Roche looks unlikely to be able to move its DLBCL drug Columvi earlier in the treatment sequence after experts in an FDA advisory committee joined the agency in questioning the regional imbalance of clinical trial data. What’s more, both FDA Commissioner Martin Makary, M.D., and longtime FDA oncology chief Richard Pazdur, M.D., have a new message for drug developers.
A panel of experts on the FDA’s Oncologic Drugs Advisory Committee voted 8 to 1 Tuesday that results from Roche’s phase 3 Starglo trial are not applicable to a U.S. patient population. The trial evaluated the combination of Columvi and the chemotherapy regimen GemOx in second- or later-line transplant-ineligible DLBCL.
A patient representative cast the only yes vote.
Inconsistent results
The experts’ concerns stemmed from an apparent overall survival data imbalance between Asian and non-Asian regions in the Starglo trial. Overall, the trial showed a statistically significant 38% survival improvement for Columvi-chemo over a Rituxan-chemo combo (R-GemOx) during a longer-term follow-up. The magnitude of benefit was 41% at the primary analysis, which was the focus of Tuesday’s meeting.
But an FDA analysis found that Columvi’s life extension benefit was markedly stronger, at 61%, in Asian countries, whereas its risk of death was 6% higher than the control arm in non-Asian countries.
Roche pegged the inconsistency to the control arm outperforming historical data in non-Asian countries. Historically, the median overall survival length of R-GemOx has been in the range of eight months to 13.5 months, according to Roche. In the Asian population of Starglo, the number landed in that range, at 8.2 months. However, the number reached 27.8 months for the non-Asian group. By comparison, the overall survival performance of Columvi-chemo was consistent across regions.
Roche further attributed the surprise showing to the wider usage of highly efficacious subsequent therapies such as CAR-T therapies in non-Asian countries, as well as a higher proportion of patients with primary refractory disease—a high-risk factor—in the Columvi arm in North America.
In the control arm, 40% of patients in the non-Asian regions received a novel subsequent treatment, versus 24% in Asian countries. The numbers were 19% and 6% in the Columvi arm.
After adjusting for novel follow-on therapies, the overall survival outcome in ex-Asia countries moved in favor of the Columvi arm with a 27% death reduction, according to a Roche analysis.
However, the FDA pointed out additional imbalances on efficacy endpoints that were unlikely to be affected by subsequent treatments. For example, Columvi performed worse in non-Asian countries compared with the Asian region on the measures of progression-free survival and complete response rate, although those comparisons remained in the Roche drug’s favor.
Low US representation
During Tuesday’s public meeting, the FDA went beyond the efficacy analysis to also question the composition of the trial participants and the relevance of Starglo’s control arm design.
Based on utilization and claims data, the FDA estimated that only 2% to 8% of second-line DLBCL patients in the U.S. use R-GemOx. Roche’s estimate for the number was 18%, whereas all types of Rituxan-chemo combinations account for nearly half of cases.
As Starglo enrolled transplant-ineligible patients, the FDA dug into the reason for ineligibility and found that “patient refused transplant” was cited in 65% in patients enrolled in Asian countries versus 7% in non-Asian countries.
The FDA chose to highlight transplant refusals because, as a subjective measure, it’s challenging to characterize who those patients are, an FDA official noted during the meeting.
In response, Roche noted that of the total 95 patients who refused transplant before enrolling in Starglo, refusal was the sole reason for only 29 patients.
The Starglo trial enrolled 48% of its total population in Asia, including 34% Chinese patients, whereas U.S. patients only made up 9% of the trial population. Roche attributed the pattern to slow enrollment in the U.S. because of the pandemic when the study was launched about five years ago.
In an unconventional move, FDA Commissioner Martin Makary, M.D., together with newly appointed Center for Biologics Evaluation and Research Director Vinay Prasad, M.D., dropped by early during Tuesday morning’s meeting, where Pazdur was present.
“Dr. Pazdur and I have talked about the importance of your contributions and a few of these emerging issues that we have to consider as an agency,” Makary said. “What do we do with trials where all of the enrollees are in China? Are these trials that we can make big decisions [on] in our national interest? So those are questions that are in front of you, among many other questions.”
Despite all the hypotheses that Roche came up to explain the imbalances, the FDA expert panel was unconvinced.
“It boils down to one thing: We don’t know,” Christopher Lieu, M.D., from University of Colorado Cancer Center, said before voting against Roche.
“When you deal with this much of a discrepancy in overall survival benefit, then you’re certainly not proving that it works in a North American population,” Lieu said. “And here, it’s really questionable as to whether there’s even, like a trend towards either no difference or even harm.”
Mark Conaway, Ph.D., from the University of Virginia, said he was already unsure of the applicability of Starglo’s results to a U.S. population when he knew there were just 25 U.S. participants.
“And when you add in the part about the differential results, I have even less confidence about the generalizability,” he said.
The U.S. FDA is looking for increased enrollment of clinical trials in the U.S., Pazdur said.
“Unfortunately, if you take a look at all the oncology trials that come to us, only about 20% of the population is derived from the United States,” Pazdur said during the meeting. “We’d like to see robust increased enrollment in the United States.”
“I’d like to make sure that people understand this is going to be an area that the Oncology Center of Excellence is looking at quite closely to ensure that control arms are basically applicable to the United States and represent really current standards here in the United States, not just the weakest control arm that we have that’s not referable so much to this study,” he said.
“People are developing drugs for marketing in the United States, so it should address our interest here in the United States,” Pazdur continued. “And I just want to make sure that this is clearly delineated here.”
Editor's Note: The story was updated to show that the Columvi regimen showed a 41% overall survival benefit at the primary analysis