Customize your JAMA Network experience by selecting one or more topics from the list below.
Sun M, Choueiri TK, Hamnvik OR, et al. Comparison of Gonadotropin-Releasing Hormone Agonists and Orchiectomy: Effects of Androgen-Deprivation Therapy. JAMA Oncol. 2016;2(4):500–507. doi:10.1001/jamaoncol.2015.4917
Androgen-deprivation therapy (ADT) through surgical castration is equally effective as medical castration in controlling prostate cancer (PCa). However, the adverse effect profiles of both ADT groups have never been compared.
To provide a comparative effectiveness analysis of the adverse effects of gonadotropin-releasing hormone agonists (GnRHa) vs bilateral orchiectomy in a homogeneous population.
Design, Setting, and Participants
A population-based cohort of 3295 men with metastatic PCa between January 1995 and December 2009 66 years or older was selected from the Surveillance, Epidemiology, and End Results (SEER) Medicare-linked database.
Orchiectomy or GnRHa.
Main Outcomes and Measures
Any fractures, peripheral arterial disease, venous thromboembolism, cardiac-related complications, diabetes mellitus, and cognitive disorders. To minimize treatment group biases, the inverse probability of treatment was weighted using the propensity score. Multivariable competing risk regression models were performed with the adjustment of all-cause mortality. Secondary analyses examined the effect of increasing duration of GnRHa treatment. Multivariable logistic regression models examined expenditures.
Overall, 3295 men with a primary diagnosis of metastatic PCa treated with GnRHa or orchiectomy were identified between years 1995 and 2009, and in adjusted analyses, patients who received a bilateral orchiectomy had significantly lower risks of experiencing any fractures (hazard ratio [HR], 0.77; 95% CI, 0.62-0.94; P = .01), peripheral arterial disease (HR, 0.65; 95% CI, 0.49-0.87; P = .004), and cardiac-related complications (HR, 0.74; 0.58-0.94; P = .01) compared with those treated with GnRHa. No statistically significant difference was noted between orchiectomy and GnRHa for diabetes and cognitive disorders. In individuals treated with GnRHa for 35 months or more, the increased risk for GnRHa compared with orchiectomy was noted for fractures (HR, 1.80), peripheral arterial disease (HR, 2.25), venous thromboembolism (HR, 1.52), cardiac-related complications (HR, 1.69), and diabetes mellitus (HR, 1.88) (P ≤ .01 for all). At 12 months after PCa diagnosis, the median total expenditures was not significantly different between GnRHa and orchiectomy.
Conclusions and Relevance
Gonadotropin-releasing hormone agonist therapy is associated with higher risks of several clinically relevant adverse effects compared with orchiectomy.
Androgen-deprivation therapy (ADT) achieved with surgical or medical castration, leading to disruption of the hypothalamic-pituitary-gonadal axis, has been a cornerstone in the management of metastatic prostate cancer (PCa) for the past half-century.1 Indeed, the effectiveness of ADT for treatment of advanced PCa has been demonstrated in several clinical studies. For example, ADT is supported by level 1 evidence as neoadjuvant, concurrent, or adjuvant treatment to radiotherapy for men with high-risk and locally-advanced PCa.2 In men with metastatic PCa, results from the SWOG-9346 trial3 suggest that the survival outcomes between intermittent and continuous ADT were inconclusive. However, a 20% greater mortality risk with intermittent ADT cannot be ruled out. Given that the study failed to demonstrate noninferiority of intermittent ADT with respect to survival compared with continuous ADT, continuous ADT currently represents the standard of care for metastatic PCa according to national guidelines.4 That said, 2 meta-analyses5,6 on randomized clinical trials reported no survival difference between intermittent and continuous ADT. However, there is mounting evidence that ADT is linked to significant adverse effects, such as cardiovascular events,7,8 diabetes,8,9 acute kidney injury,10 and bone loss.11,12
In contemporary years, cosmetic and psychological concerns have nearly eliminated the use of bilateral orchiectomy in the United States, even for patients who will need lifelong ADT.13 However, one of the publications that first noted an increased risk for cardiac events on gonadotropin-releasing hormone agonists (GnRHa) indirectly observed that patients who received orchiectomy may not have the same increased risk of cardiac events.8 Some have hypothesized that the adverse cardiac events may be somehow linked to the GnRHa themselves rather than the hypogonadism effect.14 Hence, there is a possibility that orchiectomy could be associated with fewer adverse effects on multiple health domains compared with long-term GnRHa use.
