What Tumor Mutational Burden Does and Does Not Tell You

Tumor mutational burden, usually shortened to TMB, counts the somatic mutations in a tumor and expresses them as mutations per megabase of sequenced DNA. It is a rough proxy for how many abnormal proteins, called neoantigens, a tumor might display for the immune system to recognize. As a biomarker for immune checkpoint immunotherapy, TMB works best as a continuous, probabilistic signal: higher generally means better odds of response, lower means worse odds, with wide overlap in between. What it does not give you is a clean yes-or-no verdict, and the single 10 mutations per megabase line drawn through it means different things in different tumors and on different assays.

What TMB is actually measuring

Checkpoint inhibitors like anti-PD-1 and anti-PD-L1 antibodies release a brake on T cells so they can attack cancer. For that attack to happen, T cells first have to see something foreign. Mutations in a tumor's DNA can produce altered proteins that look foreign, and a tumor carrying more mutations is, on average, more likely to present neoantigens the immune system can target. TMB is a genomic estimate of that potential. It is counted from next-generation sequencing, either across the whole exome or, far more commonly in the clinic, from a targeted gene panel that samples a portion of the genome and extrapolates.

The key word is potential. TMB counts mutations; it does not confirm that any of them generated a real neoantigen, that the neoantigen was presented, or that a T cell ever recognized it. It is an upstream surrogate several steps removed from the biology that actually determines whether immunotherapy works.

Where the 10 mutations per megabase threshold came from

In June 2020, the FDA granted pembrolizumab accelerated, tumor-agnostic approval for previously treated adults with unresectable or metastatic TMB-high solid tumors, defined as 10 or more mutations per megabase on a specific companion diagnostic. As the FDA approval summary describes, this was the first cancer approval based on TMB and among the first based on a biomarker rather than the tumor's site of origin.

The evidence came from a subset of the KEYNOTE-158 trial. Among patients whose tumors met the TMB-high cutoff, the objective response rate was about 29 percent, compared with roughly 6 percent in those below it, and a majority of responses lasted a year or longer. That is a meaningful separation, and it justified an option for patients who had run out of standard choices. It also helps to be precise about what that number is: a response rate in a pre-treated, biomarker-selected population, not a survival guarantee and not a promise that crossing the line changes an individual outcome.

Why one threshold does not travel cleanly across tumor types

The 10 mutations per megabase cutoff was a pragmatic choice, not a biologically derived constant. It was not statistically inferred from efficacy data, and the trial that supported it did not demonstrate a clear overall survival benefit tied to that exact line. Two limitations follow directly.

First, response varied widely by tumor type even among TMB-high patients in KEYNOTE-158, from strong responses in some cohorts to single-digit rates in others. The same TMB number did not carry the same meaning everywhere.

Second, and more fundamental, TMB and immune recognition are not linked in every cancer. A 2021 analysis by McGrail and colleagues in Annals of Oncology found that high TMB predicted checkpoint response in cancers such as melanoma, lung, and bladder, where mutation load correlated with CD8 T-cell infiltration. In cancers such as glioma, breast, and prostate, that correlation broke down, and high TMB did not predict benefit. The mechanism explains the discordance: TMB only helps when more mutations actually translate into more immune infiltration. In tumors where that chain is broken, a high count is a number without a consequence.

Why the same tumor can score differently on different assays

Even setting biology aside, the measurement itself is not fixed. TMB from a targeted panel is an estimate of what whole-exome sequencing would find, and that estimate shifts with panel size, which regions are covered, and how the bioinformatics pipeline counts variants, including whether it filters out germline and known cancer-driver mutations. A tumor scored on one platform may land above the cutoff and below it on another.

The Friends of Cancer Research TMB Harmonization Project studied this directly. Across many laboratories using different panels, in-silico and empirical work showed that panel design and analysis pipelines were real drivers of variation in the reported number. The project developed a calibration approach, aligning each panel to a common reference derived from The Cancer Genome Atlas, which reduced the spread of panel values around the whole-exome benchmark for the large majority of samples, along with a public tool to improve reproducibility. That effort matters precisely because, without alignment, a fixed threshold like 10 mutations per megabase does not correspond to the same underlying biology from one assay to the next.

How to read a TMB result

A high TMB result is best understood as one input that raises the estimated probability of benefit, strongest in tumor types where mutation load tracks with immune infiltration, and it should be interpreted alongside the assay used, the tumor type, and other markers such as microsatellite instability and PD-L1. A low result lowers the estimated odds but does not rule benefit out, and a value sitting near the cutoff should be read as genuinely uncertain rather than decisively positive or negative. TMB narrows the question; it does not close it.

This article is educational and not medical advice. Decisions about immunotherapy and biomarker testing belong in a conversation with the treating oncology team, who can weigh the specific tumor type, assay, and clinical situation.