Prevention of Bone Metastases in Patients with High-risk Nonmetastatic Prostate Cancer Treated with Zoledronic Acid: Efficacy and Safety Results of the Zometa European Study (ZEUS)

ZA has long-term benefits in patients with CRPC and BMs. It is approved for prevention of skeletal-related events in metastatic genitourinary cancers and recommended for routine use in CRPC according to the European Association of Urology guidelines [16] . ZA was the subject of a metastasis prevention trial in nonmetastatic CRCP patients that was closed early because the data safety monitoring board halted the study after accrual of 398 of 991 planned patients, since the event rate was lower than expected [17] . Oral clodronate trials showed improved OS in men with metastatic PCa starting ADT but no evidence of effect in men with nonmetastatic PCa [6] . Smith et al. demonstrated that denosumab significantly increased BM-free survival by a median of 4.2 mo compared with placebo and also significantly delayed ttBMs [18] .

The additive anticancer effect of ZA combined with endocrine therapy in patients with early stage breast cancer was reported in the AZURE trial. Although the primary end point of disease-free survival was not met in the overall study population, subgroup analyses showed significant benefits for disease-free survival and OS in women ≥5 yr after menopause [19] . A significant anticancer effect of the addition of ZA on disease-free survival in premenopausal women with endocrine-responsive breast cancer taking anastrozole or tamoxifen was reported in the ABCSG-12 trial [20] .

In our study there was no difference in the occurrence of BMs between treatment groups. At 4 yr, the local urologist/radiologist diagnosed BMs in 17.1% of the patients in the ZAG and 17.0% in the CG with a difference between proportions of 0.1% (95% CI, −4.4 to 4.7) in favour of the CG. When looking at the lower limit of the 95% CI, the difference in favour of ZA was 4.4% at the highest point, which was lower than the assumed difference of at least 6% used for sample size calculation. A maximum risk reduction of 4.4% is equivalent to treating 23 patients with ZA to prevent BMs in 1 patient at 4 yr. When taking into account the associated patient burden of treatment with ZA (eg, AEs, frequent visits to the hospital for intravenous infusions, costs), the potential benefit of ZA was insufficient and clinically irrelevant. The central review results led to similar conclusions. When specifically focussing on the subgroup receiving ADT as in the breast cancer trials, no differences could be detected. The ZA dosage may be important because in the AZURE trial, patients received more intense ZA treatment: 4 mg 3 to 4-weekly ×6, every 3 mo ×8, and every 6 mo ×5 thereafter [19] . In the ABCSG-12 trial, however, a less intense ZA administration schedule was used: 4 mg every 6 mo for 3 yr [20] . In our study, ZA was administered every 3 mo for ≤4 yr, suggesting that frequency of ZA administration may not be the only reason for failure to prevent BMs. However, the main mechanisms of the prevention of BMs by ZA in breast cancer may be different in PCa. If the BM-preventive effect of ZA in PCa is not exerted in the bone where sufficient drug levels are maintained up to 240 d but is based on its effect on the mevalonate pathway and the immune system (effects likely to be dose dependent), we cannot exclude the possibility that ZA could prevent BMs when used more intensively than the 3-mo regimen in our study [11], [12], [13], and [21]. The ZOOM trial data support this view because bone markers (N-terminal telopeptide) remained low in the patients treated monthly but increased in patients treated every 3 mo, suggesting that accumulation of ZA in the bone did not ensure extended bone turnover suppression [22] .

BMs were assessed according to standard procedures where BSs were performed when symptoms appeared or PSA increased. BS hot spots are sensitive but not specific for metastases, and they are indistinguishable from other causes such as fractures or inflammation/infection without confirmative imaging. Unfortunately, only 53% of the patients with a positive BS had confirmative imaging, which reflects real-life practice patterns as a limitation of our study.

For a more accurate understanding of the number of BMs, the protocol was amended so that BIP images could be retrospectively collected for independent and blinded expert central review (F.L.). Participating centres were not originally selected based on their ability to generate BIP images for central review. Consequently, we collected paired BIPs from only 612 patients, another study limitation. Remarkably, more patients in the ZAG had equivocal findings compared with the CG, specifically because PSA categories mainly showed incongruously low PSA levels. It was likely that ZA also affected bone, which was reflected in bone/muscle pain or arthralgia. Because most of the equivocal BIP results were derived from not-for-cause BIPs, it is unlikely that symptoms could explain the higher rate of equivocal BIP results in the ZAG.

Another limitation of our study was our heterogeneous population. Patients treated locally with curative intent had fewer BMs compared with those who were not. Similarly, patients receiving early ADT were potentially more prone to developing BMs.

Another possible limitation of our study was the open-label design resulting in patients in the ZAG attending clinic more frequently compared with the CG, which may have resulted in differential outcomes reporting. In our study, more AEs were reported in the ZAG, which may be biased by the more frequent visit schedule. However, according to the known side-effect profile of ZA, more AEs can be expected. A more frequent visit schedule could lead to more BIPs with more patients diagnosed as metastatic in the ZAG compared with the CG. As shown in Supplemental Table 2, the number of BIPs was comparable in both arms. Thus such a bias seems unlikely.

A multivariate Cox proportional-hazards analysis showed that geographic region was an independent risk factor for BMs. Patients in Sweden, Norway, and the Netherlands had double the likelihood to develop BMs compared with Germany. This equated with the effect of a Gleason score of 8–10 or node-positive disease. Although we may speculate that genetic or other high-risk factors (eg, a higher positive margin rate) could explain the differences in the occurrence of BMs between countries, the reasons for these findings are presently unclear.

A major AE of BPs is ONJ, which occurs in 1–7% of patients who have undergone dental extraction while receiving BPs [23] and [24]. The frequency of ONJ increased when more risk factors such as corticosteroids, radiotherapy, and chemotherapy were present. Dental extractions or poor dental hygiene are other prerequisites for ONJ development [14], [23], and [24]. Once this information was published, these patients were excluded, and procedures to improve dental hygiene were recommended prior to ZA administration. After this amendment, no additional patients reported ONJ in our study.

Prostate cancer (PCa) is characterised by its propensity for bone metastases (BMs) that occur in >80% of patients with advanced PCa [1] . BMs result in morbidity and are associated with pain, pathologic fractures, and decreased survival. The standard treatment of patients with metastases is androgen deprivation therapy (ADT) using surgical or medical castration. The 5-yr survival rate of patients with metastatic or locally advanced disease subjected to ADT is 35% [2] . If considerable numbers of PCa patients receive prolonged ADT, decreased bone mineral density, increased risk for osteoporosis, and skeletal fractures can ensue [2] and [3].

