NEW YORK – Adding to growing evidence for the association of tumor mutational burden with patient response to immunotherapies, another study describing successful validation of a TMB readout appeared in a journal last week, describing efforts by a group of European pathologists to bring this new biomarker to their lung cancer patients.
Published in the Journal of Pathology, the study was conducted at several institutions, led by a team from the University Hospital Basel in Switzerland. Investigators reported that they were able to stratify their patients such that a tertile-based TMB cutoff system pusing Thermo Fisher Scientific's Oncomine TMB sequencing product, divided patients into groups with significantly different drug responses in a small cohort treated with a variety of checkpoint inhibitors.
When they divided the NSCLC patients in their sample set based on these tertiles, those with a high-TMB designation had a 64 percent response rate to immune checkpoint inhibitor therapy compared to just 29 percent in those fitting into the low-TMB group.
Also reflective of findings in other studies, the highest response rates in the cohort were in patients who had both high-TMB and positive PD-L1 test results, confirming that the two biomarkers appear to be complementary.
Philip Jermann, head of molecular assay development at the Basel hospital's Institute of Medical Genetics and Pathology, said in an interview that his group has been successfully using the TMB Oncomine assay for their patients since validating it as described in the study.
"We are a public hospital, and in pathology we are running routine diagnostics, and of course we want to be up to date with where the state of the art is," he said. "We have been doing PD-L1 for a long time and everybody knows it tends a little bit to predict, but it's not that great of a biomarker, so of course there's a need for more tools."
"We have seen all the recent clinical trials that have evaluated tumor mutational burden … and we were convinced that introducing this could add a benefit, so we wanted to implement this into an in-house test," said Jermann. He added that with the study, they wanted to see if they got similar results in their patients that had been seen in clinical trials.
In the study, Jermann and colleagues measured TMB retrospectively using the Thermo Fisher Oncomine assay in 76 metastatic non-small cell lung cancer patients and used data on response and outcomes to define a cutoff to subcategorize the group into low, intermediate, and high TMB.
Analyzing results against patient outcomes, they found that 64 percent of the high-TMB group responded to immunotherapy treatment compared to just 33 percent of the intermediate group and 29 percent of the low-TMB patients. TMB-high patients also showed significantly longer progression-free survival and overall survival versus their comparators.
Interestingly, the highest response rates were among samples that were both PD-L1 positive and had high tumor mutational load. Overall, PD-L1 status and TMB were not correlated. But, according to the authors, combining PD-L1 expression and TMB increased the predictive power, shifting the area under the receiver operating curve from 0.63 for TMB, or 0.62 for PD-L1, to 0.65 for patients who were double positives.
Jermann noted that like PD-L1 TMB is not a perfect predictor, and the field will need to continue to investigate the biological and molecular determinants of why a patient does or does not respond to immunotherapy. "We are actually already working on the next study … looking at immune cells and the tumor microenvironment, [which] may be an additional piece in the puzzle," he said. "But, we are not satisfied in terms of patient stratification for immunotherapy as-is. Not at all."
TMB has gained attention very quickly, with companies like Foundation Medicine and others now offering actionable calculation of patient status as part of their existing tissue or liquid biopsy sequencing services. As other tests are expected to join the market, and various labs push forward to validate their own in-house workflows, attention is increasingly being paid to the need for standardization of what are likely to be multiple methodologies with different and non-transferrable cutoff points.
In Europe, hospital labs generally favor in-house testing over send-out services like those offered by Foundation and others, but that potentially makes the implementation of TMB even more subject to issues of cross-compatibility.
Jermann and colleagues agreed that the need for standardization across different assays and diagnostic centers is urgent, but they wrote that it is promising that the Oncomine assay they chose to implement is currently being evaluated as part of the larger harmonization efforts already taking place.
"Our patient cohort consisted of a very heterogeneous population. The number of treatment lines prior to ICI treatment as well as the time from biopsy acquisition to ICI treatment was highly variable," the group added. "However, despite this heterogeneity, TMB association with treatment response was significant, underlining the robustness and potential power of [this] biomarker."
In addition to the Basel group, Thermo Fisher has also recently highlighted TMB validation efforts by other groups using its products. A French team, led by Nice Hospital pathologist Paul Hofman, published a study last month in the journal Cancers, for example, concluding that TMB assessments performed in their hospital's clinical laboratory using the Oncomine assay provided similar results and predictive value compared to what was returned by Foundation Medicine.
The company said that it is heartening that both studies, though conducted independently in different countries and addressing patients treated with different immune checkpoint inhibitors, yielded largely the same cutoff to define low, intermediate, and high TMB.
Jermann echoed this observation. "What we really like to see, of course, was that the cutoff that we defined was very similar to what has been observed before for this tumor type. It was a sanity check for us as well," he said.
He also stressed that his team's results and other similar studies must be viewed in a specific context. Increasingly, research is making clear that while TMB may be predictive across different cancer types, each are likely to have their own spectrum, and therefore different cutoff points for therapeutic prediction.
Jermann added that the team is paying closer attention now to developing the best methodology for them to report TMB to their oncologist colleagues. TMB itself is a value, but the clinical interpretation requires connecting that value with a specific cutoff and associated clinical implication.
When Foundation Medicine garnered FDA approval for its FoundationOne CDx, for example, it had to grapple with adjusting its TMB reporting. The company has at different times returned just a quantitative report, without indicating how exactly the numbers in question are associated with higher or lower likelihood of immunotherapy benefit, or has classified results as falling into defined ranges.
For the Basel team, "it is important that physicians are making treatment decisions and not us," Jermann said. As such, when the group analyzes patient samples they currently just give the resulting TMB value. However, they are also considering how to dd more information for doctors, potentially by reporting not just the patient result but also a median for all samples tested in that tumor type, in this case NSCLC.
"Then you can put it in relation to all the other samples from this indication and the oncologist [can] make the decision," he said.