Our objective was to perform a direct comparison of adverse effects of GnRHa vs orchiectomy vs no ADT in men with metastatic PCa. Given that providing value and cost-effective care has never been more important, we also sought to perform an updated analysis of the differential expenditures according to ADT groups in the contemporary era.
The purpose of our study was to compare the adverse effect profiles of medical vs surgical castration.
Medical castration was associated with higher risks of several clinically relevant adverse effects compared with orchiectomy.
Surgical castration was associated with lower risks of any fractures (hazard ratio [HR], 0.77; 95% CI, 0.62-0.94), peripheral arterial disease (HR, 0.65; 95% CI, 0.49-0.87), and cardiac-related complications (HR, 0.74; 95% CI, 0.58-0.94) compared with medical castration.
In some patients who need permanent androgen suppression, surgical castration may represent a suitable alternative to medical castration.
An institutional review board waiver was obtained prior to conducting this study in accordance with institutional regulation when dealing with deidentified administrative data.
Data originated from the Surveillance, Epidemiology, and End Results (SEER) program of cancer registries that collect clinical demographic and cause of death information for persons with cancer and the Medicare claims for covered health care services from the time of a person’s Medicare eligibility until death. To link SEER with Medicare data, the participating SEER registries send individual identifiers for all persons in their files, which are matched with identifiers contained in Medicare’s master enrolment file. For the linkage, 93% of persons 65 years and older in the SEER files were matched to the Medicare enrollment file.15
Patients with a primary diagnosis of PCa who were 66 years or older between January 1995 and December 2009 were identified (n = 28 749). Patients who were not enrolled in both Parts A and B Medicare 12 months before and 12 months after their diagnosis (n = 11 810), those part of a health maintenance organization (HMO), or those diagnosed on autopsy or death certificate only (n = 5725) were not considered. Patients who had nonmetastatic PCa (n = 3854) were removed from our analyses. This resulted in 7360 patients with a primary diagnosis of metastatic PCa. From these, those who had both GnRHa and orchiectomy (n = 91 [1.2%]), radical prostatectomy (n = 150 [0.7%]), radiotherapy (n = 1176 [16.0%]), both (n = 3 [0.4%]), or those who did not receive any therapy (n = 2002 [27.2%]) were omitted. This resulted in 4336 patients. Patients with unknown marital status (n = 274 [6.3%]), unknown race (n = 9 [0.2%]), unknown region (n = 703 [16.2%]), and unknown income (n = 55 [1.3%]) were also removed. A total of 3295 patients remained in our study. Patients who received bilateral orchiectomy (Healthcare Common Procedure Coding System [HCPCS] codes 54520, 54521, 54522, 54530 or 54535, or the International Classifications of Diseases, 9th Revision [ICD-9] code 624) were differentiated from those who received 1 or more doses of GnRHa (HCPCS codes J1950, J9202, J9217, J9218, J9219).8,12 Only those who received GnRHa or orchiectomy as primary cancer therapy within 12 months of metastatic PCa diagnosis were included. Of note, patients who received antiandrogens as monotherapy or in combination with GnRHa, as well as men who received GnRH antagonists, were also not considered in the analyses since FDA approval was only obtained in 2008.
The primary end points examined in the current analysis were comprised of 6 major adverse effects: any fractures, peripheral arterial disease, venous thromboembolism, cardiac-related complications, diabetes mellitus, and cognitive disorders. These adverse outcomes were selected on the basis of their effect on a patient’s quality of life, the potential for increased health care costs, and previously described association with ADT use.8,12,16,17 Age at diagnosis, race (white, black, other), marital status (single, unmarried), SEER regions (East, Northern Plain, Pacific Coast, Southwest), county-specific income levels, as well as year of diagnosis, were considered as covariates. Biopsy Gleason score was dichotomized as 7 or lower vs 8 to 10. Baseline comorbidities were computed according to the Klabunde-modified Charlson comorbidity index (CCI).18
Finally, the expenditures associated with use of ADT was assessed as a seventh end point. To accomplish this, all Medicare health expenditures 12 months after PCa diagnosis were extracted based on claims from inpatient, outpatient, and physician services and summed. Using each patient as his own control, we subtracted health expenditures accrued 12 months prior to use of ADT to estimate the baseline annual health charges for each patient, as previously described.19 All expenditures are reported in 2009 US dollars.
