Tumor mutational burden, or TMB, counts the number of DNA mutations inside a cancer’s genome and uses that tally to predict whether immunotherapy is likely to work. The core idea is straightforward: tumors carrying more mutations tend to produce more abnormal proteins on their surfaces, which makes them easier for the immune system to recognize and attack. In 2020, this reasoning led the FDA to approve the immunotherapy drug pembrolizumab for any solid tumor with a TMB of 10 or more mutations per megabase of DNA, regardless of where the cancer originated. But the story behind that approval, and the practical reality of using TMB in the clinic, is considerably more complicated than a single number might suggest.
What TMB Actually Measures
TMB is defined as the average number of somatic mutations per megabase in a tumor’s sequenced DNA.1PubMed. Tumour mutational burden: an overview for pathologists “Somatic” means these mutations were acquired during a person’s lifetime rather than inherited. They accumulate from DNA replication errors, exposure to carcinogens, or failures in the cell’s DNA repair machinery. The count varies enormously across cancer types: melanomas and lung cancers driven by ultraviolet light and tobacco smoke often carry hundreds of mutations per megabase, while many pediatric cancers and blood cancers carry fewer than five.2PubMed Central. Tumor Mutational Burden as a Predictive Biomarker in Solid Tumors
Not every mutation matters equally. Most somatic mutations are “passenger” mutations that do nothing useful or harmful. A smaller subset produces altered proteins that the immune system can potentially distinguish from normal tissue. These altered proteins, called neoantigens, get displayed on the surface of cancer cells. When immune cells encounter a neoantigen that looks foreign, they can mount an attack. The logic behind TMB is essentially a numbers game: the more mutations a tumor has, the more neoantigens it is likely to generate, and the better the odds that at least some of those neoantigens will provoke a strong immune response.3PubMed Central. The role of neoantigens and tumor mutational burden in cancer immunotherapy: advances, mechanisms, and perspectives
Why It Matters for Immunotherapy
Immune checkpoint inhibitors work by releasing the brakes the immune system places on itself. Normally, cancer cells exploit those brakes to hide from T cells. Drugs like pembrolizumab block a protein called PD-1, allowing T cells to recognize and kill tumor cells. The catch is that T cells still need something foreign-looking to latch onto. A tumor with very few neoantigens may simply not give the immune system enough targets, even after the brakes are released. That is why TMB has become a useful proxy: it estimates, indirectly, how visible a tumor might be to T cells.4Journal for ImmunoTherapy of Cancer. Tumor mutation burden for predicting immune checkpoint blockade response: the more, the better
The landmark approval came from KEYNOTE-158, a multicenter trial that enrolled patients across nine tumor types. Among patients whose tumors had a TMB of 10 or more mutations per megabase, the overall response rate to pembrolizumab was about 29%, and more than half of those responses lasted at least a year.5PubMed Central. FDA Approval Summary: Pembrolizumab for the Treatment of Tumor Mutational Burden-High Solid Tumors A pan-tumor retrospective analysis further supported the idea that a single TMB cutoff could work across cancer types, independent of PD-L1 expression or microsatellite instability status.6PubMed Central. Tumor mutational burden predicts the efficacy of pembrolizumab monotherapy: a pan-tumor retrospective analysis of participants with advanced solid tumors
How TMB Is Measured
The gold standard for counting mutations is whole-exome sequencing, which reads all the protein-coding regions of the genome. But whole-exome sequencing is expensive and slow, so in practice most hospitals use targeted gene panels that sequence a curated set of several hundred cancer-related genes and then extrapolate a TMB estimate from that smaller snapshot. The FDA-approved companion test for the pembrolizumab approval is FoundationOne CDx, a panel covering about 324 genes.
How well do these panels match up with whole-exome results? In lung cancer samples, one study found that roughly 92% of cases were classified the same way by both methods, though about 8% of tumors called “high” by whole-exome sequencing were missed by the targeted panel.7PubMed Central. Paired analysis of tumor mutation burden calculated by targeted deep sequencing panel and whole exome sequencing in non-small cell lung cancer A broader analysis comparing five large gene panels against whole-exome sequencing found consistently strong correlations across platforms.8Scientific Reports. Analysis of tumor mutational burden: correlation of five large gene panels with whole exome sequencing These correlations are reassuring in aggregate, but individual patients can still be misclassified, and smaller panels are less reliable than larger ones.