Accumulating clinical evidence has confirmed the value of bisphosphonates (BPs) in BMs, preventing ADT-related bone loss, reducing morbidity and pain, and improving survival in patients with castration-resistant PCa (CRPC) [4], [5], [6], and [7]. Zoledronic acid (ZA; Zometa, Novartis, Basel, Switzerland), developed as a third-generation nitrogen-containing BP with increased potency relative to other BPs, induces direct inhibition of PCa cells in vitro, activation of cytotoxic T cells toward a range of malignant cell types, and inhibition of tumour-mediated angiogenesis that may render the bone microenvironment less conducive to micrometastatic growth [8], [9], [10], and [11].

We investigated the efficacy of 3-mo ZA for BM prevention in high-risk nonmetastatic PCa patients. A 3-mo interval was chosen for practical and financial reasons. In addition, data indicated that ZA has a long bone retention time, with approximately 50% of the peak dose persisting there 240 d after administration, and pharmacokinetics demonstrating a multiphasic plasma disposition with an initial half-life of 0.2–1.4 h, followed by 39–4526 h for subsequent elimination, indicating that ZA levels may persist for 3 mo [12] and [13].

ZA has long-term benefits in patients with CRPC and BMs. It is approved for prevention of skeletal-related events in metastatic genitourinary cancers and recommended for routine use in CRPC according to the European Association of Urology guidelines [16] . ZA was the subject of a metastasis prevention trial in nonmetastatic CRCP patients that was closed early because the data safety monitoring board halted the study after accrual of 398 of 991 planned patients, since the event rate was lower than expected [17] . Oral clodronate trials showed improved OS in men with metastatic PCa starting ADT but no evidence of effect in men with nonmetastatic PCa [6] . Smith et al. demonstrated that denosumab significantly increased BM-free survival by a median of 4.2 mo compared with placebo and also significantly delayed ttBMs [18] .

The additive anticancer effect of ZA combined with endocrine therapy in patients with early stage breast cancer was reported in the AZURE trial. Although the primary end point of disease-free survival was not met in the overall study population, subgroup analyses showed significant benefits for disease-free survival and OS in women ≥5 yr after menopause [19] . A significant anticancer effect of the addition of ZA on disease-free survival in premenopausal women with endocrine-responsive breast cancer taking anastrozole or tamoxifen was reported in the ABCSG-12 trial [20] .

In our study there was no difference in the occurrence of BMs between treatment groups. At 4 yr, the local urologist/radiologist diagnosed BMs in 17.1% of the patients in the ZAG and 17.0% in the CG with a difference between proportions of 0.1% (95% CI, −4.4 to 4.7) in favour of the CG. When looking at the lower limit of the 95% CI, the difference in favour of ZA was 4.4% at the highest point, which was lower than the assumed difference of at least 6% used for sample size calculation. A maximum risk reduction of 4.4% is equivalent to treating 23 patients with ZA to prevent BMs in 1 patient at 4 yr. When taking into account the associated patient burden of treatment with ZA (eg, AEs, frequent visits to the hospital for intravenous infusions, costs), the potential benefit of ZA was insufficient and clinically irrelevant. The central review results led to similar conclusions. When specifically focussing on the subgroup receiving ADT as in the breast cancer trials, no differences could be detected. The ZA dosage may be important because in the AZURE trial, patients received more intense ZA treatment: 4 mg 3 to 4-weekly ×6, every 3 mo ×8, and every 6 mo ×5 thereafter [19] . In the ABCSG-12 trial, however, a less intense ZA administration schedule was used: 4 mg every 6 mo for 3 yr [20] . In our study, ZA was administered every 3 mo for ≤4 yr, suggesting that frequency of ZA administration may not be the only reason for failure to prevent BMs. However, the main mechanisms of the prevention of BMs by ZA in breast cancer may be different in PCa. If the BM-preventive effect of ZA in PCa is not exerted in the bone where sufficient drug levels are maintained up to 240 d but is based on its effect on the mevalonate pathway and the immune system (effects likely to be dose dependent), we cannot exclude the possibility that ZA could prevent BMs when used more intensively than the 3-mo regimen in our study [11], [12], [13], and [21]. The ZOOM trial data support this view because bone markers (N-terminal telopeptide) remained low in the patients treated monthly but increased in patients treated every 3 mo, suggesting that accumulation of ZA in the bone did not ensure extended bone turnover suppression [22] .

BMs were assessed according to standard procedures where BSs were performed when symptoms appeared or PSA increased. BS hot spots are sensitive but not specific for metastases, and they are indistinguishable from other causes such as fractures or inflammation/infection without confirmative imaging. Unfortunately, only 53% of the patients with a positive BS had confirmative imaging, which reflects real-life practice patterns as a limitation of our study.

For a more accurate understanding of the number of BMs, the protocol was amended so that BIP images could be retrospectively collected for independent and blinded expert central review (F.L.). Participating centres were not originally selected based on their ability to generate BIP images for central review. Consequently, we collected paired BIPs from only 612 patients, another study limitation. Remarkably, more patients in the ZAG had equivocal findings compared with the CG, specifically because PSA categories mainly showed incongruously low PSA levels. It was likely that ZA also affected bone, which was reflected in bone/muscle pain or arthralgia. Because most of the equivocal BIP results were derived from not-for-cause BIPs, it is unlikely that symptoms could explain the higher rate of equivocal BIP results in the ZAG.

Another limitation of our study was our heterogeneous population. Patients treated locally with curative intent had fewer BMs compared with those who were not. Similarly, patients receiving early ADT were potentially more prone to developing BMs.

Another possible limitation of our study was the open-label design resulting in patients in the ZAG attending clinic more frequently compared with the CG, which may have resulted in differential outcomes reporting. In our study, more AEs were reported in the ZAG, which may be biased by the more frequent visit schedule. However, according to the known side-effect profile of ZA, more AEs can be expected. A more frequent visit schedule could lead to more BIPs with more patients diagnosed as metastatic in the ZAG compared with the CG. As shown in Supplemental Table 2, the number of BIPs was comparable in both arms. Thus such a bias seems unlikely.

A multivariate Cox proportional-hazards analysis showed that geographic region was an independent risk factor for BMs. Patients in Sweden, Norway, and the Netherlands had double the likelihood to develop BMs compared with Germany. This equated with the effect of a Gleason score of 8–10 or node-positive disease. Although we may speculate that genetic or other high-risk factors (eg, a higher positive margin rate) could explain the differences in the occurrence of BMs between countries, the reasons for these findings are presently unclear.

