20], and [21].
sPCa is associated with increased age, PSA level, and a small prostate volume [22]. Although not statistically significant, the patient characteristics in our study (Table 1) confirm these relations, but the PSA levels in patients with sPCa detected with TRUS-Bx (Group 2) did not confirm this relation.
mpMRI evaluation can be difficult in the apex given the small size of this region and its location at the margin of the prostate [12]. Also, some focal lesions might be un-noticed on standard DWI protocols due to signal received from surrounding benign prostatic tissue which overshadows the lesion [11]. This may be overcome by using high b-values (1400–2000 s/mm2) [11] and [16]. In our diagnostic protocol we used a calculated b-value of 1400 s/mm2 which may not be sufficient. Furthermore, from the 67 lesions seen on mpMRI which were negative on MR-Bx only three lesions contained sPCa on TRUS-Bx, which suggests these CSLs were false-positive on mpMRI rather than being missed with MR-Bx.
The relatively high involvement of dorsolateral segments in sPCa lesions missed with MR-Bx might be explained by subcapsular tumors [12]. Critical evaluation of small areas with abnormal signal intensity on DWI and dynamic contrast enhanced images in the absence of a readily apparent mass abutting the capsule is recommended [12]. Furthermore, since the exact needle location is not known with TRUS-Bx some lateral biopsy samples can be false-positive. This may occur in cases where the Bx core is located in both the medial and lateral segments but the tumor is confined within the medial segment.
Ukimura et al [2] recommends sampling according to a medial sextant pattern with ≥4 cores from the lateral peripheral zone. Furthermore, this review advises obtaining samples from the anterior apex, anterior lateral horn, and anterior transition zone in the repeat Bx setting [2]. The results found in our study confirm these recommendations.
To our knowledge this is the first study which describes the differences between MR- and TRUS-Bx regarding cancer location in Bx-naïve men at risk for PCa. The results have clinical implications for both radiologists and urologists. For example, to optimize Bx sites for TRUS-Bx and systematic sampling with MRI targeted TRUS-biopsy [23]. Based on our observation it is reasonable to obtain additional samples at initial TRUS-Bx from the anterior apex and anterior midprostate since these are the most common sites where cancer is missed. Adding cores in the extreme anterior apex has been shown to be beneficial [24]. Similarly, it might be reasonable to obtain additional random Bx from dorsolateral and apical regions during mpMRI targeted TRUS-Bx to improve detection of sPCa without the need for systematic 12-core Bx [25].
Regarding costs and patients comfort TRUS-MRI fusion is an interesting alternative for MR-guided Bx [26]. Although there is no study comparing mpMRI targeted TRUS-Bx and in-bore MR-Bx, the visual feedback regarding the accuracy of needle placement makes the in-bore MR-guided biopsy, in our opinion, more accurate.
In our study only 18% of the patients with a PI-RADS 3 lesion on mpMRI had sPCa. This figure was 89% in patients with a PI-RADS 4–5 lesion. Although considerable efforts have been undertaken to determine the optimal threshold for biopsy, this is still a topic of debate [27]. Based on the findings from our study it might be interesting to investigate a region-dependent threshold for biopsy. For example, a threshold of PI-RADS 3 for lesions in the apical and dorsolateral regions and a threshold of PI-RADS 4–5 for lesions elsewhere in the prostate.
Some limitations of this study should be addressed. Firstly, there is no universally accepted and validated definition for sPCa [28]. The modality-specific definition used in our study make the results difficult to compare with other studies. Furthermore, in the case in which a patient has multiple low-risk lesions, this definition can diagnose a patient as having sPCa without having a discrete sPCa lesion. However, this definition was created to allow comparison of two to four MR-Bx cores with 12-cores obtained with TRUS-Bx.
Secondly, whole mount prostatectomy would be the most reliable tool for histopathologic evaluation but is ethically impossible in patients with no proven cancer on biopsy. An alternative approach is to use 5-mm transperineal template prostate mapping [4]. The need for general anesthesia and increased complication rate makes this a highly invasive technique and presents challenges in recruiting patients for clinical trials when less invasive alternatives are available [4].
Thirdly, a subanalysis of patients who underwent prostatectomy would provide an excellent reference standard; however, this may have introduced selection bias. By comparing the Bx results between MR- and TRUS-Bx we were able to directly compare differences between these modalities in a large patient group.
Both MR- and TRUS-Bx have difficulties in detecting apical lesions. MR-Bx most often missed cancer with involvement of the dorsolateral part (58%) and TRUS-Bx missed lesions with involvement of the anterior part (79%).
Author contributions: Martijn Gerjan Schouten 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: Schouten, van der Leest, Futterer, Barentsz.
Acquisition of data: Schouten, van der Leest, Pokorny, Thompson, Barentsz, Futterer.
Analysis and interpretation of data: Schouten, van der Leest, Futterer.
Drafting of the manuscript: van der Leest, Schouten, Pokorny, Hoogenboom, Barentsz, Thompson, Futterer.
Critical revision of the manuscript for important intellectual content: Barentsz, Futterer, Thompson.
Statistical analysis: Schouten, Futterer.
Obtaining funding: Futterer, Barentsz.
Administrative, technical, or material support: Thompson, Barentsz.
Supervision: Barentsz, Futterer.
Other: None.
Financial disclosures: Martijn Gerjan Schouten 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: The Dutch Cancer Society.
Study demonstrated that systematic and MRI-targeted biopsy miss clinically significant prostate cancer in different areas of the prostate gland. Systematic TRUS biopsy most commonly missed csPCa in anterior prostate, while MRI-targeted biopsy in dorsolateral area of the gland, whereas both techniques have difficulties in detecting apical lesions. These results may help in biopsy planning.