Is there still a role for intermittent androgen-deprivation therapy for prostate cancer? No.
In the 1990s, Bruchovsky et al1 and Sato et al2 reported that in the androgen-dependent mammary carcinoma and LNCaP (androgen-sensitive human prostate adenocarcinoma cells) prostate cancer models the proportion of residual “androgen-independent” stem cells, surviving androgen withdrawal, could adapt to androgen-depleted environment and that intermittent reexposure to androgens reduces the proportion of surviving cells and delay emergence of androgen independence, thus providing the basis for the intermittent androgen deprivation (IAD) hypothesis. Clinically it was proposed that IAD allows recovery of gonadal function and consequently reexposes the disease to intrinsic testosterone during off-treatment period. The objective was to delay emergence of androgen independence (as defined 26 years ago) and thus improve outcomes, including progression-free and overall survival (OS) and quality of life (QOL) compared with continuous androgen deprivation (CAD).
In a recent review3 and editorial,4 we provided an in-depth discussion of 7 fully reported randomized phase 3 clinical trials comparing IAD with CAD, highlighted key limitations of each, and assessed validity of authors conclusions based on each trial’s result. All studies underestimated outcomes assumptions. With exception of the SWOG S9346 trial,5 all were underpowered for OS and/or had short follow-up time to capture the necessary number of events to allow for adequate testing of null hypothesis. Other limitations include broad inclusion criteria (stage M0, M1 disease); variable androgen-deprivation approaches; several did not select for responsive disease; use of agents of unproven efficacy, such as cyproterone acetate (not FDA-approved for prostate cancer), in SEUG (South European Uroncological Group) trials 9401 and 9901 and the TULP (Therapy Upgrading Life in Prostate Cancer) trial.4 Most studies used broadly defined and/or unvalidated criteria for time to progression to castration resistance (TTP); of note, TTP artificially favors IAD. None of the trials demonstrated significant and/or sustainable impact on QOL with IAD.
S9346, JPR7,6 and TAP 227 were conducted in homogeneous populations and designed with OS as the primary end point. TAP 22 was a “superiority study.” Despite the small sample size, results indicate that IAD had worse median survival compared with CAD (3.5 vs 4.3 years). The 2 largest trials, PR76 in hormone-naïve nonmetastatic (HNNM) prostate-specific antigen (PSA) relapse postradiation and our own S9346 trial5 in metastatic hormone-sensitive prostate cancer, were designed as noninferiority (NI) studies. To adequately interpret these studies, it is critical to understand the basis for the actual design assumptions of NI studies. Specifically, the noninferiority margin (NIM) must be considered clinically acceptable to allow for reliable and clinically meaningful conclusions. The NIM can translate into varying absolute survival differences depending on the patient population and their expected survival; for example, if the expected median survival of the control arm is 5 years, a 1.25 NIM translates into a 4-year median survival in the experimental arm. Hence, an absolute difference of 1 year less survival between the arms would presumably be considered noninferior.
The PR7 trial, designed by the National Cancer Institute of Canada,6 evaluated IAD vs CAD without a no-treatment or deferred-treatment control arm. It should be noted that the standard of care in the tested population remains undefined and the choice of any androgen-deprivation therapy in M0 disease remains purely conjectural; therefore, the PR7 trial does not provide useful information to define the standard of care for this population. The PR7 trial was designed with an NIM of 1.25, implying that the investigators determined a priori that a 1.8-year reduction in OS with IAD is clinically acceptable and noninferior. The observed median survival was 9.1 years with CAD and 8.8 years with IAD. The investigators concluded that IAD was noninferior compared with CAD. However, based on the observed OS and hazard ratio (HR) confidence interval, a 1.6-year reduction in OS with IAD cannot be ruled out with 95% confidence. At the time of PR7 trial report,6 less than 45% of deaths were related to prostate cancer or treatment and follow-up time was clearly very short (6.9 years) for this population. Interestingly, despite a short follow-up, the IAD arm had more deaths from prostate cancer.
In the S9346 trial5 the results show that IAD had worse median survival compared with CAD: 5.1 vs 5.8 years. The observed HR upper limit of 1.23 with median survival of 5.8 years in the CAD arm translates into an absolute survival difference of 1-year in favor of CAD. The QOL data revealed that IAD was associated with better erectile function and mental health at month 3 (but not thereafter). Interestingly, treatment-related grades 3 and 4 adverse events, including cardiovascular events, were not different between the arms. Furthermore, we recently examined long-term, late adverse events in elderly patients randomized to IAD or CAD in the S9346 trial using Medicare claims. Contrary to initial assumptions, older men receiving IAD had no apparent reduction in bone, endocrine, or cognitive adverse events and had an increased incidence of ischemic and thrombotic events.8
When IAD was introduced, it was assumed that testosterone recovery and PSA dynamics would remain relatively constant and predictable with IAD cycles. However, Keizman et al9 reported that PSA doubling times become shorter and testosterone recovery is progressively slower with repeated cycles. Because retreatment criteria are often based on PSA levels (not testosterone recovery), it is likely that after a few cycles, retreatment is instituted before full testosterone recovery, thus not reflecting the acute intermittent reexposure of prostate cancer cells to exogenous testosterone in the experiments of Bruchovsky et al1 and Sato et al.2 Isaacs et al10 reported that acute exposure of prostate cancer cells, growing in an androgen-depleted environment, to exogenous testosterone resulted in cell death and resensitization to subsequent testosterone withdrawal. These data suggest that intermittent supraphysiologic testosterone administration may be a more precise mode to test the IAD hypothesis. The authors10 suggest that the underlying mechanisms involve induction of androgen-receptor (AR) degradation and reversal of other potential AR-based adaptation processes involved in progression to castration-resistant prostate cancer. Cyclical testosterone administration is currently undergoing clinical testing.
In summary, 26 years after its introduction, the IAD concept remains unsupported by clinical trials data despite thousands of patients and millions of dollars. Attempts to derive reliable conclusions based on meta-analyses of aggregate studies data are limited by patients’ and outcomes’ heterogeneity.
Compared with the 1980s and 1990s when the concept of IAD was introduced, we have come a long way at multiple levels. Our understanding of the biology and mechanisms of progression to castration resistance has significantly evolved. Current biological knowledge support concepts of clonal selection and adaptation via AR-dependent and independent mechanisms. Androgen-receptor signaling remains constitutively active despite castration, a principle validated by abiraterone and enzalutamide phase 3 trials which argue that persistent AR antagonism and selective pressure is the more biologically appropriate therapy and argues strongly against routine use of IAD in the clinic.
The evolving clinical/biological data and the unprecedented OS improvements with docetaxel plus ADT,3 strongly argue that it is time to move on and focus our efforts on translating compelling discoveries into therapies that are most likely to result in true value and meaningful impact for men with prostate cancer.
Corresponding Author: Maha Hussain, MD, Departments of Internal Medicine and Urology, University of Michigan, 7314 Comprehensive Cancer Center, 1500 E Medical Center Dr, Ann Arbor, MI 48109-5946 (email@example.com).
Published Online: August 25, 2016. doi:10.1001/jamaoncol.2016.2650
Conflict of Interest Disclosures: None reported.
Hussain M, Eisenberger M. Intermittent Androgen DeprivationPrimum Non Nocere—“First Do NO Harm”. JAMA Oncol. 2016;2(12):1533-1534. doi:10.1001/jamaoncol.2016.2650