Descriptive statistics were used to describe the patient population according to ADT status. Differences between treatment groups and categorical baseline characteristics were compared using the χ2 test; the Wilcoxon rank-sum test was used to compare differences in median values and distributions of continuous factors in non-normally distributed populations.
First, to account for potential treatment selection biases arising from nonrandom allocation of patients to different ADT groups, we relied on the inverse probability of treatment weighting (IPTW) using the propensity score.20 It uses weights based on the propensity score to create a synthetic sample in which the distribution of measured baseline covariates is independent of treatment assignment. Covariates included were age, CCI, year of diagnosis, race, Gleason score, income, marital status, and region.
Second, to determine the effect of ADT type, primary analyses consisted of competing risk regression models that examined each major adverse effect. In all these models, GnRHa was used as the referent category. All models were adjusted for all-cause mortality. We chose to rely on competing risk regression models, as described by Fine and Gray,21 instead of the conventional Cox regression models because competing risk regression models control for the loss of patients due to death at various time points. Of interest, when the same outcomes were examined using Cox regression models, none of the examined end points showed any statistically significant differences between GnRHa and orchiectomy. Indeed, the 2 treatment groups demonstrated significant variability in patient censoring (log-rank P < .001) and overall survival (log-rank P = .029). For this reason, we chose to rely on the competing risk regression models, although it is important to acknowledge the strengths in both model types.22
Furthermore, for each end point of interest that was tested, patients with a baseline condition of the end point studied within that model were excluded. For example, 714 patients with a baseline fracture-related diagnosis 12 months prior to PCa diagnosis were excluded when looking at the risk of experiencing any fracture according to ADT status.
Third, secondary analyses were conducted by examining duration of GnRHa exposure quartiles (1-9, 10-17, 18-34, ≥35 months).12 In these analyses, orchiectomy was used as the referent category. Finally, expenditure-effective comparisons between treatment groups were made using multivariable logistic regression models. In this step, more expensive treatment 12 months after PCa diagnosis was defined according to total expenditures above the 75th percentile ($21 451.65). All statistical testing was 2-sided with a level of significance set at P < .05. All analyses were performed with SAS version 10 (SAS Institute).
Overall, 3295 men with a primary diagnosis of metastatic PCa treated with GnRHa (n = 2866 [87.0%]) or orchiectomy (n = 429 [13.0%]) between 1995 and 2009 were identified (Table 1). Statistically significant differences were observed according to age, baseline CCI, income, region, and year of diagnosis (P = .03 for all comparisons) (Table 1). The 3-year overall survival rate was 46% (95% CI, 44%-48%) for GnRHa treatment vs 39% (95% CI, 35%-44%) for orchiectomy (log-rank P = .03). Following the inverse probability of treatment weighting, the 2 groups were well balanced (Table 2).
In adjusted analyses, orchiectomy was associated with significantly lower risks of any fracture (HR, 0.77; 95% CI, 0.62-0.94), peripheral arterial disease (HR, 0.65; 95% CI, 0.49-0.87), and cardiac-related complications (HR, 0.74; 95% CI, 0.58-0.94) (Table 3) compared with GnRHa. Our results also showed that patients who received GnRHa for the longest duration (≥35 months) were at the greatest risks of experiencing 5 of 6 of the examined end points: any fracture (HR, 1.80), peripheral arterial disease (HR, 2.25), venous thromboembolism (HR, 1.52), cardiac-related complications (HR, 1.69), and diabetes mellitus (HR, 1.88) (P ≤ .001 for all comparisons) (Table 4).
At 1 year after PCa diagnosis, the median total expenditures were $9726.98 for orchiectomy vs $8478.46 for GnRHa (P = .24) (Table 5). After adjusting for treatment propensity scores, no difference with respect to total expenditures greater than the 75th percentile was observed between GnRHa and orchiectomy (odds ratio, 1.14; 95% CI, 0.88-1.45).
