Somatic Tumor Variants (MSI, MRD)

Tumor genomic testing addresses questions distinct from germline interpretation: what drives this cancer, what therapy will work, is the patient in remission, and is there an underlying hereditary syndrome? Somatic variant classification uses the AMP/ASCO/CAP 4-tier system (2017), separate from the ACMG/AMP germline framework covered in the ACMG Variant Classification leaf.

• Paired tumor/normal sequencing (gold standard): the matched normal (typically blood or buccal) defines the germline background, so variants present only in tumor are clearly somatic. Required for definitive separation when a hereditary cancer syndrome is on the differential.
• Tumor-only sequencing: the more common clinical workflow. Distinguishing germline from somatic relies on bioinformatic filters:
  • Population frequency filters: variants above a gnomAD threshold (often 0.1-1%) are flagged as likely germline.
  • ClinVar/HGMD annotation: known germline pathogenic variants are flagged for confirmation testing.
  • Variant allele frequency (VAF) patterns: heterozygous germline ~50%, homozygous germline ~100%, somatic typically variable and usually below 50% depending on tumor purity, ploidy, and clonality.
• Incidental germline findings: tumor-only assays may detect likely-pathogenic variants in cancer-predisposition genes (BRCA1/2, MLH1, TP53). Many laboratories require germline confirmation testing on a separate normal sample before reporting these as hereditary.

Phenotype of mismatch repair deficiency (dMMR), classically caused by biallelic loss of MLH1, MSH2, MSH6, or PMS2 (germline in Lynch syndrome; somatic via MLH1 promoter hypermethylation in sporadic colorectal and endometrial cancer).

• Bethesda 5-marker panel (PCR-based): BAT-25, BAT-26 (mononucleotide), D2S123, D5S346, D17S250 (dinucleotide). Compares tumor to matched normal.
  • MSI-H: 2 or more markers unstable (40% or more)
  • MSI-L: 1 marker unstable
  • MSS: no markers unstable
• NGS-based MSI scoring: tools such as MSIsensor, mSINGS, and MANTIS quantify instability across hundreds to thousands of microsatellite loci within an exome or panel sequencing run. Increasingly replaces the 5-marker panel.
• Immunohistochemistry (IHC) for MMR proteins is a complementary screen; absent staining of MLH1/PMS2 or MSH2/MSH6 indicates dMMR.
• Clinical use:
  • Lynch syndrome screening: universal tumor MSI/IHC testing on colorectal and endometrial cancers, with reflex MLH1 promoter methylation testing to distinguish sporadic from germline cases.
  • Immune checkpoint inhibitor eligibility: pembrolizumab has FDA tumor-agnostic approval for MSI-H or dMMR solid tumors (the first such approval, 2017).

Total somatic non-synonymous mutations per megabase of sequenced coding region. High TMB (commonly defined as 10 mut/Mb or more) predicts response to immune checkpoint inhibitors because heavily mutated tumors generate more neoantigens.

• Pembrolizumab has tumor-agnostic FDA approval for TMB-H (10 mut/Mb or more) solid tumors (2020).
• TMB calculation depends on panel size; small panels (less than 1 Mb) produce noisy estimates, so 1.5 Mb is a commonly cited minimum.
• Caveats: hypermutator phenotypes from POLE/POLD1 exonuclease-domain mutations and from MSI-H produce extremely high TMB but represent distinct biology.

Detection of residual circulating tumor DNA (ctDNA) after definitive therapy to predict relapse before clinical or imaging recurrence.

• Tumor-informed assays (e.g., Signatera): a custom panel is built from the patient's specific somatic variants identified at diagnosis, then deep sequencing of plasma cfDNA detects those variants at very low VAF. Sensitivity often reaches 0.01% VAF or below.
• Tumor-agnostic / fixed-panel assays (e.g., Guardant Reveal): use a predefined gene panel plus methylation signatures; no upfront tumor profiling required, but typically less sensitive than tumor-informed assays.
• Clinical use: post-resection surveillance in colorectal, breast, and bladder cancer; adjuvant therapy escalation for MRD-positive patients in clinical trials (e.g., DYNAMIC, COBRA).

Distinct from the 5-tier ACMG/AMP germline classification.

• Tier I - Strong clinical significance: variants with FDA-approved therapy in the patient's tumor type, or included in professional guidelines (NCCN). Example: BRAF V600E in melanoma.
• Tier II - Potential clinical significance: variants with FDA-approved therapy in a different tumor type, evidence from clinical trials, or strong preclinical data. Example: BRAF V600E in colorectal cancer (off-label therapy plus EGFR blockade).
• Tier III - Unknown clinical significance: variants not previously observed in cancer or in population databases at minor allele frequencies inconsistent with germline polymorphism; analogous to germline VUS.
• Tier IV - Benign or likely benign: common polymorphisms or variants without any cancer association. Generally not reported.