The Standardization Problem
One of the persistent headaches with TMB is that different testing platforms can return different numbers for the same tumor. A panel that covers more genes will capture more mutations, and different bioinformatics pipelines make different decisions about which mutations to count. The Friends of Cancer Research harmonization project found that panels need to cover at least 667 kilobases of genomic territory to reliably classify tumors as TMB-high or TMB-low across the range of cutoffs used in practice.9PubMed. Aligning tumor mutational burden (TMB) quantification across diagnostic platforms: phase II of the Friends of Cancer Research TMB Harmonization Project
Even the “10 mutations per megabase” cutoff is not as universal as it sounds. When researchers compared results from FoundationOne CDx, MSK-IMPACT (Memorial Sloan Kettering’s panel), and whole-exome sequencing using 10 as the universal cutoff, inter-platform agreement was good between panels but slightly lower between panels and whole-exome results. The reason is that gene panels are enriched for frequently mutated cancer genes, so their per-megabase estimates tend to run a bit higher than whole-exome averages. One analysis found that using a cutoff of 8 mutations per megabase for whole-exome sequencing produced better alignment with the panel-based cutoff of 10.10PubMed Central. Universal cutoff for tumor mutational burden in predicting the efficacy of anti-PD-(L)1 therapy for advanced cancers
Technical factors also matter. Tumor purity, meaning what fraction of the sequenced tissue is actually cancer versus normal cells, affects the count. If a biopsy contains too many normal cells, low-frequency mutations get drowned out. One study found that targeted panels were largely unaffected by tumor purity, while whole-exome estimates shifted significantly as purity dropped.11PubMed Central. Clinical advantage of targeted sequencing for unbiased tumor mutational burden estimation in samples with low tumor purity Sequencing depth, or how many times each DNA fragment is read, adds another variable. Below a certain depth, rare mutations are missed; above it, sequencing artifacts can inflate the count.12PubMed Central. Bias and inconsistency in the estimation of tumour mutation burden These are not just theoretical concerns. They mean that the same tumor biopsied at two different hospitals could receive two different TMB classifications depending on which platform is used, how deeply it is sequenced, and how much tumor tissue was in the sample.
TMB Does Not Predict Equally Across All Cancer Types
A large study of more than 8,000 patients across 24 cancer types found that higher TMB was associated with better survival on checkpoint inhibitor therapy in a stepwise fashion: compared to patients with fewer than 5 mutations per megabase, those with 10 to 20 had about a 21% lower risk of death, and those with 20 or more had roughly half the risk. This held across most tumor types tested, including lung cancer, melanoma, urothelial cancer, endometrial cancer, gastric cancer, and breast cancer.13PubMed Central. Tumor mutational burden and survival on immune checkpoint inhibition in >8000 patients across 24 cancer types
Small-cell lung cancer stood out as a notable exception, showing no clear survival benefit from high TMB despite being a heavily mutated tumor type. Colorectal cancer had its own wrinkle: the benefit held for microsatellite-stable tumors overall, but the effect was less consistent than in other cancers. These exceptions matter because they illustrate that mutation count alone does not capture everything about how a tumor interacts with the immune system. The tumor microenvironment, the specific genes that are mutated, and the patient’s own immune landscape all play roles that a simple mutation tally cannot fully account for.
Where TMB Meets Other Biomarkers
TMB is not the only biomarker used to select patients for immunotherapy. PD-L1 expression, a measure of how much of the immune-checkpoint protein a tumor displays on its surface, was approved as a companion biomarker before TMB was. Microsatellite instability (MSI), a sign of defective DNA mismatch repair, is another well-established marker. These three overlap in some patients but diverge in others, and the clinical question is whether combining them improves prediction.
