NEW YORK (GenomeWeb) – A study evaluating the combination of Bristol-Myers Squibb's immunotherapies Opdivo (nivolumab) and Yervoy (ipilimumab) may support regulatory approval for the regimen as an option for lung cancer patients with a large number of tumor mutations that are identified by Foundation Medicine's next-generation sequencing companion diagnostic.
This week, BMS executives announced that with the help of Foundation Medicine's FoundationOne CDx they've been able amend the design of the CheckMate-227 trial in line with the latest understanding of tumor biology and evaluate the efficacy of the Opdivo/Yervoy combination based on patients' tumor mutational burden (TMB), a relatively new biomarker for personalizing immunotherapy. Without reporting the actual data from CheckMate-227, BMS said the trial has shown that advanced NSCLC patients with high tumor mutational burden treated with Opdivo and Yervoy as a front-line option had significantly higher progression-free survival compared to those on chemotherapy ― regardless of their PD-L1 expression status.
The US Food and Drug Administration has approved Opdivo in nine cancer settings, including for previously treated, advanced NSCLC, and as a single agent and in combination with Yervoy for metastatic melanoma patients. For NSCLC and melanoma patients considering Opdivo, the PD-L1 IHC 28-8 pharmDx test, developed by Agilent Technologies subsidiary Dako, is FDA-approved as a complementary diagnostic for guiding treatment strategy.
Meanwhile, the FDA last year approved FoundationOne CDx, which gauges 324 genes, as well as microsatellite instability and TMB. Specifically, the agency approved companion diagnostic indications for a handful of genes that can be used to identify best responders to 15 FDA-approved treatments in five tumor types. The other genes on the panel, including genomic signatures, such as TMB, don't have the agency's blessing as companion diagnostics, but can still be used to inform the clinical management of patients. The FDA-approved label for FoundationOne CDx states that the "clinical validity of TMB ... has not been established."
In light of BMS' announcements regarding CheckMate-227 this week, Dave Fabrizio, who leads cancer immunotherapy efforts at Foundation Medicine, said the study data support tumor mutational burden as a predictive biomarker for combination immunotherapy in NSCLC. "It's clear the data are registration enabling," Fabrizio said. "We're eager to extend the conversation with healthcare authorities to explore this option."
To date, Foundation has only retrospectively studied the role of TMB as a predictive biomarker for immunotherapy. “This is the first time that we're able to prove the hypothesis and demonstrate prospectively that TMB as a biomarker for immunotherapy has clinical utility,” he said.
Foundation's collaboration with BMS began in early 2017, around the time the drugmaker announced that it would not seek accelerated approval of the Opdivo/Yervoy combination in first-line lung cancer treatment.
Several months before, the CheckMate-026 study had failed to meet its primary endpoint of progression-free survival and showed similar overall survival in PD-L1-positive NSCLC patients treated with Opdivo or platinum-based doublet chemotherapy. The setback in the profitable first-line NSCLC setting was a precision oncology fumble, industry observers said at the time, because BMS had chosen to enroll patients whose tumors expressed PD-L1 in 5 percent or more of cells ― a cutoff that favored a larger intent-to-treat population at the cost of homing in on which patients would benefit from treatment.
"This was really around the time that BMS was gathering data on CheckMate-026 … and we were learning a lot about how tumor mutational burden could have helped predict response by doing a retrospective analysis within that study," Fabrizio said.
When investigators from CheckMate-026 published their data in the New England Journal of Medicine in mid-2017, they discussed an exploratory analysis involving around 300 patients to assess how TMB impacted outcomes. They retrospectively performed exome sequencing on samples from this subset and found that patients on Opdivo who were TMB high ― defined as having 243 or more mutations ― had higher response rates and longer progression-free survival than those on chemotherapy. Although overall survival was still similar between the Opdivo- and chemo-treated groups in the TMB-high subset, the authors noted that 68 percent of patients in the chemo arm crossed over to receive Opdivo.
"As those data were being understood, CheckMate-227 was ongoing," looking at the Opdivo/Yervoy combination in NSCLC, Fabrizio recalled. "We were talking to BMS quite a bit about TMB and how to calculate it, and understanding the relationship between whole-exome sequencing and the FoundationOne CDx calculation of TMB."
Foundation performed analysis to show the correlation between TMB as determined by its test and by exome sequencing, which helped BMS evolve its regulatory strategy for the Opdivo/Yervoy combination, and "understand that TMB was perhaps a more predictive biomarker for immunotherapy in first-line lung cancer than immunohistochemistry staining of PD-L1," Fabrizio said.
The 2,500-patient CheckMate-227 trial was initially divided into two sections ― in the first section researchers are evaluating the Opdivo/Yervoy combination versus Opdivo monotherapy and doublet chemo in NSCLC patients with PD-L1 expression in 1 percent or more tumor cells; and in the second section they are studying the combination, Opdivo plus chemo, and chemo alone in PD-L1 non-expressers.
After consulting with the FDA, BMS decided to amend the study while it was still blinded so it could track overall survival in PD-L1 expressers and progression-free survival in patients with high TMB regardless of PD-L1 status. The overall survival primary endpoint in PD-L1 expressers is slated to read out later this year or early next year. The progression-free survival primary endpoint in high TMB patients will be presented at an upcoming medical conference. BMS also plans to track overall survival in the TMB subpopulation, but that will take longer to read out.