A major AE of BPs is ONJ, which occurs in 1–7% of patients who have undergone dental extraction while receiving BPs [23] and [24]. The frequency of ONJ increased when more risk factors such as corticosteroids, radiotherapy, and chemotherapy were present. Dental extractions or poor dental hygiene are other prerequisites for ONJ development [14], [23], and [24]. Once this information was published, these patients were excluded, and procedures to improve dental hygiene were recommended prior to ZA administration. After this amendment, no additional patients reported ONJ in our study.

Prostate cancer (PCa) is characterised by its propensity for bone metastases (BMs) that occur in >80% of patients with advanced PCa [1] . BMs result in morbidity and are associated with pain, pathologic fractures, and decreased survival. The standard treatment of patients with metastases is androgen deprivation therapy (ADT) using surgical or medical castration. The 5-yr survival rate of patients with metastatic or locally advanced disease subjected to ADT is 35% [2] . If considerable numbers of PCa patients receive prolonged ADT, decreased bone mineral density, increased risk for osteoporosis, and skeletal fractures can ensue [2] and [3].

Accumulating clinical evidence has confirmed the value of bisphosphonates (BPs) in BMs, preventing ADT-related bone loss, reducing morbidity and pain, and improving survival in patients with castration-resistant PCa (CRPC) [4], [5], [6], and [7]. Zoledronic acid (ZA; Zometa, Novartis, Basel, Switzerland), developed as a third-generation nitrogen-containing BP with increased potency relative to other BPs, induces direct inhibition of PCa cells in vitro, activation of cytotoxic T cells toward a range of malignant cell types, and inhibition of tumour-mediated angiogenesis that may render the bone microenvironment less conducive to micrometastatic growth [8], [9], [10], and [11].

We investigated the efficacy of 3-mo ZA for BM prevention in high-risk nonmetastatic PCa patients. A 3-mo interval was chosen for practical and financial reasons. In addition, data indicated that ZA has a long bone retention time, with approximately 50% of the peak dose persisting there 240 d after administration, and pharmacokinetics demonstrating a multiphasic plasma disposition with an initial half-life of 0.2–1.4 h, followed by 39–4526 h for subsequent elimination, indicating that ZA levels may persist for 3 mo [12] and [13].

ZA has long-term benefits in patients with CRPC and BMs. It is approved for prevention of skeletal-related events in metastatic genitourinary cancers and recommended for routine use in CRPC according to the European Association of Urology guidelines [16] . ZA was the subject of a metastasis prevention trial in nonmetastatic CRCP patients that was closed early because the data safety monitoring board halted the study after accrual of 398 of 991 planned patients, since the event rate was lower than expected [17] . Oral clodronate trials showed improved OS in men with metastatic PCa starting ADT but no evidence of effect in men with nonmetastatic PCa [6] . Smith et al. demonstrated that denosumab significantly increased BM-free survival by a median of 4.2 mo compared with placebo and also significantly delayed ttBMs [18] .

The additive anticancer effect of ZA combined with endocrine therapy in patients with early stage breast cancer was reported in the AZURE trial. Although the primary end point of disease-free survival was not met in the overall study population, subgroup analyses showed significant benefits for disease-free survival and OS in women ≥5 yr after menopause [19] . A significant anticancer effect of the addition of ZA on disease-free survival in premenopausal women with endocrine-responsive breast cancer taking anastrozole or tamoxifen was reported in the ABCSG-12 trial [20] .

In our study there was no difference in the occurrence of BMs between treatment groups. At 4 yr, the local urologist/radiologist diagnosed BMs in 17.1% of the patients in the ZAG and 17.0% in the CG with a difference between proportions of 0.1% (95% CI, −4.4 to 4.7) in favour of the CG. When looking at the lower limit of the 95% CI, the difference in favour of ZA was 4.4% at the highest point, which was lower than the assumed difference of at least 6% used for sample size calculation. A maximum risk reduction of 4.4% is equivalent to treating 23 patients with ZA to prevent BMs in 1 patient at 4 yr. When taking into account the associated patient burden of treatment with ZA (eg, AEs, frequent visits to the hospital for intravenous infusions, costs), the potential benefit of ZA was insufficient and clinically irrelevant. The central review results led to similar conclusions. When specifically focussing on the subgroup receiving ADT as in the breast cancer trials, no differences could be detected. The ZA dosage may be important because in the AZURE trial, patients received more intense ZA treatment: 4 mg 3 to 4-weekly ×6, every 3 mo ×8, and every 6 mo ×5 thereafter [19] . In the ABCSG-12 trial, however, a less intense ZA administration schedule was used: 4 mg every 6 mo for 3 yr [20] . In our study, ZA was administered every 3 mo for ≤4 yr, suggesting that frequency of ZA administration may not be the only reason for failure to prevent BMs. However, the main mechanisms of the prevention of BMs by ZA in breast cancer may be different in PCa. If the BM-preventive effect of ZA in PCa is not exerted in the bone where sufficient drug levels are maintained up to 240 d but is based on its effect on the mevalonate pathway and the immune system (effects likely to be dose dependent), we cannot exclude the possibility that ZA could prevent BMs when used more intensively than the 3-mo regimen in our study [11], [12], [13], and [21]. The ZOOM trial data support this view because bone markers (N-terminal telopeptide) remained low in the patients treated monthly but increased in patients treated every 3 mo, suggesting that accumulation of ZA in the bone did not ensure extended bone turnover suppression [22] .

BMs were assessed according to standard procedures where BSs were performed when symptoms appeared or PSA increased. BS hot spots are sensitive but not specific for metastases, and they are indistinguishable from other causes such as fractures or inflammation/infection without confirmative imaging. Unfortunately, only 53% of the patients with a positive BS had confirmative imaging, which reflects real-life practice patterns as a limitation of our study.

For a more accurate understanding of the number of BMs, the protocol was amended so that BIP images could be retrospectively collected for independent and blinded expert central review (F.L.). Participating centres were not originally selected based on their ability to generate BIP images for central review. Consequently, we collected paired BIPs from only 612 patients, another study limitation. Remarkably, more patients in the ZAG had equivocal findings compared with the CG, specifically because PSA categories mainly showed incongruously low PSA levels. It was likely that ZA also affected bone, which was reflected in bone/muscle pain or arthralgia. Because most of the equivocal BIP results were derived from not-for-cause BIPs, it is unlikely that symptoms could explain the higher rate of equivocal BIP results in the ZAG.

Another limitation of our study was our heterogeneous population. Patients treated locally with curative intent had fewer BMs compared with those who were not. Similarly, patients receiving early ADT were potentially more prone to developing BMs.