TMB and PD-L1 do not correlate well with each other. In one analysis of non-small-cell lung cancer, there was no statistical relationship between a tumor’s TMB and its PD-L1 level. Yet a prediction model that incorporated both outperformed either biomarker alone.14ASCO Annual Meeting. Tumor mutational burden and PD-L1 expression as predictors of response to immunotherapy in NSCLC In resectable lung cancer, both higher PD-L1 and higher TMB were independently associated with achieving a complete pathological response after neoadjuvant immunotherapy, reinforcing that they capture different aspects of tumor biology.15PubMed Central. Development and validation of a PD-L1 and tumor mutational burden-based predictive model for pathological complete response after neoadjuvant immunotherapy in resectable lung cancer
The overlap between TMB and MSI is particularly informative in colorectal cancer. Nearly all microsatellite-unstable colorectal tumors also have high TMB, which makes sense because defective DNA repair leads to an accumulation of mutations. But about 3% of microsatellite-stable colorectal tumors also turn out to be TMB-high, and those patients showed responses to PD-1 inhibitors despite having stable microsatellites.16PubMed Central. Beyond microsatellite testing: assessment of tumor mutational burden identifies subsets of colorectal cancer who may respond to immune checkpoint inhibition Testing for TMB in addition to MSI expanded the pool of potentially eligible patients by more than 50% in that study. Separate work confirmed that the immune microenvironment of microsatellite-stable, TMB-high colorectal tumors looked more like that of microsatellite-unstable tumors than like typical microsatellite-stable tumors, suggesting a genuine biological basis for the overlap.17PubMed Central. A next-generation sequencing-based strategy combining microsatellite instability and tumor mutation burden for comprehensive molecular diagnosis of advanced colorectal cancer
Why High TMB Does Not Guarantee a Response
Even among patients whose tumors have very high TMB (20 or more mutations per megabase), only about 45% respond to checkpoint inhibitors.18Trends in Cancer. Tumor Mutational Burden: A Biomarker for Cancer Therapy Several layers of biology explain the gap. First, only a small fraction of mutations produce neoantigens that the immune system can actually recognize. Second, even when neoantigens exist, the tumor’s molecular context determines whether T cells can reach and kill the cancer cells. Tumors with active beta-catenin signaling or PTEN loss, for instance, tend to exclude T cells from the tumor interior. Others recruit immunosuppressive cells that dampen the T-cell response.19PubMed Central. Tumor Mutational Burden as a Predictor of Immunotherapy Response: Is More Always Better? And some tumors lose the ability to present antigens on their surface altogether, rendering their neoantigens invisible regardless of how many they carry.
There is also the question of which mutations matter. Not all mutations in a tumor are equal. Clonal mutations, present in every cancer cell, are more immunologically relevant than subclonal mutations, which exist only in a subset of cells. In advanced urothelial cancer, the average percentage of clonal mutations in immunotherapy responders was about twice that of non-responders. Clonal TMB separated responders from non-responders far better than total TMB did, while subclonal TMB showed no significant difference between the two groups.20PubMed Central. The impact of mutational clonality in predicting the response to immune checkpoint inhibitors in advanced urothelial cancer This suggests that the standard TMB measurement, which lumps all mutations together, misses an important qualitative dimension.
Measuring TMB From a Blood Draw
Traditional TMB measurement requires a tissue biopsy, which is invasive, sometimes risky, and occasionally not feasible depending on where the tumor sits. Blood-based TMB, which analyzes fragments of tumor DNA circulating in the bloodstream, is an appealing alternative. In patients with adequate circulating tumor DNA (at least 1% of the total DNA in plasma), blood-based TMB correlated well with tissue-based TMB and detected TMB-high tumors with over 80% sensitivity.21The Journal of Liquid Biopsy. Validity and utility of blood tumor mutational burden (bTMB) is dependent on circulating tumor DNA (ctDNA) shed: SCRUM-Japan MONSTAR-SCREEN A separate validation study found a similar correlation, though a handful of cases had dramatically higher blood-based TMB than tissue-based TMB, possibly reflecting tumor heterogeneity or DNA shed from metastatic sites not captured in the original biopsy.22Clinical Chemistry. Analytical and Clinical Validation of Cell-Free Circulating Tumor DNA Assay for the Estimation of Tumor Mutational Burden
The limitation is that blood-based TMB depends heavily on how much tumor DNA is actually in the bloodstream. Some cancers, particularly brain tumors and certain low-grade malignancies, shed very little DNA into circulation, which makes blood-based measurement unreliable for those patients. For cancers that do shed adequately, though, liquid biopsy offers a practical advantage: it can be repeated over time to track how TMB changes with treatment, without requiring repeated invasive procedures.
How Treatment Changes TMB Over Time
TMB is typically measured at diagnosis or before starting immunotherapy, but it is not a static number. A pan-cancer analysis found that anticancer treatments, particularly chemotherapy, tend to increase TMB. The effect was most pronounced in cancer types that start with relatively low mutation counts, such as breast, prostate, and pediatric cancers. Chemotherapy was correlated with increased TMB in most cancer types studied, and the fraction of tumors classified as TMB-high rose significantly after chemotherapy in breast, prostate, and bladder cancers as well as glioma.23PubMed Central. Pan-cancer analysis of heterogeneity of tumor mutational burden and genomic mutation under treatment pressure
This has practical implications. A tumor that tested TMB-low before chemotherapy might test TMB-high afterward, potentially opening the door to immunotherapy. Conversely, the new mutations accumulated under chemotherapy pressure may be subclonal and therefore less immunogenic than the mutations already present at diagnosis. Whether chemotherapy-induced TMB elevation translates into better immunotherapy outcomes remains an open and actively studied question.
