On June 17, the FDA approved checkpoint blocker Keytruda (pembrolizumab, Merck & Co. Inc.) “for the treatment of adult and pediatric patients with unresectable or metastatic tumor mutational burden-high (TMB-H) [≥10 mutations/megabase (mut/Mb)] solid tumors, as determined by an FDA-approved test, that have progressed following prior treatment and who have no satisfactory alternative treatment options.”

Keytruda itself is already approved for a subset of tumors with TMB-high, those who have microsatellite instability (MSI-high or dMMR).

But the new approval opens up a much larger patient cohort for Keytruda.

“TMB is a measure of the somatic mutation rate,” David Fabrizio told BioWorld, and “MSI is one of many [conditions] that can lead to a TMB-high state.”

Fabrizio is the vice president of translation strategy of Foundation Medicine Inc., whose Foundationone CDx test was approved as a companion diagnostic to measure TMB and help identify patients who may be appropriate for treatment with Keytruda.

MSI-high is a proxy for a high somatic mutation rate due to a deficiency in DNA mismatch repair. But there are tumors that have DNA repair defects without being MSI-high. Environmental exposure to mutagens, including cigarette smoke and UV radiation, can also result in TMB-high tumors. In the Keynote-158 trial that was the basis for the new approvals, only 15 of 120 TMB-high patients were MSI-high.

Keytruda and Foundationone CDx received accelerated approval based on a higher response rate in TMB-H patients. In that trial, whose results were presented at the 2019 European Society of Medical Oncology (ESMO) Congress, 28.3% of TMB-high patients and 6.5% of TMB-low patients responded to Keytruda.

In 2017, MSI-high tumors were the first subset of TMB-high tumors to receive a tumor-agnostic FDA nod because “the MSI tests didn’t require complex technology like next-generation sequencing” (NGS), Fabrizio said. The capability to test for MSI-high status “arose well before NGS became widely available and widely adopted.”

Foundation’s test sequences a panel of 324 genes to look for point mutations, copy number alterations, insertions/deletions (indels) and fusions.

TMB is itself a proxy for how many neoantigens a tumor is likely to have. Large numbers of neoantigens, which are specific to tumors, improve the odds that when T cells are freed from inhibition via checkpoint blockade, they will find a tumor-specific antigen to sink their claws into.

Fabrizio credited the company’s success partly to the TMB harmonization project, a multi-institutional consortium led by the nonprofit Friends of Cancer Research. The goal of the project is to create consistent methods and standards to assess TMB, so that the decision of whether a patient is a good candidate for immunotherapy will not depend on which institution they are being treated at, or which diagnostic test they are being tested with.

Lack of standardization has been one issue, though not the only one, plaguing the use of PD-1 or PD-L1 expression level as a standardized diagnostic test for Keytruda and other PD-1/PD-L1 inhibitors.

The TMB harmonization project, whose members include the National Cancer Institute, the FDA, more than 20 diagnostics and biopharma companies, and a dozen academic institutions under the leadership of the nonprofit Friends of Cancer Research, most recently presented data on calibration approaches for the alignment of TMB measurements on clinical samples at the AACR virtual annual meeting in April, and has published multiple papers and conference presentations with guidelines for universal definitions and validation standards

“Without that, I really don’t think we would be where we are today,” Fabrizio said.

Foundation has a liquid biopsy that is currently under priority review by the FDA, and part of the company’s research is into understanding the relationship between blood and tissue biopsies.

Fabrizio said that “another area where we want to have a lot of focus is in the monitoring space,” for example, to assess tumor burden levels and trends postsurgically, and catch relapses early.

Earlier in the research endeavor, researchers at Stanford University reported in the June 15, 2020, online issue of Nature Machine Intelligence that they had developed an algorithm, Image2TMB, that could classify lung adenocarcinomas as high- or low mutational burden based on reading histopathological slides. The authors wrote that their approach “has the predictive power of a targeted sequencing panel of ~100 genes.” The authors also showed that their algorithm was able to estimate tumor mutational burden along a continuous scale, albeit with lower accuracy.

Foundation is exploring AI approaches, along with many others, as a diagnostic tool, though Fabrizio said that he thinks it will need to be “complemented with comprehensive genomic profiling” to work well.

Nevertheless, he said, “that is the point of entry for any diagnostic test – generating a slide from a biopsy. Using that initial data to make an inference about something that’s as important as TMB is a good idea.”

The patients most likely to benefit from an AI approach – which, the Stanford team pointed out, can be performed in minutes – are relapsed patients with advanced disease who have exhausted their other options.

“If you think of the reality of that situation,” Fabrizio said, “every day probably matters.”

No Comments