Skip to main content
Premium Trial:

Request an Annual Quote

At AACR, Immuno-Oncology Clinical Trials Offer Mixed Messages on Predictive Biomarkers


CHICAGO (GenomeWeb) – Results from several highly anticipated immuno-oncology clinical trials were released yesterday at the annual meeting of the American Association for Cancer Research. All the studies included some analysis of molecular response biomarkers, but with variable takeaways for the future of companion testing in this rapidly accelerating field.

Amongst the studies, three of which also appeared simultaneously in the New England Journal of Medicine, was a report from Bristol-Myers Squibb on the efficacy of a combined treatment regimen of Opdivo (nivolumab) and Yervoy (ipilimumab).

The company shared data from the CheckMate 227 trial that it had hinted at earlier this year, demonstrating that the immuno-oncology (IO) drug combination resulted in significantly better progression-free survival for patients with a high tumor mutational burden as assessed by Foundation Medicine's comprehensive targeted next-generation sequencing panel.

BMS didn't begin CheckMate 227 with plans to look at TMB as one of its endpoints, but was able to add an analysis of mutational burden after retrospective data from other trials began to accumulate showing that it might be a crucial biomarker for immunotherapy response.

In the new data presented at AACR, researchers demonstrated that TMB effectively and significantly stratified responders to the IO combination. Among patients with tumors that had more than 10 mutations per megabase (a cutoff point developed in a separate trial, CheckMate 568, and applied in CheckMate 227) the percentage of patients with progression-free survival at one year on combined immunotherapy was triple that of those treated with chemo.

In contrast, for patients below the TMB threshold there was essentially no difference in PFS between the immunotherapy and chemo arms of the trial.

Notably, although earlier studies of immuno-oncology monotherapy have indicated that PD-L1 remains important — with patients who are both TMB-high and high PD-L1 expressers doing much better than those who have high TMB but no PD-L1 — the superiority of the Yervoy-Opdivo combination in CheckMate 227 was strong across various subgroups, including those who were completely negative for PD-L1.

Discussing the results, Memorial Sloan Kettering Cancer Center oncologist Matthew Hellmann said that the findings contribute to the establishment of TMB as an independent and reliable biomarker.

Although PFS was improved with the immunotherapy combination in an all-comer population in the trial — overall, about twice as many IO-treated patients as chemo-treated were still responding at one year regardless of TMB levels — "the application of TMB effectively identified a subgroup of patients with true clinical benefit," added Naiyer Rizvi, director of thoracic oncology at Columbia University Medical Center.

As evidence like the results from CheckMate-227 continue to suggest that TMB testing should be incorporated into practice, Rizvi also highlighted how wider clinical implementation will require changes in testing practices to move comprehensive sequencing up front in patient care, rather than beginning a workup with IHC and FISH assays, or smaller targeted panels.

Encouragingly, he argued, the availability of Foundation Medicine's test, and the quick turnaround the company has pushed itself to offer, present a good argument for feasibility of more widespread TMB testing.

Yale Pathology Professor David Rimm said that despite the promising results of CheckMate 227, there are reasons to be cautious about clinical implementation of TMB testing.

BMS's announcement about pursuing assays with Illumina after having used Foundation Medicine in its trials so far illustrates the unanswered questions, he added. "The Illumina TruSight Oncology 500 panel is a completely different assay. So, is 10 [mutations per megabase] still going to be the cutoff point?" Rimm asked.

"Lots of different labs are going to want to do this, and we have to make sure pathologists know what they are up against," in terms of standardization and concordance. "I'm not even sure TMB is standardizable, but it's definitely not standardized yet.

The question of cutoff point is a significant one for Foundation Medicine as well if it is to transition the TMB readout from its LDT test to future companion diagnostic indications.

The FDA approved the company's FoundationOne CDx last fall, but only a handful of genes on the larger panel actually have agency-approved companion indications.

That doesn't mean it can't be used clinically. Foundation has been marketing TMB analysis as part of its LDT service for more than a year and reported last September that it had analyzed tens of thousands of patients since making the analysis part of its FoundationOne report.

When performing the non-FDA approved version of FoundationOne in its CLIA lab, Foundation has reported TMB both quantitatively and qualitatively, but with the CDx version of the test now approved by FDA, the company has said it is planning to report the biomarker as a numerical value only. In this way, if TMB is at some point approved as a companion to a specific therapy, Foundation could adjust the format of the report to reflect a specific cutoff based on FDA drug labelling.

BMS has not made a defined commitment to pursuing FDA recognition of the Foundation Medicine TMB readout alongside the Yervoy-Opdivo combination, especially not exclusively. In fact, the company said last week that it was committing to a separate development deal for IVD TMB assays with Illumina.

Steve Averbuch, head of the BMS precision medicine group, declined to discuss any specific plans for discussions with regulators, but said that the company believes that the data presented yesterday from CheckMate-227 provides the clinical validation data that would be necessary to support a CDx approval for Foundation's TMB readout.

In respect to the future of clinical TMB testing and its plans with Illumina, Averbuch said that as a drug company, BMS wants to have "appropriate, high-standard, validated TMB testing broadly available to patients, full stop."

