乌鸦传媒

A pair of PD-1 immunotherapies used in advanced hepatocellular carcinoma (HCC) met sharply different receptions from FDA advisers reviewing accelerated approvals that failed to show benefit in confirmatory trials.

In considering the two agents for previously treated HCC, the Oncologic Drugs Advisory Committee (ODAC) on Thursday unanimously recommended that FDA maintain its accelerated approval of pembrolizumab (Keytruda), by a vote of 8-0, but split over nivolumab (Opdivo), with a 5-4 vote in favor of revoking the indication.

The divided decision largely stemmed from whether ongoing trials would offer new evidence to support the indications. ODAC members took a wait-and-see approach with pembrolizumab -- given the imminent readout of KEYNOTE-394, a placebo-controlled second-line trial being conducted in Asia. But with nivolumab, they remained unconvinced that any ongoing trial from drugmaker Bristol Myers Squibb (BMS) would adequately provide insight for the current single-agent indication.

"There really were no post-marketing trials planned that specifically addressed efficacy of [nivolumab] monotherapy in the second-line setting," said Pamela Kunz, MD, of Yale University School of Medicine in New Haven, Connecticut, discussing why she voted against nivolumab maintaining the indication.

The recent first-line approval of PD-L1 inhibitor atezolizumab (Tecentriq) in combination with bevacizumab, a VEGF inhibitor, played a large role in the discussion as well, with members pointing out that up to 20% of HCC patients remain ineligible for bevacizumab due to the risk of bleeding or clotting events, leaving the door open for the prior standard, sorafenib (Nexavar), then subsequent treatment with immunotherapy.

"There was an enormous drought in the therapeutic landscape for HCC between the bringing us sorafenib, and then what I view as the paradigm shift in IMbrave150, with atezolizumab and bevacizumab," noted Mark Lewis, MD, of Intermountain Healthcare in Murray, Utah, discussing his vote in support of pembrolizumab. "With the imminent maturation of KEYNOTE-394, it seems to me premature to withdraw the approval at this time."

Pembrolizumab Sails By, for Now

The 2018 for pembrolizumab in HCC was based on findings from the single-arm KEYNOTE-224 trial, which included 104 patients with Child-Pugh class A liver impairment who progressed on or were unable to tolerate sorafenib. The study notched an overall response rate (ORR) of 17%, with more than half of the patients maintaining responses for a year or longer.

, the confirmatory phase III trial, appeared positive at a glance, but missed on its dual co-primary endpoints per the study's prespecified statistical design.

The trial randomized 413 advanced HCC patients previously treated with sorafenib to best supportive care plus either pembrolizumab or placebo. Median overall survival (OS) reached 13.9 months with pembrolizumab, as compared with 10.6 months with placebo (HR 0.78, 95% CI 0.61-0.99), while progression-free survival was 3.0 months versus 2.8 months, respectively (HR 0.72, 95% CI 0.57-0.90). The ORR was 18.3% with the checkpoint inhibitor, with a median duration of response of 13.8 months.

"Although the statistics on the confirmatory study did not meet the specified endpoint, the results seem remarkably similar to the earlier study, and while the response rate is relatively low, the duration of responses -- and this is true of immunotherapy in general -- the responses are often remarkably long," said ODAC Chair Philip Hoffman, MD, of the University of Chicago.

"I was also persuaded by the unmet need question," he added. "There are still a fair number of patients who are not going to be suitable for treatment with bevacizumab upfront."

KEYNOTE-394, results of which are expected in the coming month, "will either clearly support the continued indication," said Christopher Lieu, MD, of the University of Colorado in Aurora, "or simply prove that pembrolizumab may not be effective enough to reach prespecified statistical boundaries."

No Such Luck for Nivolumab

"I do think there are patients who benefit from nivo monotherapy, I just think that the data show that the benefit is not there for the population as a whole," said Anthony Sung, MD, of the Duke Adult Blood and Marrow Transplant Clinic in Durham, North Carolina, explaining his vote against retaining the indication.

The 2017 accelerated approval of single-agent nivolumab was based on a that involved 154 HCC patients with Child-Pugh class A cirrhosis previously treated with sorafenib. The confirmed ORR was 14.3%, with 55% of patients achieving responses of at least a year.

But the confirmatory trial, , a first-line trial in unresectable HCC, narrowly failed to show an OS advantage with nivolumab versus sorafenib (HR 0.85, 95% CI 0.72-1.02, P=0.075), with roughly 20% of patients in the control arm going on to receive an anti-PD-1 agent.

"This indication is certainly more complicated than our prior discussion, and that's mainly because the confirmatory study that does not reach statistical significance is in a completely different line of therapy," said Lieu, who in fact did vote to keep the approval. "There is a late but somewhat believable separation of curves, but the lack of statistical significance is certainly problematic."

Further complicating sponsor BMS's position was the more recent in HCC, which demonstrated an ORR of 33%.

BMS maintained that single-agent nivolumab might still be appropriate for less-fit patients unlikely to tolerate the added toxicities associated with ipilimumab, a CTLA-4 checkpoint inhibitor, and should therefore stay on the market, which perplexed Richard Pazdur, MD, acting director of the Office of Oncologic Diseases at FDA's Center for Drug Evaluation and Research.

"We can't contrive or just make up an indication here," said Pazdur. "Is the data there in the single-arm nivo study of patients that could not tolerate ipi-nivo? That's what we're being asked here. Could I see that data?"

But BMS didn't have it on hand.

"This was an extremely nuanced and difficult decision," said Lewis, who voted in favor of revoking the indication. "As a clinician, I have to tell you there's a certain je ne sais quoi when I look at a patient in front of me, as to whom I think will tolerate an immunotherapy doublet."

"That only comes with practice and experience, and that is again part of being a doctor, it's part of being an oncologist," he added. "The problem here is trying to distill that gestalt down into data and specifically into eligibility criteria."

Lewis said he didn't see any upcoming data that would "specifically address the risk-benefit calculus" with nivolumab as a single agent in this setting, with the next trials to readout involving nivolumab-ipilimumab versus sorafenib-lenvatinib (Lenvima) in the advanced setting (CheckMate 9DW) and nivolumab versus placebo as adjuvant therapy for patients undergoing surgery (CheckMate 9DX).

Thursday's session was part of a 3-day ODAC review of accelerated approvals for checkpoint inhibitors that had confirmatory trials that failed to demonstrate further benefit -- so-called dangling approvals -- and included indications in triple-negative breast cancer, metastatic urothelial cancer, and gastric and gastroesophageal junction cancers.

Although the FDA is not required to follow its advisory committees' recommendations, it typically does. Split votes are generally taken as no recommendation one way or the other.

Comment