In the current study, we sought to directly compare the adverse-effect profiles of GnRHa vs orchiectomy and, subsequently, the expenditures between the 2 approaches. Primarily, our findings indicate that GnRHa was associated with significantly higher risks of fractures, peripheral arterial disease, and cardiac-related complications compared with surgical castration. When GnRHa was stratified into 4 different categories according to duration of therapy, our findings indicate that duration of GnRHa exposure of 35 months or more had a direct negative effect on the patients’ health with respect to fractures, peripheral arterial disease, venous thromboembolism, cardiac-related complications, and diabetes mellitus compared with orchiectomy. Cognitive disorders including major depression and dementia were not shown to be different between orchiectomy and GnRHa.
The fewer ADT-related toxic effects associated with orchiectomy relative to GnRHa noted in this study suggest that differences in the physiologic effects of orchiectomy vs GnRHa therapy has clinically significant implications. The specific biological mechanisms need to be better elucidated, but several theories may be put forth. First, effect on endogenous androgens may not be equivalent between the 2 therapies. Gonadotropin-releasing hormone agonist therapy is associated with an initial surge in testosterone levels. It is possible that the adverse effects associated with these treatments are precipitated by this initial androgen surge. However, the increase in GnRHa side-effects with time suggest that the initial brief increase of testosterone levels may not be entirely to blame. More relevant are the long-term testosterone levels. Both orchiectomy and GnRHa eventually reduce testosterone levels to very low levels, although 15% of patients treated with GnRHa have levels above the levels achieved after bilateral orchiectomy.23 The residual androgen activity is a result of adrenal androgen production, as well as possible breakthrough testicular production in the case of GnRHa therapy. The latter is thought to be the cause of the persistent microsurges in testosterone levels seen in about 6% of patients treated with GnRHa.24 However, these microsurges have unclear relevance when it comes to PCa outcomes and adverse effects. It is also possible that higher testosterone levels are seen in patients taking GnRHa therapy owing to noncompliance, which is not relevant after orchiectomy. However, higher testosterone levels should improve bone density and thus reduce fracture rates25 and the incidence of cardiovascular disease.26 Therefore, testosterone levels cannot explain the totality of the findings shown herein.
An alternative explanation could be related to differing effects on other hormones. Whereas orchiectomy leads to low anti-Müllerian hormone (AMH) levels and high luteinizing hormone and follicle-stimulation hormone levels, GnRHa are associated with unchanged AMH levels and low or low-normal follicle-stimulation hormone and luteinizing hormonelevels.27 Another hormonal change that occurs with both orchiectomy and GnRHa therapy is a reduction in inhibin-B levels.27 However, the actual clinical relevance of these changes on our current findings are unknown and remain speculative at best.
Against a backdrop of increasing demand for better value in American health care and the continued need for long-term castration in select men with advanced PCa, these results suggest that some consideration should be given to better incorporate the use of bilateral orchiectomy for men who need permanent androgen suppression. This option may be advantageous for the patient (ie, increased avoidance of long-term adverse effects) and possibly the health care system. Both GnRHa and bilateral orchiectomy represent equally effective ADT for locally advanced or metastatic PCa. However, since the introduction of GnRHa, surgical castration has been on the decline.13 Indeed, 7 of 10 patients would opt for GnRHa given the choice.28 Certainly, the reason for this dramatic shift is multifactorial. The preference of medical over surgical castration is likely largely associated with patients’ fear of the less desirable cosmetics, psychological factors about self-image, stress of the surgical procedure, and the irreversibility of the intervention. However, several investigators have shown that the actual psychological effect of surgical castration may not be as significant as it is perceived to be. For example, Bonzani et al29 examined quality of life and body image among medically and surgically castrated patients, and found no difference. Data from the US Prostate Cancer Outcomes Study showed that GnRHa-treated men were more likely to worry about their disease and more often expressed a general dissatisfaction with their overall health compared with those who received an orchiectomy (67% vs 51%) (P = .004).30 Apart from the patient’s perspective, other reasons for the dramatic rise of GnRHa therapy may possibly include the lack of discussion of both treatment options in an unbiased fashion, the urge to provide treatment in the setting of a rising prostate-specific antigen, and/or financial incentives to clinicians. Our results highlight the importance of continuing to present patients with the pros and cons of medical vs surgical castration.
