Module 3: Pan-Cancer Biomarkers: TMB, MSI, VAF, CNV & Loss of Heterozygosity
1. Tumor Mutational Burden (TMB)
Tumor Mutational Burden (TMB) quantifies the total number of non-synonymous somatic mutations per megabase (Mb) of examined coding genome. Mechanistically, tumors harboring high mutational loads generate elevated frequencies of novel peptide sequences (neoantigens). These neoantigens are processed and presented on tumor cell surfaces via MHC-I complexes, provoking cytotoxic CD8+ T-cell antitumor immune responses.
$$\text{TMB (mut/Mb)} = \frac{\text{Count of Eligible Somatic Non-Synonymous Coding Variants}}{\text{Total Interrogated Coding Target Space (Mb)}}$$In June 2020, the FDA approved Pembrolizumab (anti-PD-1) for adult and pediatric patients with unresectable or metastatic tumor mutational burden-high ($\text{TMB} \ge 10\text{ mut/Mb}$) solid tumors that have progressed following prior treatment and have no satisfactory alternative options (based on the KEYNOTE-158 trial). Comprehensive Genomic Profiling panels (e.g. FoundationOne CDx, MSK-IMPACT) require at least $0.8 - 1.2\text{ Mb}$ of sequenced coding territory to provide reliable, low-variance TMB estimations.
2. Microsatellite Instability (MSI)
Microsatellites are short, repetitive nucleotide tandem repeat sequences (1–6 bp repeats, e.g. $(A)_n$ or $(CA)_n$) distributed throughout the genome. When the cellular DNA Mismatch Repair (MMR) machinery—encoded by MLH1, MSH2, MSH6, and PMS2—is inactivated through somatic mutation, promoter hypermethylation, or germline Lynch syndrome mutations, DNA polymerases slip across these repeat tracts during replication.
This failure to repair slippage creates length alterations across microsatellite loci, termed Microsatellite Instability-High (MSI-H). Computational algorithms (such as MSIsensor-pro and mSINGS) quantify the percentage of mutated microsatellite loci directly from targeted NGS reads:
$$\text{MSI Score (\%)} = \frac{\text{Unstable Microsatellite Sites}}{\text{Total Evaluated Microsatellite Sites}} \times 100\%$$Tumors with $\text{MSI Score} \ge 20\% - 30\%$ are classified as MSI-H and are strongly hypersensitive to immune checkpoint blockade regardless of tissue of origin.
3. Variant Allele Fraction (VAF) & Tumor Purity Deconvolution
The Variant Allele Fraction (VAF) is the proportion of sequenced sequencing reads at a given genomic locus that support the mutant allele relative to total depth:
$$\text{VAF} = \frac{\text{Alternative (Mutant) Read Depth (AD)}}{\text{Total Read Depth (DP)}}$$A variant's VAF is governed by three biological and technical parameters: Tumor Purity ($p$), Clonal Fraction ($c$), and Copy Number Status ($CN$). For a heterozygous somatic mutation in a diploid tumor with purity $p$:
$$\text{Expected VAF} = \frac{p \cdot c}{2(1 - p) + 2p} = \frac{p \cdot c}{2}$$For example, if a biopsy specimen contains 40% tumor cells ($p = 0.40$), a clonal heterozygous driver mutation will present at $\text{VAF} = 0.40 / 2 = 0.20$ ($20\%$). Variants with VAF significantly below expected clonal levels represent subclonal sub-populations indicative of intratumor heterogeneity.
4. Copy Number Alterations (CNVs) and Loss of Heterozygosity (LOH)
| Genomic Alteration | Biochemical & Cytogenetic Event | Therapeutic & Clinical Impact | Key Diagnostic Examples |
|---|---|---|---|
| Focal Gene Amplification | High-level copy-number gain ($\ge 6 - 8$ copies) restricted to a discrete chromosomal region. | Overexpression of oncogenic tyrosine kinases; predictive of targeted inhibitor efficacy. | ERBB2 (HER2) amplification in breast/gastric cancer (Trastuzumab sensitivity); MET amplification in NSCLC (TKI resistance). |
| Homozygous Deletion | Biallelic complete loss of both copies of a tumor suppressor gene (0 copies remaining). | Loss of cell-cycle checkpoint control or DNA repair competency. | CDKN2A/B homozygous deletion in glioblastoma and lung cancer; PTEN loss in prostate cancer. |
| Loss of Heterozygosity (LOH) | Loss of one parental allele across a chromosome arm or locus, frequently accompanied by duplication of the remaining mutant allele (Copy-Neutral LOH). | Satisfies Knudson's Two-Hit Hypothesis for tumor suppressor inactivation. | Somatic loss of wild-type TP53 or BRCA1 allele in patients harboring a single hit, rendering tumor cells completely defective in repair. |
Common Pitfalls & Troubleshooting Checklist
- Java Heap Exhaustion (GATK / Picard): If processes terminate with
java.lang.OutOfMemoryError, set your environment variable explicitly:export _JAVA_OPTIONS="-Xmx16g -Djava.io.tmpdir=/tmp". - Index Timestamp Mismatch: If tools complain that a BAM index is missing or out of date, regenerate with
samtools index -@ 4 sample.bam. - Conda Channel Conflicts: Avoid ABI incompatibilities between bioconda and conda-forge by setting strict priority:
conda config --set channel_priority strict. - Chromosome Naming Contamination: Always verify whether reference FASTA headers use
chr1or1; divergent naming causes silent alignment drops.