Pediatric Cancers and Low-TMB Tumors
Most pediatric cancers have very low TMB compared to adult cancers, which has raised questions about whether TMB testing is even useful in children. One challenge is that at low mutation counts, the variability between testing platforms gets worse. A study of pediatric tumors found that while panel-based and whole-exome TMB correlated well overall, this was largely driven by a handful of hypermutated cases. When only tumors below 10 mutations per megabase were analyzed, the correlation dropped substantially, meaning that for the vast majority of pediatric tumors, current platforms struggle to reliably distinguish TMB-high from TMB-low.24PubMed Central. Assessment of Tumor Mutational Burden in Pediatric Tumors by Real-Life Whole-Exome Sequencing and In Silico Simulation of Targeted Gene Panels
That does not mean low-TMB tumors are invisible to the immune system. Even in pediatric medulloblastoma, which typically carries very few mutations, researchers have identified patient-specific neoantigens capable of triggering T-cell responses.25PubMed. Low mutational load in pediatric medulloblastoma still translates into neoantigens as targets for specific T-cell immunotherapy The mutations may be few, but if even one happens to produce a strongly immunogenic neoantigen, that can be enough. This points to a fundamental limitation of TMB as a biomarker: it measures quantity, not quality. A tumor with five mutations, one of which is highly visible to the immune system, could respond better to immunotherapy than a tumor with fifty mutations that are all immunologically bland.
Beyond Counting Mutations
Researchers are increasingly looking at refinements to the blunt count of mutations per megabase. One avenue focuses on the type of mutation. A pan-cancer analysis found that insertion and deletion mutations (indels) that cause frameshifts in the reading frame of a gene generate roughly three times more high-affinity neoantigens per mutation than simple single-letter changes in DNA. When restricted to neoantigens that are truly novel, meaning the normal version of the protein would not trigger an immune response, frameshift indels were nine times more enriched than point mutations.26The Lancet Oncology. Insertion-and-deletion-derived tumour-specific neoantigens and the immunogenic phenotype: a pan-cancer analysis A tumor with a moderate overall TMB but a high proportion of frameshift indels might be more immunogenic than its raw number suggests.
Gene expression signatures of the tumor microenvironment offer another complementary dimension. A T cell-inflamed gene expression profile has been shown to independently predict response to anti-PD-1 therapy when combined with TMB. Patients with both high TMB and a T cell-inflamed microenvironment had the best outcomes, and the two markers were not correlated with each other, meaning they captured genuinely different biology.27European Journal of Cancer. Tumour mutational burden as a biomarker for immunotherapy: Current data and emerging concepts The likely future of biomarker-guided immunotherapy involves composite scores that weigh TMB alongside mutation type, clonality, immune microenvironment signatures, and possibly blood-based monitoring, rather than relying on any single number.
Cost and Access Barriers
Comprehensive genomic profiling for TMB is not cheap. Whole-genome or whole-exome sequencing can run into thousands of dollars, and even panel-based tests carry significant costs that vary by country and healthcare system. A cost-effectiveness analysis found that while immunotherapy itself was cost-effective compared with chemotherapy across multiple testing scenarios, adding TMB testing to guide treatment selection did not clearly justify its expense when compared with no testing or PD-L1 testing alone. Among TMB testing methods, tissue-based panel testing was the most cost-effective approach.28Journal of Clinical Oncology. Cost-effectiveness of PD-L1 testing and tumor mutational burden testing of immune checkpoint inhibitors for non-small cell lung cancer A separate modeling study for advanced lung cancer concluded that whole-genome sequencing-based TMB testing was not currently cost-effective for treatment selection compared with PD-L1 alone, though it could become so if sequencing costs continue to fall.29PubMed Central. Exploring the Cost Effectiveness of a Whole-Genome Sequencing-Based Biomarker for Treatment Selection in Patients with Advanced Lung Cancer Ineligible for Targeted Therapy
For patients in resource-limited settings or those whose tumors cannot be biopsied easily, access to TMB testing is a real barrier. The development of validated blood-based TMB tests and continued drops in sequencing prices may eventually democratize access, but for now, the availability of TMB testing varies widely across institutions and countries. In practice, many oncologists still use PD-L1 as the primary gatekeeper for immunotherapy and reserve TMB testing for cases where PD-L1 results are equivocal or negative but clinical suspicion of immunotherapy benefit remains high.