Another possible limitation of our study was the open-label design resulting in patients in the ZAG attending clinic more frequently compared with the CG, which may have resulted in differential outcomes reporting. In our study, more AEs were reported in the ZAG, which may be biased by the more frequent visit schedule. However, according to the known side-effect profile of ZA, more AEs can be expected. A more frequent visit schedule could lead to more BIPs with more patients diagnosed as metastatic in the ZAG compared with the CG. As shown in Supplemental Table 2, the number of BIPs was comparable in both arms. Thus such a bias seems unlikely.

A multivariate Cox proportional-hazards analysis showed that geographic region was an independent risk factor for BMs. Patients in Sweden, Norway, and the Netherlands had double the likelihood to develop BMs compared with Germany. This equated with the effect of a Gleason score of 8–10 or node-positive disease. Although we may speculate that genetic or other high-risk factors (eg, a higher positive margin rate) could explain the differences in the occurrence of BMs between countries, the reasons for these findings are presently unclear.

A major AE of BPs is ONJ, which occurs in 1–7% of patients who have undergone dental extraction while receiving BPs [23] and [24]. The frequency of ONJ increased when more risk factors such as corticosteroids, radiotherapy, and chemotherapy were present. Dental extractions or poor dental hygiene are other prerequisites for ONJ development [14], [23], and [24]. Once this information was published, these patients were excluded, and procedures to improve dental hygiene were recommended prior to ZA administration. After this amendment, no additional patients reported ONJ in our study.

Prostate cancer (PCa) is characterised by its propensity for bone metastases (BMs) that occur in >80% of patients with advanced PCa [1] . BMs result in morbidity and are associated with pain, pathologic fractures, and decreased survival. The standard treatment of patients with metastases is androgen deprivation therapy (ADT) using surgical or medical castration. The 5-yr survival rate of patients with metastatic or locally advanced disease subjected to ADT is 35% [2] . If considerable numbers of PCa patients receive prolonged ADT, decreased bone mineral density, increased risk for osteoporosis, and skeletal fractures can ensue [2] and [3].

Accumulating clinical evidence has confirmed the value of bisphosphonates (BPs) in BMs, preventing ADT-related bone loss, reducing morbidity and pain, and improving survival in patients with castration-resistant PCa (CRPC) [4], [5], [6], and [7]. Zoledronic acid (ZA; Zometa, Novartis, Basel, Switzerland), developed as a third-generation nitrogen-containing BP with increased potency relative to other BPs, induces direct inhibition of PCa cells in vitro, activation of cytotoxic T cells toward a range of malignant cell types, and inhibition of tumour-mediated angiogenesis that may render the bone microenvironment less conducive to micrometastatic growth [8], [9], [10], and [11].

We investigated the efficacy of 3-mo ZA for BM prevention in high-risk nonmetastatic PCa patients. A 3-mo interval was chosen for practical and financial reasons. In addition, data indicated that ZA has a long bone retention time, with approximately 50% of the peak dose persisting there 240 d after administration, and pharmacokinetics demonstrating a multiphasic plasma disposition with an initial half-life of 0.2–1.4 h, followed by 39–4526 h for subsequent elimination, indicating that ZA levels may persist for 3 mo [12] and [13].

ZA has long-term benefits in patients with CRPC and BMs. It is approved for prevention of skeletal-related events in metastatic genitourinary cancers and recommended for routine use in CRPC according to the European Association of Urology guidelines [16] . ZA was the subject of a metastasis prevention trial in nonmetastatic CRCP patients that was closed early because the data safety monitoring board halted the study after accrual of 398 of 991 planned patients, since the event rate was lower than expected [17] . Oral clodronate trials showed improved OS in men with metastatic PCa starting ADT but no evidence of effect in men with nonmetastatic PCa [6] . Smith et al. demonstrated that denosumab significantly increased BM-free survival by a median of 4.2 mo compared with placebo and also significantly delayed ttBMs [18] .

The additive anticancer effect of ZA combined with endocrine therapy in patients with early stage breast cancer was reported in the AZURE trial. Although the primary end point of disease-free survival was not met in the overall study population, subgroup analyses showed significant benefits for disease-free survival and OS in women ≥5 yr after menopause [19] . A significant anticancer effect of the addition of ZA on disease-free survival in premenopausal women with endocrine-responsive breast cancer taking anastrozole or tamoxifen was reported in the ABCSG-12 trial [20] .

In our study there was no difference in the occurrence of BMs between treatment groups. At 4 yr, the local urologist/radiologist diagnosed BMs in 17.1% of the patients in the ZAG and 17.0% in the CG with a difference between proportions of 0.1% (95% CI, −4.4 to 4.7) in favour of the CG. When looking at the lower limit of the 95% CI, the difference in favour of ZA was 4.4% at the highest point, which was lower than the assumed difference of at least 6% used for sample size calculation. A maximum risk reduction of 4.4% is equivalent to treating 23 patients with ZA to prevent BMs in 1 patient at 4 yr. When taking into account the associated patient burden of treatment with ZA (eg, AEs, frequent visits to the hospital for intravenous infusions, costs), the potential benefit of ZA was insufficient and clinically irrelevant. The central review results led to similar conclusions. When specifically focussing on the subgroup receiving ADT as in the breast cancer trials, no differences could be detected. The ZA dosage may be important because in the AZURE trial, patients received more intense ZA treatment: 4 mg 3 to 4-weekly ×6, every 3 mo ×8, and every 6 mo ×5 thereafter [19] . In the ABCSG-12 trial, however, a less intense ZA administration schedule was used: 4 mg every 6 mo for 3 yr [20] . In our study, ZA was administered every 3 mo for ≤4 yr, suggesting that frequency of ZA administration may not be the only reason for failure to prevent BMs. However, the main mechanisms of the prevention of BMs by ZA in breast cancer may be different in PCa. If the BM-preventive effect of ZA in PCa is not exerted in the bone where sufficient drug levels are maintained up to 240 d but is based on its effect on the mevalonate pathway and the immune system (effects likely to be dose dependent), we cannot exclude the possibility that ZA could prevent BMs when used more intensively than the 3-mo regimen in our study [11], [12], [13], and [21]. The ZOOM trial data support this view because bone markers (N-terminal telopeptide) remained low in the patients treated monthly but increased in patients treated every 3 mo, suggesting that accumulation of ZA in the bone did not ensure extended bone turnover suppression [22] .

BMs were assessed according to standard procedures where BSs were performed when symptoms appeared or PSA increased. BS hot spots are sensitive but not specific for metastases, and they are indistinguishable from other causes such as fractures or inflammation/infection without confirmative imaging. Unfortunately, only 53% of the patients with a positive BS had confirmative imaging, which reflects real-life practice patterns as a limitation of our study.

