Raising the Bar for Therapeutic Trials in Advanced Prostate Cancer

It was not until 2004 that the first clinical trials in metastatic castration-resistant prostate cancer (mCRPC) demonstrated a survival benefit for men with this disease. The results of the TAX327 and SWOG 9916 trials demonstrated extended survival and improved quality-of-life benefit of docetaxel, offering a new option for patients with mCRPC. Since then, phase III clinical trials of sipuleucel-T, cabazitaxel, abiraterone, enzalutamide, and radium-223 have established the efficacy of these agents for the treatment of mCRPC. Furthermore, recent work also demonstrates a significant improvement in overall survival with earlier use of docetaxel in the metastatic castration-naive setting. Despite these notable successes, failure has been much more common in mCRPC phase III clinical trials (Table 1). As demonstrated in the article accompanying this editorial, there are several reasons why the phase III trial of cabozantinib versus prednisone might have been negative, despite the apparently promising results of earlier-stage trials. Improvements in progressionfree survival, bone scan response, circulating tumor cell conversion—end points commonly associated with overall survival benefits—were seen in the phase II clinical trials of cabozantinib. However, despite these promising signals, declines in serum prostate-specific antigen concentrations—an imperfect and yet important marker of antitumor activity—were rare in prior phase II studies of cabozantinib, and PSA responses remained uncommon in the phase III trial. Moreover, it is possible that changes in bone scintigraphy seen with cabozantinib reflected, at least in part, interference with the imaging modality rather than true antitumor activity. This hypothesis is supported by the observation that enzalutamide and abiraterone (agents that have well-established antitumor activity inmCRPC and produce substantial improvements in progression-free and overall survival) do not cause the dramatic improvements in bone scintigraphy that are seen with cabozantinib. Thus, the true level of antitumor activity of cabozantinib was likely lower than hoped for. Other factors that are not unique to cabozantinib may also have contributed to the negative outcome of this phase III trial. These factors include the molecular and clinical heterogeneity of the unselected patient population studied, access to postprotocol life-prolonging therapies, and early discontinuation of cabozantinib because of toxicity. However, even if the trial proved to be positive in a statistical sense, would it have produced a clinically meaningful result? Would it have yielded a new, well-tolerated treatment option that patients with mCRPC and their providers would embrace? Although we will not know the answer to those questions, concern is raised by themagnitude of overall survival improvement this trial was designed to detect. The trial sample size was selected based on the assumption that the median survival in the control armwould be 7months, and success was defined as a 25% reduction in the risk of death (a hazard ratio of 0.75). This might be a meaningful amount of benefit to a patient whose life expectancy is 3 years, but it translates into only a 2.3-month improvement in median survival in the patient population tested. Even acknowledging that treatment benefit varies considerably between patients, one is left to ask: Is this a bar worth jumping over? Furthermore, the trial was designed with 90% power. If we were to repeat the power calculation with the commonly used 80%