Despite its strengths, our study is not devoid of limitations. Overall limitations reflect challenges related to exposure assessment in Medicare claims, including inaccuracies and the time frame of treatment ascertainment.31 Primarily, it is also limited by its retrospective design. Specifically, patients were not randomized to receive GnRHa, orchiectomy, or no ADT. Inherent differences with respect to number of patients, and patient and clinical characteristics between the 2 groups may be present but unmeasured in our analyses. Second, given that the current database is limited to men 65 years and older, the results cannot be directly applicable to younger patients. Other considerations (ie, intermittent vs continuous ADT) in men younger than 65 years with a primary diagnosis of metastatic PCa may need to be evaluated in conjunction with risks of adverse effects. Third, PSA data were not available due to a coding issue, precluding the ability to adjust for this variable in our analyses. This represents an important limitation of the study. There is a need for prospective studies of GnRHa compared with orchiectomy and other ADT treatments (ie, GnRH antagonists). These would provide convincing level 1 evidence and help to determine whether the risks of GnRHa are real and whether the risks are unique to specific types of ADT. Furthermore, many unmeasured confounders (ie, body mass index, tobacco use, and genetic predisposition) could not be accounted for. As disease outcomes were identified using diagnosis codes in administrative data, we also cannot exclude the possibility that men with more frequent doctor visits were more likely to be diagnosed and treated with these adverse events, although we expect that even patients who underwent orchiectomy would continue to have frequent visits for PSA and symptom checks to evaluate for possible emergence of castration-resistant disease. That said, patients who underwent orchiectomy may have actually been recommended as such because of their noncompliance. In the sub-analyses comparing duration of GnRHa to orchiectomy may be biased, as those still receiving treatment beyond 35 months after diagnosis may be significantly different with respect to health and disease status compared with all orchiectomy patients. We sought to address this limitation by adjusting for confounders and, most importantly, by accounting for all-cause mortality using competing risk models. Hence, the number of deaths, if greater in the orchiectomy group, was accounted for when measuring the effect of ADT on side effects.
Gonadotropin-releasing hormone agonist therapy was associated with higher risks in 3 of 6 examined adverse effects: any fractures, peripheral arterial disease rates, and cardiac-related complications. In some patients who need permanent androgen suppression, surgical castration may represent a suitable alternative to GnRHa. However, other considerations must be contemplated when deciding between medical or surgical castration (ie, young age, intermittent ADT).
Corresponding Author: Quoc-Dien Trinh, MD, Center for Surgery and Public Health, Division of Urologic Surgery, Brigham and Women’s Hospital/Dana-Farber Cancer Institute, Harvard Medical School, 45 Francis St, Urology-ASB II-3, Boston, MA 02115 (email@example.com).
Published Online: December 23, 2015. doi:10.1001/jamaoncol.2015.4917.
Author Contributions: Dr Trinh had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Choueiri, Nguyen, Trinh, and Ms Sun contributed equally to this article.
Study concept and design: Sun, Trinh.
Acquisition, analysis, or interpretation of data: Sun, Choueiri, Hamnvik, Preston, De Velasco, Jiang, Loeb, Nguyen, Trinh.
Drafting of the manuscript: Sun, Trinh.
Critical revision of the manuscript for important intellectual content: Sun, Choueiri, Hamnvik, Preston, De Velasco, Jiang, Loeb, Nguyen, Trinh.
Statistical analysis: Sun, Preston, Jiang.
Obtained funding: Choueiri.
Administrative, technical, or material support: Choueiri.
Study supervision: Choueiri, De Velasco, Loeb, Nguyen, Trinh.
Conflict of Interest Disclosures: Dr Loeb served as an advisory board member and receiving an honorarium for a lecture from Bayer. Dr Nguyen has served as a consultant for Medivation and Genome Dx. No other conflicts are reported.
Funding/Support: Dr Loeb is supported by the Laura and Isaac Perlmutter New York University Cancer Center and the Louis Feil Charitable Lead Trust. Dr Trinh is supported by an unrestricted educational grant from the Vattikuti Urology Institute and the Professor Walter Morris-Hale Distinguished Chair in Urologic Oncology at Brigham and Women’s Hospital.
Role of the Funder/Sponsor: The Laura and Isaac Perlmutter New York University Cancer Center, Louis Feil Charitable Lead Trust, Vattikuti Urology Institute, and Brigham and Women’s Hospital had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.