For a more accurate understanding of the number of BMs, the protocol was amended so that BIP images could be retrospectively collected for independent and blinded expert central review (F.L.). Participating centres were not originally selected based on their ability to generate BIP images for central review. Consequently, we collected paired BIPs from only 612 patients, another study limitation. Remarkably, more patients in the ZAG had equivocal findings compared with the CG, specifically because PSA categories mainly showed incongruously low PSA levels. It was likely that ZA also affected bone, which was reflected in bone/muscle pain or arthralgia. Because most of the equivocal BIP results were derived from not-for-cause BIPs, it is unlikely that symptoms could explain the higher rate of equivocal BIP results in the ZAG.

Another limitation of our study was our heterogeneous population. Patients treated locally with curative intent had fewer BMs compared with those who were not. Similarly, patients receiving early ADT were potentially more prone to developing BMs.

Another possible limitation of our study was the open-label design resulting in patients in the ZAG attending clinic more frequently compared with the CG, which may have resulted in differential outcomes reporting. In our study, more AEs were reported in the ZAG, which may be biased by the more frequent visit schedule. However, according to the known side-effect profile of ZA, more AEs can be expected. A more frequent visit schedule could lead to more BIPs with more patients diagnosed as metastatic in the ZAG compared with the CG. As shown in Supplemental Table 2, the number of BIPs was comparable in both arms. Thus such a bias seems unlikely.

A multivariate Cox proportional-hazards analysis showed that geographic region was an independent risk factor for BMs. Patients in Sweden, Norway, and the Netherlands had double the likelihood to develop BMs compared with Germany. This equated with the effect of a Gleason score of 8–10 or node-positive disease. Although we may speculate that genetic or other high-risk factors (eg, a higher positive margin rate) could explain the differences in the occurrence of BMs between countries, the reasons for these findings are presently unclear.

A major AE of BPs is ONJ, which occurs in 1–7% of patients who have undergone dental extraction while receiving BPs [23] and [24]. The frequency of ONJ increased when more risk factors such as corticosteroids, radiotherapy, and chemotherapy were present. Dental extractions or poor dental hygiene are other prerequisites for ONJ development [14], [23], and [24]. Once this information was published, these patients were excluded, and procedures to improve dental hygiene were recommended prior to ZA administration. After this amendment, no additional patients reported ONJ in our study.

Prostate cancer (PCa) is characterised by its propensity for bone metastases (BMs) that occur in >80% of patients with advanced PCa [1] . BMs result in morbidity and are associated with pain, pathologic fractures, and decreased survival. The standard treatment of patients with metastases is androgen deprivation therapy (ADT) using surgical or medical castration. The 5-yr survival rate of patients with metastatic or locally advanced disease subjected to ADT is 35% [2] . If considerable numbers of PCa patients receive prolonged ADT, decreased bone mineral density, increased risk for osteoporosis, and skeletal fractures can ensue [2] and [3].

Accumulating clinical evidence has confirmed the value of bisphosphonates (BPs) in BMs, preventing ADT-related bone loss, reducing morbidity and pain, and improving survival in patients with castration-resistant PCa (CRPC) [4], [5], [6], and [7]. Zoledronic acid (ZA; Zometa, Novartis, Basel, Switzerland), developed as a third-generation nitrogen-containing BP with increased potency relative to other BPs, induces direct inhibition of PCa cells in vitro, activation of cytotoxic T cells toward a range of malignant cell types, and inhibition of tumour-mediated angiogenesis that may render the bone microenvironment less conducive to micrometastatic growth [8], [9], [10], and [11].

We investigated the efficacy of 3-mo ZA for BM prevention in high-risk nonmetastatic PCa patients. A 3-mo interval was chosen for practical and financial reasons. In addition, data indicated that ZA has a long bone retention time, with approximately 50% of the peak dose persisting there 240 d after administration, and pharmacokinetics demonstrating a multiphasic plasma disposition with an initial half-life of 0.2–1.4 h, followed by 39–4526 h for subsequent elimination, indicating that ZA levels may persist for 3 mo [12] and [13].

ZA has long-term benefits in patients with CRPC and BMs. It is approved for prevention of skeletal-related events in metastatic genitourinary cancers and recommended for routine use in CRPC according to the European Association of Urology guidelines [16] . ZA was the subject of a metastasis prevention trial in nonmetastatic CRCP patients that was closed early because the data safety monitoring board halted the study after accrual of 398 of 991 planned patients, since the event rate was lower than expected [17] . Oral clodronate trials showed improved OS in men with metastatic PCa starting ADT but no evidence of effect in men with nonmetastatic PCa [6] . Smith et al. demonstrated that denosumab significantly increased BM-free survival by a median of 4.2 mo compared with placebo and also significantly delayed ttBMs [18] .

The additive anticancer effect of ZA combined with endocrine therapy in patients with early stage breast cancer was reported in the AZURE trial. Although the primary end point of disease-free survival was not met in the overall study population, subgroup analyses showed significant benefits for disease-free survival and OS in women ≥5 yr after menopause [19] . A significant anticancer effect of the addition of ZA on disease-free survival in premenopausal women with endocrine-responsive breast cancer taking anastrozole or tamoxifen was reported in the ABCSG-12 trial [20] .

In our study there was no difference in the occurrence of BMs between treatment groups. At 4 yr, the local urologist/radiologist diagnosed BMs in 17.1% of the patients in the ZAG and 17.0% in the CG with a difference between proportions of 0.1% (95% CI, −4.4 to 4.7) in favour of the CG. When looking at the lower limit of the 95% CI, the difference in favour of ZA was 4.4% at the highest point, which was lower than the assumed difference of at least 6% used for sample size calculation. A maximum risk reduction of 4.4% is equivalent to treating 23 patients with ZA to prevent BMs in 1 patient at 4 yr. When taking into account the associated patient burden of treatment with ZA (eg, AEs, frequent visits to the hospital for intravenous infusions, costs), the potential benefit of ZA was insufficient and clinically irrelevant. The central review results led to similar conclusions. When specifically focussing on the subgroup receiving ADT as in the breast cancer trials, no differences could be detected. The ZA dosage may be important because in the AZURE trial, patients received more intense ZA treatment: 4 mg 3 to 4-weekly ×6, every 3 mo ×8, and every 6 mo ×5 thereafter [19] . In the ABCSG-12 trial, however, a less intense ZA administration schedule was used: 4 mg every 6 mo for 3 yr [20] . In our study, ZA was administered every 3 mo for ≤4 yr, suggesting that frequency of ZA administration may not be the only reason for failure to prevent BMs. However, the main mechanisms of the prevention of BMs by ZA in breast cancer may be different in PCa. If the BM-preventive effect of ZA in PCa is not exerted in the bone where sufficient drug levels are maintained up to 240 d but is based on its effect on the mevalonate pathway and the immune system (effects likely to be dose dependent), we cannot exclude the possibility that ZA could prevent BMs when used more intensively than the 3-mo regimen in our study [11], [12], [13], and [21]. The ZOOM trial data support this view because bone markers (N-terminal telopeptide) remained low in the patients treated monthly but increased in patients treated every 3 mo, suggesting that accumulation of ZA in the bone did not ensure extended bone turnover suppression [22] .

