Tumor Mutational Burden (TMB)

TMB is calculated as the total number of non-synonymous somatic single nucleotide variants (SNVs) and small insertions/deletions (indels) per megabase of the sequenced coding genome.

• Whole Exome Sequencing (WES) is the gold standard, covering ~30-50 Mb of coding DNA
• Targeted gene panels (e.g., MSK-IMPACT, FoundationOne CDx) estimate TMB from 0.8-1.5 Mb and require rigorous calibration to WES
• Synonymous mutations, germline variants, and driver mutations are typically filtered out
• Allele frequency thresholds (often ≥5%) are applied to ensure clonal representation

Defining TMB-High (TMB-H) status requires a clinically validated cutoff, which varies by cancer type and assay platform. The pan-tumor approval of pembrolizumab uses ≥10 mutations per megabase as a tissue-agnostic threshold.

• Cutoffs are often derived from the top quintile or decile of the population distribution
• Cancer-specific thresholds may be more predictive; for example, TMB-H in colorectal cancer is often defined differently than in lung cancer
• Harmonization efforts like the Friends of Cancer Research TMB Harmonization Project aim to standardize cutoffs across assays
• Blood-based TMB (bTMB) from liquid biopsies requires distinct thresholds due to lower circulating tumor DNA shedding

The mechanistic rationale for TMB as an immunotherapy biomarker is that somatic mutations generate neoantigens—novel peptide sequences presented on the tumor cell surface by major histocompatibility complex (MHC) class I molecules.

• Higher TMB increases the probability of producing immunogenic neoantigens recognized by CD8+ T cells
• Not all mutations create neoantigens; only those altering amino acid sequences and binding MHC with sufficient affinity are relevant
• Clonal neoantigens (present in all tumor cells) are more immunogenic than subclonal ones
• Computational pipelines like NetMHCpan predict peptide-MHC binding affinity to estimate neoantigen burden from mutation calls

TMB distributions vary dramatically across cancer types, reflecting differences in mutagenic exposures and DNA repair deficiencies.

• Ultra-high TMB: Cancers with mismatch repair deficiency (dMMR) or UV-induced mutagenesis (melanoma) often exceed 50 mut/Mb
• Moderate TMB: Non-small cell lung cancer (NSCLC) in smokers averages 8-10 mut/Mb
• Low TMB: Pediatric cancers and liquid tumors (leukemias) often have <1 mut/Mb
• This variability means a single pan-cancer TMB-H cutoff may misclassify patients in low-mutation tumor types
• TMB percentile ranking within a cancer type is sometimes more informative than absolute value

TMB calculation is highly dependent on the bioinformatics pipeline, including alignment, variant calling, and filtering parameters.

• Variant allele frequency (VAF) thresholds must balance sensitivity for low-frequency subclonal mutations against false positives from sequencing artifacts
• Germline filtering requires either matched normal samples or population databases like gnomAD; tumor-only pipelines risk inflating TMB with rare germline variants
• Panel-to-exome calibration uses linear regression or conversion factors to harmonize TMB estimates
• FFPE-induced artifacts (cytosine deamination) must be computationally corrected using tools like GATK FilterMutectCalls or TNscope
• Reproducibility studies show inter-assay TMB correlation coefficients of 0.7-0.9, indicating significant but imperfect concordance

Despite FDA approvals, TMB has significant limitations that motivate ongoing research into composite biomarkers.

• TMB-H does not guarantee ICI response: Only 20-40% of TMB-H patients respond, indicating neoantigen quality matters more than quantity
• Intratumoral heterogeneity means a single biopsy may not capture the full mutational landscape; subclonal mutations may not generate uniformly presented neoantigens
• MHC class I loss or beta-2-microglobulin (B2M) mutations can abrogate neoantigen presentation regardless of TMB
• T-cell exhaustion and immunosuppressive tumor microenvironments can override high neoantigen load
• Emerging composite biomarkers integrate TMB with PD-L1 expression, TIL density, and immune gene expression signatures (e.g., T-effector signature) for improved prediction

A hypermutation phenotype caused by defective DNA mismatch repair (dMMR). MSI-High tumors accumulate frameshift mutations in repetitive DNA tracts, generating abundant neoantigens. While MSI-H and high TMB frequently overlap, they are mechanistically distinct biomarkers. MSI is a binary or categorical assay (PCR-based or NGS), whereas TMB is a continuous quantitative measure. Both independently predict immune checkpoint inhibitor response, and the FDA has approved pembrolizumab for MSI-H/dMMR solid tumors regardless of tissue origin.

The total number of novel peptide sequences presented by major histocompatibility complex (MHC) molecules on the tumor cell surface, derived from somatic mutations. High TMB is a surrogate for elevated neoantigen burden, but not all mutations generate immunogenic neoantigens. Computational pipelines predict MHC-I binding affinity using tools like NetMHCpan to identify clonal neoantigens most likely to trigger T-cell recognition. Neoantigen quality—including clonality and foreignness (similarity to self)—may be more predictive of immunotherapy response than raw TMB alone.

A functional defect in the cellular machinery that corrects base-base mismatches and insertion-deletion loops during DNA replication. dMMR results from germline mutations (Lynch syndrome) or somatic silencing (MLH1 promoter hypermethylation) of MLH1, MSH2, MSH6, or PMS2 genes. dMMR tumors accumulate orders of magnitude more mutations than MMR-proficient tumors, directly elevating TMB. Immunohistochemistry for MMR protein expression and MSI PCR testing are standard clinical assays, while TMB quantification via next-generation sequencing provides a genome-wide mutation count.

A protein-level biomarker measured by staining tumor and immune cells for Programmed Death-Ligand 1 expression. Unlike TMB, which is a genomic assay, PD-L1 IHC is assessed on tissue slides using various antibody clones (22C3, 28-8, SP142, SP263) with different scoring thresholds. PD-L1 and TMB are independent, non-redundant biomarkers—some patients with low PD-L1 but high TMB respond to checkpoint inhibitors, and vice versa. Combined assessment may improve patient stratification for anti-PD-1/PD-L1 therapy.

The gold-standard method for TMB quantification, sequencing approximately 1-2% of the genome covering all protein-coding exons (~30-50 megabases). WES-based TMB is calculated by counting non-synonymous somatic mutations per megabase after filtering out germline variants using matched normal DNA. The FoundationOne CDx and MSK-IMPACT targeted panels provide validated TMB estimates correlated with WES. Harmonization efforts like the Friends of Cancer Research TMB Harmonization Project aim to standardize TMB calculation across platforms and establish universal cutoffs.

The therapeutic context driving TMB's clinical relevance. ICIs are monoclonal antibodies blocking CTLA-4, PD-1, or PD-L1 to reactivate exhausted T-cells against tumors. High TMB predicts improved objective response rates and progression-free survival across multiple cancer types, most notably in non-small cell lung cancer (NSCLC), melanoma, and bladder cancer. The CheckMate-227 trial established TMB ≥10 mut/Mb as a predictive cutoff for first-line nivolumab plus ipilimumab in NSCLC, leading to FDA approval of TMB as a companion diagnostic.