"As a therapeutic company, we are not wedded or committed to any one model of providing TMB testing to patients worldwide," he added, saying that both Foundation Medicine and future Illumina TruSight-based IVDs are valid solutions, and their use in the clinic is going to be subject to the needs of different marketplaces around the world.

For Foundation, a companion diagnostic path for TMB isn't tied to BMS alone. The company has also said that it is advancing a blood-based version of its TMB test as a companion to Roche/Genentech's immunotherapy drug Tecentriq (atezolizumab) in first-line treatment of non-small cell lung cancer patients.

Averbuch said that BMS is now collecting samples for both tissue-based and blood-based TMB analyses across its clinical trials, along with a variety of other biomarkers, including PD-L1 and gene expression.

Because TMB is, in essence, a simple quantification of tumor mutations, there are also multiple other companies that have discussed pursuing assay commercialization, and potentially, innumerable labs that could create their own tests for mutational burden.

Johns Hopkins and Memorial Sloan Kettering were some of the first academic institutions to publish on the predictive nature of TMB in relation to immunotherapy response.

Personal Genome Diagnostics, originally a Hopkins spinout, announced in February that it has reached an agreement with MSKCC to develop and commercialize tissue- and blood-based products that include assessment of TMB status.

PGDx CEO Doug Ward said in a statement at the time that the IP agreement with MSKCC reinforces other existing license positions with Hopkins, and that the firm is "confident that the combined patent estate adequately protects our extensive investment and development plans for [its] TMB-enabling tests."


As the future of TMB in clinical practice appears to be coalescing, based on presentations at AACR and other clinical research meetings, the utility of other immunotherapy biomarkers like PD-L1 is becoming hazier.

This was illustrated by the CheckMate 227 results, which were significant regardless of patients' PD-L1 status, but also in several other clinical trials reported at the meeting, which also raise questions about the necessity of PD-L1 testing as the indications for immunotherapy expand.

In one study, Merck's Keynote-189, researchers demonstrated that the company's immunotherapy Keytruda (pembrolizumab) plus chemotherapy offered significantly longer overall survival and progression-free survival compared with chemotherapy alone for newly diagnosed metastatic non-squamous NSCLC patients.

Keytruda plus pemetrexed and carboplatin-based chemotherapy was approved by the FDA in this indication based on the phase II cohort G of the KeyNote-021 study last year, but discussants at the AACR session said that the treatment regimen has not yet been widely adopted, as clinicians awaited results of the newly published Phase III Keynote-189.

Although the study recruited patients regardless of PD-L1 status, investigators did analyze the results across PD-L1 subcategories, concluding that although there were differences, with high expression of PD-L1 conferring even better responses, the benefit of the IO plus chemotherapy regimen remained significant in every group, including PD-L1 negative patients.

Discussing the results, Yale Cancer Center Professor Roy Herbst raised one caveat to this, which is that while an overall survival benefit was statistically significant for all the PD-L1 groups, the difference in progression-free survival in PD-L1 negative patients didn't quite reach significance.

The trial also didn't address what might be a crucial question for clinicians, which is how responses in patients with high PD-L1 expression in this trial might compare to responses seen with Keytruda monotherapy, which is also approved for first-line treatment of lung cancer patients who test positive for PD-L1 expression above 50 percent.

"We already use pembro in these patients with greater than 50 percent PD-L1 expression, so should we now be switching to this combination?" Herbst asked.

Based on a quick comparison of the data from the Keynote-189 trial and the earlier Keynote-24 study, Herbst said that certain metrics look identical while others may be a bit better for the combined treatment. Survival at one year was 73 percent in both studies. But the response rates and hazard ratio both look a bit better in Keynote-189.

Another limitation of the trial was that it excluded patients with ALK and EGFR mutations, so it can't be used to answer what Herbst argued remains an important open question — whether these patients might also benefit from a combination immunotherapy and chemotherapy approach.

Another study presented at the meeting did provide some hints at this, though. Researchers shared an analysis of efficacy across a variety of biomarker groups, including various PD-L1 expression subsets, and EGFR/ALK mutations from Roche/Genentech's Phase III IMpower-150, which tested a combination of the company's Tecentriq (atezolizumab), Avastin (bevacizumab) and chemo compared to Avastin and chemo alone.

One important aspect of the analysis was a retrospective comparison of different PD-L1 assays, prompted by efforts like the BluePrint project, which have found that the SP-142 assay initially applied in the trial identifies fewer PD-L1-positive patients than other tests developed alongside other immunotherapies.

Researchers in the trial were able to look at SP-263 antibody testing in about 500 patients from the trial, and despite evidence of non-concordance of the assays in other research, they reported that essentially all the PD-L1 expression subgroups defined by either assay, including PD-L1-negative patients, showed improved outcomes on the triple-therapy combination.

Patients with EGFR and ALK alterations also had better progression-free survival on the full combination compared to just chemo and Avastin.

Finally, in a third study exploring a new use of immunotherapy in the neoadjuvant lung cancer setting, researchers reported that patients with stage 1-3 NSCLC had higher pathologic response when given BMS' Opdivo before surgery regardless of PD-L1 expression. In contrast, TMB did appear to predict the rate of response.

Despite the results though, presenters for both the Keynote and IMpower trials as well as for the BMS CheckMate-227 study all said that they believe that PD-L1 analysis still has relevance and should continue to be performed in the clinic.