"This is both a TMB story and an Opdivo/Yervoy combination story. Both are important," BMS Chief Scientific Officer Thomas Lynch said during a call to relay the study's success and 2017 financials to analysts. Yervoy has shown to be efficacious across tumors types and based on the data seen in CheckMate-227 and other studies evaluating the combination, Lynch said Yervoy was "a big part of the efficacy," and it was not due to just Opdivo.
At the same time he highlighted TMB as the kind of predictive biomarker that doctors are asking for to help them identify which patients will benefit from immune-oncology treatments. Oncologists have been underwhelmed by PD-L1 expression status as a predictive biomarker for immunotherapy, since it doesn't allow for neat segmentation of responders from non-responders.
But, because PD-L1 testing is still used by doctors to determine which patients should receive Opdivo, "there are some people not getting immunotherapy because they are PD-L1 negative," Lynch said. "TMB gives us the opportunity to identify patients who clearly benefit from low-dose Yervoy added to Opdivo [in NSCLC]."
BMS defined high TMB as 10 or more mutations per megabase. However, there are different ways to define this cutoff, as one analyst pointed out during BMS' call to discuss CheckMate-227. In fact, when performing the non-FDA approved version of FoundationOne in its CLIA-certified lab, Foundation reports TMB quantitatively (as mutations/megabase) and qualitatively in terms of high (20 or more mutations/megabase), intermediate (between 6 and 19 mutations/megabase), and low (5 or fewer mutations/megabase). Fabrizio estimated that around 7 percent, 35 percent, and 55 percent of patients fall into the high, intermediate, and low categories, respectively.
Comparatively, BMS' TMB high cutoff in CheckMate-227 of 10 or more mutations/megabase would include 45 percent of NSCLC patients in the study. (The study excludes NSCLC patients with EGFR and ALK mutations.) Lynch noted during the call that BMS is evaluating outcomes according to TMB status across some 100 studies, which might reveal different cutoffs depending on the setting. Fabrizio also expects that cohort-specific or study-specific cutoffs will be pursued for TMB.
As Foundation transitions from reporting TMB within the CLIA version of the test to now evaluating it in a CDx context, the company is planning to report the biomarker according to the cutoff that is established under regulatory approval. "We're planning to report TMB on the front page of that report as a numerical value only, without the qualitative status of high, intermediate, and low," Fabrizio said. "If we see an approval for TMB as a CDx, however, then the reporting will adjust based on the label information using that specific cutoff." In addition, the FoundationOne CDx report will include a "professional services section" with more detailed information on the application of TMB in additional cancer settings from published studies.
Foundation has also been trying to map the complex biomarker landscape for immunotherapy response. For example, last year, Foundation researchers presented an analysis of how PD-L1 expression, TMB, and genomic alterations interact to impact patients' ability to respond to immunotherapy. The researchers assessed PD-L1 and TMB status of nearly 800 NSCLC tumor samples and identified the samples that were TMB high but had negative PD-L1 expression, or had high PD-L1 expression but were TMB low.
Testing these discordant samples for genomic alterations, researchers found that the TMB high/PD-L1 negative samples tended to also have STK11 loss-of-function alterations that may play a part in helping tumor cells evade an immune attack. Meanwhile, samples that were TMB low/PD-L1 high were enriched for BRAF mutations that may help trigger immune responses against cancer cells.
"Although further ongoing analysis is required, our study suggests that the combination of TMB, PD-L1 staining, and genomics may provide the best prediction of response to checkpoint inhibitors," the authors wrote in an abstract.
Though this is a ripe area of research at Foundation, the focus for CheckMate-227 is essentially TMB, Fabrizio said.
During BMS's earnings call on Monday, market analysts questioned whether a relatively new and complex biomarker like TMB would hinder uptake of the drug in community settings where around 70 percent of patients receive care.
While academic care centers are already familiar with TMB and many are measuring it within precision oncology programs, community oncologists may be less familiar with the biomarker, BMS Chief Commercial Officer Murdo Gordon said during the call. However, BMS plans to focus its educational efforts in the community setting and is also conducting a study in a large community-based population to investigate safety issues related to the Opdivo/Yervoy combination in TMB-high patients.
Market analysts during the call were particularly wary that the progression-free survival advantage seen in CheckMate-227 would not translate into an overall survival benefit for patients given the signals from CheckMate-026. Lynch explained that although BMS hasn't yet seen overall survival data in the study, it will track it according to TMB status. In the meantime, he believes oncologists will be able to rely on the progression-free survival benefit in the TMB-high group, given the size and power of CheckMate-227.
Credit Suisse analyst Vamil Divan wrote in a note to investors this week that while there are examples where the FDA has approved cancer drugs based on highly positive progression-free survival results, there is limited data on whether progression-free survival can be proxy for overall survival in this particular population, and as such, regulators may decide to hold out for overall survival data. "Even with regulatory approval, the commercial impact may take some time as physicians and payers may wish to see how the overall survival data matures," wrote Divan.