TMB: A Potential Pan-Cancer Biomarker for Immunotherapy?

Alexander M. Castellino, PhD

January 17, 2019

The search for a biomarker predicting response to immune checkpoint inhibitors (ICI) has been an ongoing race with no candidate sprinting across the finish line. So far, the most studied biomarkers have been the programmed cell death 1 receptor ligand (PD-L1) and tumor mutational burden (TMB), but both have shown varying degrees of success.

A new article argues for use of TMB as a potential pan-cancer biomarker, reporting a significant association between high TMB and overall survival (OS) as well as response to immune checkpoint inhibition across multiple cancer types.

The study comes from the Memorial Sloan Kettering Cancer Center in New York City, with Robert M. Samstein, MD, PhD, as lead author, and was published online January 14 in Nature Genetics.

"This is the largest pan-cancer study that confirms the predictive value of TMB for overall survival and response to ICI beyond smaller studies in certain cancer types where we have previously seen this association," co-corresponding author Luc G.T. Morris, MD, told Medscape Medical News.

"Our study shows that TMB has potential to be a pan-cancer biomarker predictive of immunotherapy response," he said.

Our study shows that TMB has the potential to be a pan-cancer biomarker predictive of immunotherapy response    Luc G.T. Morris, MD

Morris pointed out that until now the association between high TMB and response to ICI was seen in small studies in clinical trial patients with melanoma, lung cancer, and bladder cancer. He emphasized that an important discovery emerging from this latest study is the definition of high TMB appears to be different for each cancer.  

Approached for comments, David Rimm, MD, PhD, director of translational pathology at Yale University, New Haven, Connecticut, said: "I am impressed with the study and especially since the authors do not claim TMB to have 'predictive value.' This is a broad, sweeping paper where the investigators looked at TMB across a range of cancers," he told Medscape Medical News.

Study Details

The researchers examined the association between nonsynonymous somatic TMB (ie, coding mutations) as measured using the MSK-IMPACT assay and OS after treatment with ICI.

The MSK-IMPACT assay uses next-generation sequencing to identify mutations in 468 unique genes and has been approved by the US Food and Drug Administration (FDA).

The study cohort included 1662 patients whose tumors were profiled with MSK-IMPACT and who had received at least one dose of ICI therapy. Of these, 1446 (94%) of patients had stage IV or metastatic disease.

Patients received ICI alone or in combination: 9% of patients received anti-CTLA-4 immunotherapy, 76% received anti-PD-1/PD-L1 immunotherapy, and 16% received a combination of these two types of immunotherapy.

Most patients had received ICI within their FDA-approved indications (nonsmall cell lung cancer [NSCLC], n = 350; melanoma, n = 321; renal cell carcinoma [RCC], n = 151; bladder cancer, n = 215; and squamous cell carcinoma of the head and neck [SCCHN], n = 138).

OS was measured from the time of first ICI dose to the time of death or last follow-up. Median follow-up was 19 months.

In this study, TMB was defined as the number of mutations per megabase (Mb) of DNA.

Within each tumor type, TMB subgroup was defined by percentile. "We took this approach because the median and range of mutational load have been shown to vary across tumor types," Samstein and colleagues note.

Association Between High TMB and Clinical Endpoints

Firstly, across the entire cohort, a higher number of mutations was associated with improved OS. As a continuous variable (ie, across the range of mutational burden), the hazard ratio [HR] for improved survival was 0.985 (95% CI, 0.979 - 0.991). "Our finding shows that there is a dose-response relationship between TMB and clinical benefit as defined by OS," he said.

Even after patients with melanoma and NSCLC were removed from the analysis, the association between TMB and OS remained significant, indicating that the effect was not driven by these cancers alone (HR, 0.982; 95% CI, 0.974 - 0.990).

Secondly, except for glioma, the association was significant for a number of cancers, each having its own cut-offs defining high TMB. For example, the top 20% cutoff for TMB was 52.2/Mb for colorectal cancer (CRC), 17.6/Mb for bladder cancer, 30.7/Mb for melanoma, and 13.8/Mb for NSCLC.

The HR for OS (top 20% TMB vs bottom 80% TMB) was significant for CRC, SCCHN, NSCLC, and trended towards significance for melanoma, RCC, esophagogastric cancer, and bladder cancer.

Morris indicated that the effect was different across tumors and the smaller sample size for some of the tumors may explain the association not reaching significance. "Our results suggest that there is not likely to be a universal number defining high TMB that is predictive of clinical benefit to ICI across all cancer types," Morris told Medscape Medical News. "The cutoff point defining high from low TMB is likely to be different for each tumor type," he said.

