The Food and Drug Administration is looking to its Oncologic Drugs Advisory Committee for help in reconciling the fact that the results of the two pivotal phase III trials of Novartis’s Afinitor (everolimus) for neuroendocrine tumors do not support each other.
The supplemental application coming up for review on April 12 would cover advanced neuroendocrine tumors of gastrointestinal, lung, or pancreatic origin, and is based on two trials, one in pancreatic neuroendocrine tumors (PNET) and one in carcinoid tumors.
In a briefing document for the advisory committee meeting, the FDA highlights the inability of the two trials to support each other as one of four review issues. The impact of informative censoring on both studies, the effect of Afinitor on overall survival, and the use of progression-free survival (PFS) to establish clinical benefit and its impact on overall survival were also flagged as issues.
The FDA was troubled by a midstream change in primary end points in both trials. The original primary end point for both trials, which were designed and approved by the FDA under a Special Protocol Assessment, was PFS as determined by an independent radiology committee. However, in October 2009 Novartis and the FDA met to discuss "discordant evaluations of PFS by the local investigator and the IRC in the carcinoid study during the second interim analysis," the FDA’s briefing document says.
The primary end point of both trials was changed as a result – to PFS as determined by the local investigator in the PNET study, and to PFS as determined by a central adjudication committee in the carcinoid tumor study. At a pre-NDA meeting in August 2010, "FDA noted that both SPA agreements had been invalidated by this change in the primary end point and that the acceptability of the revised statistical plans would be a review issue," the agency said.
Discordant Results, Too
In the end, the results of the two trials were discordant, likely because of the discrepancies in efficacy assessments.
The PNET trial’s primary analysis indicated a statistically significant advantage in median PFS for patients taking Afinitor compared with those on placebo: 11.0 months vs. 4.6 months, per the investigator analysis.
In the primary analysis of the carcinoid tumor trial, based on the central adjudication committee, median PFS was 16.4 months for those taking the drug and 11.3 months for those taking the placebo, but it was not a statistically significant difference. According to Novartis, however, "everolimus, in combination with depot octreotide, has provided benefit for this population, despite the study not meeting its primary end point.
"Informative censoring proved to be a critical methodological challenge," Novartis added, asserting that secondary inverse probability of censoring weights analyses that corrected for this bias "provide evidence for a treatment effect." Censoring occurred in the trials when there was a dispute between the investigators and the radiology committee; the FDA briefing documents cite an example in which an investigator determined a patient to be suffering from progressive disease at cycle 7, while the radiology committee disagreed and censored the patient.
Overall survival (OS) is likely to pose another headache for Novartis, since neither trial demonstrated a benefit for patients taking Afinitor in this area. Both trials used a crossover design, allowing patients who progressed while on placebo to switch to the active arm. In the PNET trial, that ended up happening to 73% of the 203 placebo patients, according to an article in the Feb. 10 issue of the New England Journal of Medicine. The result was to confound the detection of a treatment-related survival benefit, the authors noted.
Novartis’s briefing document also notes that the crossover design in the carcinoid trial made it impossible to determine any OS benefit.
In fact, more patients on Afinitor died in this trial (100, or 46.3% of those in the active arm, compared with 85, or 39.9%, of those on placebo). In the PNET trial, 51 patients, or 24.6% of those in the active arm, died, compared with 50 of those on placebo, also constituting 24.6%. Both results were interim analyses; not enough events have occurred yet in either trial for a final analysis of OS.
This could be an especially big problem for Novartis given that the FDA’s enthusiasm for PFS end points for oncology drugs has been waning.
Afinitor is a multikinase inhibitor that forms an inhibitory complex with the mammalian target of rapamycin (mTOR). It is currently indicated for advanced renal cell carcinoma in patients who have received prior treatment with Pfizer’s Sutent (sunitinib) or Bayer’s Nexavar (sorafenib), for subependymal giant cell astrocytoma, and, under the trade name Zortress, for prophylaxis of kidney transplant rejection.