BMs were assessed according to standard procedures where BSs were performed when symptoms appeared or PSA increased. BS hot spots are sensitive but not specific for metastases, and they are indistinguishable from other causes such as fractures or inflammation/infection without confirmative imaging. Unfortunately, only 53% of the patients with a positive BS had confirmative imaging, which reflects real-life practice patterns as a limitation of our study.

For a more accurate understanding of the number of BMs, the protocol was amended so that BIP images could be retrospectively collected for independent and blinded expert central review (F.L.). Participating centres were not originally selected based on their ability to generate BIP images for central review. Consequently, we collected paired BIPs from only 612 patients, another study limitation. Remarkably, more patients in the ZAG had equivocal findings compared with the CG, specifically because PSA categories mainly showed incongruously low PSA levels. It was likely that ZA also affected bone, which was reflected in bone/muscle pain or arthralgia. Because most of the equivocal BIP results were derived from not-for-cause BIPs, it is unlikely that symptoms could explain the higher rate of equivocal BIP results in the ZAG.

Another limitation of our study was our heterogeneous population. Patients treated locally with curative intent had fewer BMs compared with those who were not. Similarly, patients receiving early ADT were potentially more prone to developing BMs.

Another possible limitation of our study was the open-label design resulting in patients in the ZAG attending clinic more frequently compared with the CG, which may have resulted in differential outcomes reporting. In our study, more AEs were reported in the ZAG, which may be biased by the more frequent visit schedule. However, according to the known side-effect profile of ZA, more AEs can be expected. A more frequent visit schedule could lead to more BIPs with more patients diagnosed as metastatic in the ZAG compared with the CG. As shown in Supplemental Table 2, the number of BIPs was comparable in both arms. Thus such a bias seems unlikely.

A multivariate Cox proportional-hazards analysis showed that geographic region was an independent risk factor for BMs. Patients in Sweden, Norway, and the Netherlands had double the likelihood to develop BMs compared with Germany. This equated with the effect of a Gleason score of 8–10 or node-positive disease. Although we may speculate that genetic or other high-risk factors (eg, a higher positive margin rate) could explain the differences in the occurrence of BMs between countries, the reasons for these findings are presently unclear.

A major AE of BPs is ONJ, which occurs in 1–7% of patients who have undergone dental extraction while receiving BPs [23] and [24]. The frequency of ONJ increased when more risk factors such as corticosteroids, radiotherapy, and chemotherapy were present. Dental extractions or poor dental hygiene are other prerequisites for ONJ development [14], [23], and [24]. Once this information was published, these patients were excluded, and procedures to improve dental hygiene were recommended prior to ZA administration. After this amendment, no additional patients reported ONJ in our study.

Prostate cancer (PCa) is characterised by its propensity for bone metastases (BMs) that occur in >80% of patients with advanced PCa [1] . BMs result in morbidity and are associated with pain, pathologic fractures, and decreased survival. The standard treatment of patients with metastases is androgen deprivation therapy (ADT) using surgical or medical castration. The 5-yr survival rate of patients with metastatic or locally advanced disease subjected to ADT is 35% [2] . If considerable numbers of PCa patients receive prolonged ADT, decreased bone mineral density, increased risk for osteoporosis, and skeletal fractures can ensue [2] and [3].

Accumulating clinical evidence has confirmed the value of bisphosphonates (BPs) in BMs, preventing ADT-related bone loss, reducing morbidity and pain, and improving survival in patients with castration-resistant PCa (CRPC) [4], [5], [6], and [7]. Zoledronic acid (ZA; Zometa, Novartis, Basel, Switzerland), developed as a third-generation nitrogen-containing BP with increased potency relative to other BPs, induces direct inhibition of PCa cells in vitro, activation of cytotoxic T cells toward a range of malignant cell types, and inhibition of tumour-mediated angiogenesis that may render the bone microenvironment less conducive to micrometastatic growth [8], [9], [10], and [11].

We investigated the efficacy of 3-mo ZA for BM prevention in high-risk nonmetastatic PCa patients. A 3-mo interval was chosen for practical and financial reasons. In addition, data indicated that ZA has a long bone retention time, with approximately 50% of the peak dose persisting there 240 d after administration, and pharmacokinetics demonstrating a multiphasic plasma disposition with an initial half-life of 0.2–1.4 h, followed by 39–4526 h for subsequent elimination, indicating that ZA levels may persist for 3 mo [12] and [13].

ZA has long-term benefits in patients with CRPC and BMs. It is approved for prevention of skeletal-related events in metastatic genitourinary cancers and recommended for routine use in CRPC according to the European Association of Urology guidelines [16] . ZA was the subject of a metastasis prevention trial in nonmetastatic CRCP patients that was closed early because the data safety monitoring board halted the study after accrual of 398 of 991 planned patients, since the event rate was lower than expected [17] . Oral clodronate trials showed improved OS in men with metastatic PCa starting ADT but no evidence of effect in men with nonmetastatic PCa [6] . Smith et al. demonstrated that denosumab significantly increased BM-free survival by a median of 4.2 mo compared with placebo and also significantly delayed ttBMs [18] .

The additive anticancer effect of ZA combined with endocrine therapy in patients with early stage breast cancer was reported in the AZURE trial. Although the primary end point of disease-free survival was not met in the overall study population, subgroup analyses showed significant benefits for disease-free survival and OS in women ≥5 yr after menopause [19] . A significant anticancer effect of the addition of ZA on disease-free survival in premenopausal women with endocrine-responsive breast cancer taking anastrozole or tamoxifen was reported in the ABCSG-12 trial [20] .