Thirdly, a significant association was also shown between TMB and progression-free survival and objective response rates for NSCLC, melanoma, esophagogastric cancer, SCCHN, and RCC.

Finally, in 5371 patients whose tumors were sequenced with MSK-IMPACT but who did not receive ICI, there was no association between higher TMB and improved OS (HR, 1.11).

The researchers noted that the top 20% cutoff defining high TMB in CRC (52.2/Mb) was high. Morris indicated that this is consistent with the patients being treated with ICI for high microsatellite instability (MSI-H). "The high mutational load in this cancer type is driven by MSI," Morris said.

"[The] finding of a significant association with overall survival in a heterogeneous cohort underscores the robustness of TMB as a predictive biomarker, thus suggesting that it is likely to be clinically meaningful," Samstein and colleagues write.

TMB Not Yet Ready for the Clinic

Yale's Rimm, who was not involved with this study, was not surprised with the findings but expressed reservations about use of TMB as a biomarker in clinical practice. "As a pathologist, if tumors were graded from 1-10 (instead of 1-3), the higher grades would have worse outcomes and prognosis compared with lower grade tumors. So it is not surprising that high TMB is associated with a response. In some ways it is similar to a quantitative assessment of tumor grade," he said.

For Rimm, the hype about TMB has yet to be realized. "I would be more of a believer if the investigators had shown their results to be corroborated using multiple concordant assays for TMB," he said.

"This is good science, but there is a concern about taking it to the clinic. The MSK-IMPACT could be seen as a 'black box' test where it is hard to determine what TMB is all about if others who measure TMB don't get the same results," he added.

Morris agreed. "In terms of using TMB in the clinic, this is not the end but the beginning of the story," he said, adding that it is not ready to be used as a companion diagnostic.

"These findings allow us, as a field, to now move forward with important prospective studies to determine if TMB can help us guide therapy and improve outcomes across different cancer types," he explained.

The team is now working to develop a more sophisticated tool. "An ideal model that will predict whether or not a tumor will respond to ICI may ultimately include multiple genomic aspects such as TMB, HLA genotype, specific genetic mutations, copy number, and measures of the tumor immune microenvironment," he said.

"However, this study indicates that TMB is going to be directionally important (high vs low)," Morris said. He also pointed out that the TMB cutoff (high vs low) will be assay specific. "We now need to generate the data necessary to identify the optimal cutoffs for high TMB in different cancers, understanding that these numbers will differ from assay to assay," he said.

"For now, as we decide how to best report TMB to clinicians, it is clear that we should not solely be using TMB in decision making, but in some cases, it may be one of many factors we may consider, just like PD-L1 staining," he noted.

"Capturing as little as 3% of the coding exome by using targeted panels such as MSK-IMPACT appears to provide a sufficient estimation of total tumor mutational load conferring predictive value for patients in whom ICI treatment is being considered," Samstein and colleagues write in the article.

"However, we do not yet have clinical evidence from a prospective clinical trial that shows patients triaged based on TMB benefit from ICI therapy," Morris said.

Rimm agreed. "We need a prospective study or a multi-institutional retrospective analysis of patients who have been treated with ICI comparing them with those who have not using a standardized TMB test that gets the same results at each site," he said. "It will be important that the value of TMB is borne out using multiple kits and multiple panels. TMB, if it's scientifically sound, should be reproducible across all users, regardless of where it originated," he added.

Several study authors are inventors on a provisional patent application filed by Memorial Sloan Kettering (MSK) related to the use of TMB in cancer immunotherapy, and some are also inventors on a patent application filed by MSK relating to the use of TMB in lung cancer immunotherapy. MSK and the inventors are reported to likely receive a share of the commercialization revenue from licensing agreements related to these patents. All other conflict of interest information for each author is listed at the end of the published report. Morris has received consulting fees from Rakuten Aspyrian and speaker fees from Physician Educational Resources. Rimm has been as a consultant for Ultivue, Perkin Elmer, Cell Signaling Technology, Merck, AstraZeneca, Agilent, Bristol-Myers Squibb, Bethyl Laboratories, Biocept, Cepheid, Navigate/Novartis, Gilead Sciences, and OptraScan.

Nat Gen. Published online January 14, 2019. Abstract

For more from Medscape Oncology, follow us on Twitter: @MedscapeOnc


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.