CHICAGO–Data from a trial of Johnson & Johnson’s blockbuster prostate cancer drug Zytiga (abiraterone) in asymptomatic or mildly symptomatic chemotherapy-naïve patients was received with widespread enthusiasm at the American Society of Clinical Oncology annual meeting, but they also raise some concerns about trial design and statistics that likely will raise flags at FDA and draw an advisory committee review.

Results from the Cougar-302 trial, which was stopped after an interim analysis showed significant improvement in progression-free survival for the abiraterone arm, were reported June 2 at the ASCO meeting in Chicago. J&J’s Janssen Research & Development LLC unit reported March 8 that the study had been halted by its independent data monitoring committee after the second planned interim analysis, but no data were revealed, piquing interest over the results Also see "Zytiga Study Supports Expansion To Early Patient Population" - Pink Sheet, 8 March, 2012.. J&J let it be known that a filing was planned for the second half, making Zytiga a potential challenger for Dendreon Corp.’s Provenge in the desirable early-stage market segment.

The study, which compared Zytiga, a CYP17a inhibitor, plus prednisone to prednisone plus placebo, had co-primary endpoints of radiological progression-free survival and overall survival. It is the first time the rPFS endpoint has been used in a large multi-site Phase III trial, and agreement between the independent central reviewer and site investigators was 78%, Charles Ryan, University Of California, San Francisco, the lead investigator on the study, pointed out during the presentation.

At the time of the analysis, when 43% of the anticipated deaths on the study (333 of a planned 773) had occurred, the median rPFS in the control arm was 8.3 months while the Zytiga arm has not yet reached the median for rPFS, already a statistically significant improvement. Median overall survival for the Zytiga group also has not been reached, but was 27.2 months in the control arm; the comparison was not yet a statistically significant advantage, though Ryan deemed it “a strong trend in favor of abiraterone.”

Zytiga also performed well on a number of secondary endpoints, including

median time to initiation of chemotherapy: 25.2 months for Zytiga versus 16.8 months for the control;
median time to deterioration in performance status: 12.3 months for Zytiga versus 10.9 months for control; and
median time to PSA progression: 11.1 months for Zytiga versus 5.6 months for the control arm.
Median time to opiate use for cancer pain was not yet achieved for the Zytiga arm, while the median for control was 23.7 months.

A final analysis for overall survival will be conducted after the 773-death endpoint is reached, but that analysis will now be confounded, noted Susan Halabi, Duke University Medical Center, who gave the talk reacting to the study. Patients on the prednisone arm were able to cross over to the treatment arm after unblinding, meaning the overall survival analysis will be muddied.

“The major concern we have about unblinding during an interim analysis is the bias introduced by the release of the data,” Halabi said. Each time an interim analysis is conducted, the trial loses statistical rigor. COU-302 was designed with three planned interim analyses and stopped after the second, at 43% of events. The third was planned for 55% of events, and Halabi suggested that had the sponsor waited for that analysis, the survival results would have reached statistical significance. The trial used a group-sequential design that was specifically developed for interim monitoring to minimize the role of subjective judgment, she noted. But the frequency of early termination is a source of frustration to regulators and others hoping to see more conclusive data.

“The price we paid when we released the data early, and we terminated the trial, is you end up with a biased estimate of clinical benefit,” Halabi said. “The observed hazard ratio was 0.75, with a confidence level between 0.61 and 0.93. Does this justify the IDMC to release the trial?”

Now that the trial is unblinded and patients have crossed over and/or gone on to subsequent therapy, “follow-up data will be biased.” Though the interim finding looks good, there is pressure for it to hold up as the data matures; “in reporting the results, we have to make sure the observed hazard ratio is going to be larger than 0.75,” she stressed.

The relationship between PFS and overall survival in this setting remains “a riddle,” she said, especially as there is discordance between PFS and prostate-specific antigen findings. “Identifying intermediate endpoints of OS remains a hot topic in patients with advanced prostate cancer.” Halabi suggested that in the absence of validated intermediate endpoints of overall survival, actual overall survival should be the primary endpoint for future trials in the castration-resistant prostate cancer population.

In the 302 study, the p-value for the overall survival endpoint was 0.0097. The p-value required for statistical significance at the 43% of events time point was 0.0008. “This raises the question why the independent DMC would terminate the trial and make the results public,” Halabi said. “The decision to stop a trial early is complex, requiring both statistical and clinical judgment,” she allowed, but “stopping a trial too early may fail to persuade the scientific community to change clinical practice.”

The decision to stop the trial early will be dissected during FDA review, and the design and significance of the findings will presumably be topics the agency brings up for discussion by the Oncologic Drugs Advisory Committee. The statistical issues around the overall survival findings could preclude any mention in labeling. But while these issues may draw regulatory scrutiny, the avid reception at ASCO and on Wall Street shows that the data satisfy the cancer community.

Zytiga was approved in 2011 for treatment of chemotherapy-resistant metastatic castration-resistant prostate cancer, with a survival benefit. (The Phase III trial supporting that application also was unblinded early after a planned IDMC analysis, though the survival benefit achieved statistical significance.) To some extent, even if overall survival cannot be proven in the pre-chemo setting, it should have an easier time gaining acceptance from clinicians because of its clear value in a later line of therapy.

Dendreon’s Provenge holds the only approval in the pre-chemotherapy space, and it has demonstrated a survival benefit of 4.1 months. But the pricey therapeutic cancer vaccine has had many challenges gaining traction in the market ([A#14120508007]).

J&J gained abiraterone in its 2009 acquisition of Cougar Biotechnology Inc. Zytiga has not only shown dominance in the increasingly competitive CRPC market, J&J is well on its way to establishing its role in other lines of therapy. In addition to the pre-chemo data, during a pre-meeting press conference, ASCO released data from a second trial showing abiraterone was effective in one third of men with localized, high-risk prostate cancer in the neoadjuvant setting ([A#29120518003]).

Zytiga Pre-chemo Data Impress At ASCO, But Early Trial Termination Could Be An Issue

More from Archive

More from Pink Sheet