In our study there was no difference in the occurrence of BMs between treatment groups. At 4 yr, the local urologist/radiologist diagnosed BMs in 17.1% of the patients in the ZAG and 17.0% in the CG with a difference between proportions of 0.1% (95% CI, −4.4 to 4.7) in favour of the CG. When looking at the lower limit of the 95% CI, the difference in favour of ZA was 4.4% at the highest point, which was lower than the assumed difference of at least 6% used for sample size calculation. A maximum risk reduction of 4.4% is equivalent to treating 23 patients with ZA to prevent BMs in 1 patient at 4 yr. When taking into account the associated patient burden of treatment with ZA (eg, AEs, frequent visits to the hospital for intravenous infusions, costs), the potential benefit of ZA was insufficient and clinically irrelevant. The central review results led to similar conclusions. When specifically focussing on the subgroup receiving ADT as in the breast cancer trials, no differences could be detected. The ZA dosage may be important because in the AZURE trial, patients received more intense ZA treatment: 4 mg 3 to 4-weekly ×6, every 3 mo ×8, and every 6 mo ×5 thereafter [19] . In the ABCSG-12 trial, however, a less intense ZA administration schedule was used: 4 mg every 6 mo for 3 yr [20] . In our study, ZA was administered every 3 mo for ≤4 yr, suggesting that frequency of ZA administration may not be the only reason for failure to prevent BMs. However, the main mechanisms of the prevention of BMs by ZA in breast cancer may be different in PCa. If the BM-preventive effect of ZA in PCa is not exerted in the bone where sufficient drug levels are maintained up to 240 d but is based on its effect on the mevalonate pathway and the immune system (effects likely to be dose dependent), we cannot exclude the possibility that ZA could prevent BMs when used more intensively than the 3-mo regimen in our study [11], [12], [13], and [21]. The ZOOM trial data support this view because bone markers (N-terminal telopeptide) remained low in the patients treated monthly but increased in patients treated every 3 mo, suggesting that accumulation of ZA in the bone did not ensure extended bone turnover suppression [22] .

BMs were assessed according to standard procedures where BSs were performed when symptoms appeared or PSA increased. BS hot spots are sensitive but not specific for metastases, and they are indistinguishable from other causes such as fractures or inflammation/infection without confirmative imaging. Unfortunately, only 53% of the patients with a positive BS had confirmative imaging, which reflects real-life practice patterns as a limitation of our study.

For a more accurate understanding of the number of BMs, the protocol was amended so that BIP images could be retrospectively collected for independent and blinded expert central review (F.L.). Participating centres were not originally selected based on their ability to generate BIP images for central review. Consequently, we collected paired BIPs from only 612 patients, another study limitation. Remarkably, more patients in the ZAG had equivocal findings compared with the CG, specifically because PSA categories mainly showed incongruously low PSA levels. It was likely that ZA also affected bone, which was reflected in bone/muscle pain or arthralgia. Because most of the equivocal BIP results were derived from not-for-cause BIPs, it is unlikely that symptoms could explain the higher rate of equivocal BIP results in the ZAG.

Another limitation of our study was our heterogeneous population. Patients treated locally with curative intent had fewer BMs compared with those who were not. Similarly, patients receiving early ADT were potentially more prone to developing BMs.

Another possible limitation of our study was the open-label design resulting in patients in the ZAG attending clinic more frequently compared with the CG, which may have resulted in differential outcomes reporting. In our study, more AEs were reported in the ZAG, which may be biased by the more frequent visit schedule. However, according to the known side-effect profile of ZA, more AEs can be expected. A more frequent visit schedule could lead to more BIPs with more patients diagnosed as metastatic in the ZAG compared with the CG. As shown in Supplemental Table 2, the number of BIPs was comparable in both arms. Thus such a bias seems unlikely.

A multivariate Cox proportional-hazards analysis showed that geographic region was an independent risk factor for BMs. Patients in Sweden, Norway, and the Netherlands had double the likelihood to develop BMs compared with Germany. This equated with the effect of a Gleason score of 8–10 or node-positive disease. Although we may speculate that genetic or other high-risk factors (eg, a higher positive margin rate) could explain the differences in the occurrence of BMs between countries, the reasons for these findings are presently unclear.

A major AE of BPs is ONJ, which occurs in 1–7% of patients who have undergone dental extraction while receiving BPs [23] and [24]. The frequency of ONJ increased when more risk factors such as corticosteroids, radiotherapy, and chemotherapy were present. Dental extractions or poor dental hygiene are other prerequisites for ONJ development [14], [23], and [24]. Once this information was published, these patients were excluded, and procedures to improve dental hygiene were recommended prior to ZA administration. After this amendment, no additional patients reported ONJ in our study.

Prostate cancer (PCa) is characterised by its propensity for bone metastases (BMs) that occur in >80% of patients with advanced PCa [1] . BMs result in morbidity and are associated with pain, pathologic fractures, and decreased survival. The standard treatment of patients with metastases is androgen deprivation therapy (ADT) using surgical or medical castration. The 5-yr survival rate of patients with metastatic or locally advanced disease subjected to ADT is 35% [2] . If considerable numbers of PCa patients receive prolonged ADT, decreased bone mineral density, increased risk for osteoporosis, and skeletal fractures can ensue [2] and [3].

Accumulating clinical evidence has confirmed the value of bisphosphonates (BPs) in BMs, preventing ADT-related bone loss, reducing morbidity and pain, and improving survival in patients with castration-resistant PCa (CRPC) [4], [5], [6], and [7]. Zoledronic acid (ZA; Zometa, Novartis, Basel, Switzerland), developed as a third-generation nitrogen-containing BP with increased potency relative to other BPs, induces direct inhibition of PCa cells in vitro, activation of cytotoxic T cells toward a range of malignant cell types, and inhibition of tumour-mediated angiogenesis that may render the bone microenvironment less conducive to micrometastatic growth [8], [9], [10], and [11].

We investigated the efficacy of 3-mo ZA for BM prevention in high-risk nonmetastatic PCa patients. A 3-mo interval was chosen for practical and financial reasons. In addition, data indicated that ZA has a long bone retention time, with approximately 50% of the peak dose persisting there 240 d after administration, and pharmacokinetics demonstrating a multiphasic plasma disposition with an initial half-life of 0.2–1.4 h, followed by 39–4526 h for subsequent elimination, indicating that ZA levels may persist for 3 mo [12] and [13].

ZA has long-term benefits in patients with CRPC and BMs. It is approved for prevention of skeletal-related events in metastatic genitourinary cancers and recommended for routine use in CRPC according to the European Association of Urology guidelines [16] . ZA was the subject of a metastasis prevention trial in nonmetastatic CRCP patients that was closed early because the data safety monitoring board halted the study after accrual of 398 of 991 planned patients, since the event rate was lower than expected [17] . Oral clodronate trials showed improved OS in men with metastatic PCa starting ADT but no evidence of effect in men with nonmetastatic PCa [6] . Smith et al. demonstrated that denosumab significantly increased BM-free survival by a median of 4.2 mo compared with placebo and also significantly delayed ttBMs [18] .

The additive anticancer effect of ZA combined with endocrine therapy in patients with early stage breast cancer was reported in the AZURE trial. Although the primary end point of disease-free survival was not met in the overall study population, subgroup analyses showed significant benefits for disease-free survival and OS in women ≥5 yr after menopause [19] . A significant anticancer effect of the addition of ZA on disease-free survival in premenopausal women with endocrine-responsive breast cancer taking anastrozole or tamoxifen was reported in the ABCSG-12 trial [20] .

In our study there was no difference in the occurrence of BMs between treatment groups. At 4 yr, the local urologist/radiologist diagnosed BMs in 17.1% of the patients in the ZAG and 17.0% in the CG with a difference between proportions of 0.1% (95% CI, −4.4 to 4.7) in favour of the CG. When looking at the lower limit of the 95% CI, the difference in favour of ZA was 4.4% at the highest point, which was lower than the assumed difference of at least 6% used for sample size calculation. A maximum risk reduction of 4.4% is equivalent to treating 23 patients with ZA to prevent BMs in 1 patient at 4 yr. When taking into account the associated patient burden of treatment with ZA (eg, AEs, frequent visits to the hospital for intravenous infusions, costs), the potential benefit of ZA was insufficient and clinically irrelevant. The central review results led to similar conclusions. When specifically focussing on the subgroup receiving ADT as in the breast cancer trials, no differences could be detected. The ZA dosage may be important because in the AZURE trial, patients received more intense ZA treatment: 4 mg 3 to 4-weekly ×6, every 3 mo ×8, and every 6 mo ×5 thereafter [19] . In the ABCSG-12 trial, however, a less intense ZA administration schedule was used: 4 mg every 6 mo for 3 yr [20] . In our study, ZA was administered every 3 mo for ≤4 yr, suggesting that frequency of ZA administration may not be the only reason for failure to prevent BMs. However, the main mechanisms of the prevention of BMs by ZA in breast cancer may be different in PCa. If the BM-preventive effect of ZA in PCa is not exerted in the bone where sufficient drug levels are maintained up to 240 d but is based on its effect on the mevalonate pathway and the immune system (effects likely to be dose dependent), we cannot exclude the possibility that ZA could prevent BMs when used more intensively than the 3-mo regimen in our study [11], [12], [13], and [21]. The ZOOM trial data support this view because bone markers (N-terminal telopeptide) remained low in the patients treated monthly but increased in patients treated every 3 mo, suggesting that accumulation of ZA in the bone did not ensure extended bone turnover suppression [22] .

BMs were assessed according to standard procedures where BSs were performed when symptoms appeared or PSA increased. BS hot spots are sensitive but not specific for metastases, and they are indistinguishable from other causes such as fractures or inflammation/infection without confirmative imaging. Unfortunately, only 53% of the patients with a positive BS had confirmative imaging, which reflects real-life practice patterns as a limitation of our study.

For a more accurate understanding of the number of BMs, the protocol was amended so that BIP images could be retrospectively collected for independent and blinded expert central review (F.L.). Participating centres were not originally selected based on their ability to generate BIP images for central review. Consequently, we collected paired BIPs from only 612 patients, another study limitation. Remarkably, more patients in the ZAG had equivocal findings compared with the CG, specifically because PSA categories mainly showed incongruously low PSA levels. It was likely that ZA also affected bone, which was reflected in bone/muscle pain or arthralgia. Because most of the equivocal BIP results were derived from not-for-cause BIPs, it is unlikely that symptoms could explain the higher rate of equivocal BIP results in the ZAG.

Another limitation of our study was our heterogeneous population. Patients treated locally with curative intent had fewer BMs compared with those who were not. Similarly, patients receiving early ADT were potentially more prone to developing BMs.

Another possible limitation of our study was the open-label design resulting in patients in the ZAG attending clinic more frequently compared with the CG, which may have resulted in differential outcomes reporting. In our study, more AEs were reported in the ZAG, which may be biased by the more frequent visit schedule. However, according to the known side-effect profile of ZA, more AEs can be expected. A more frequent visit schedule could lead to more BIPs with more patients diagnosed as metastatic in the ZAG compared with the CG. As shown in Supplemental Table 2, the number of BIPs was comparable in both arms. Thus such a bias seems unlikely.

A multivariate Cox proportional-hazards analysis showed that geographic region was an independent risk factor for BMs. Patients in Sweden, Norway, and the Netherlands had double the likelihood to develop BMs compared with Germany. This equated with the effect of a Gleason score of 8–10 or node-positive disease. Although we may speculate that genetic or other high-risk factors (eg, a higher positive margin rate) could explain the differences in the occurrence of BMs between countries, the reasons for these findings are presently unclear.

A major AE of BPs is ONJ, which occurs in 1–7% of patients who have undergone dental extraction while receiving BPs [23] and [24]. The frequency of ONJ increased when more risk factors such as corticosteroids, radiotherapy, and chemotherapy were present. Dental extractions or poor dental hygiene are other prerequisites for ONJ development [14], [23], and [24]. Once this information was published, these patients were excluded, and procedures to improve dental hygiene were recommended prior to ZA administration. After this amendment, no additional patients reported ONJ in our study.

Prostate cancer (PCa) is characterised by its propensity for bone metastases (BMs) that occur in >80% of patients with advanced PCa [1] . BMs result in morbidity and are associated with pain, pathologic fractures, and decreased survival. The standard treatment of patients with metastases is androgen deprivation therapy (ADT) using surgical or medical castration. The 5-yr survival rate of patients with metastatic or locally advanced disease subjected to ADT is 35% [2] . If considerable numbers of PCa patients receive prolonged ADT, decreased bone mineral density, increased risk for osteoporosis, and skeletal fractures can ensue [2] and [3].

Accumulating clinical evidence has confirmed the value of bisphosphonates (BPs) in BMs, preventing ADT-related bone loss, reducing morbidity and pain, and improving survival in patients with castration-resistant PCa (CRPC) [4], [5], [6], and [7]. Zoledronic acid (ZA; Zometa, Novartis, Basel, Switzerland), developed as a third-generation nitrogen-containing BP with increased potency relative to other BPs, induces direct inhibition of PCa cells in vitro, activation of cytotoxic T cells toward a range of malignant cell types, and inhibition of tumour-mediated angiogenesis that may render the bone microenvironment less conducive to micrometastatic growth [8], [9], [10], and [11].

We investigated the efficacy of 3-mo ZA for BM prevention in high-risk nonmetastatic PCa patients. A 3-mo interval was chosen for practical and financial reasons. In addition, data indicated that ZA has a long bone retention time, with approximately 50% of the peak dose persisting there 240 d after administration, and pharmacokinetics demonstrating a multiphasic plasma disposition with an initial half-life of 0.2–1.4 h, followed by 39–4526 h for subsequent elimination, indicating that ZA levels may persist for 3 mo [12] and [13].