Somatic Variant Analysis
Tumor Profiling and Precision Oncology

Understanding somatic analysis begins with understanding what makes somatic variants different from germline variants, biologically, analytically, and clinically.

The biology

Germline variants are present in every cell of the body from conception. They are inherited or arise de novo during early embryonic development, present at 50% in heterozygous carriers, and cause hereditary conditions including inherited cancer predisposition syndromes.

Somatic variants arise in individual cells after birth, through replication errors, mutagen exposure, or failed DNA repair. They are present only in cells descended from the original mutated cell. The clinical consequence: somatic variants are not at 50% VAF. They appear at the VAF of the tumor clone, which may be 80% in a pure tumor, 20% with normal contamination, or 0.1% in liquid biopsy where only a fraction of cfDNA is tumor-derived. Detecting low-VAF somatic variants with clinical confidence is the central technical challenge of somatic NGS.

Comprehensive tumor profiling must detect every variant class with therapeutic or prognostic relevance. Each type requires a specific detection approach.

Sample quality and type profoundly affect what somatic analysis can detect and how reliably it can detect it.

One of the most consequential analytical decisions in somatic NGS is whether to sequence matched normal tissue alongside the tumor. The choice shapes specificity, germline inference, and what kinds of findings the report can and cannot make.

Somatic calling differs from germline calling in sensitivity requirements, artifact management, and the databases used for filtering. Standard germline callers are tuned for 50% VAF heterozygous calls. Somatic callers must reliably operate well below that.

VAF sensitivity thresholds

• Solid tumor, FFPE, high purity: typically 5–20% minimum VAF
• Solid tumor, low purity, necrotic, or decellularized: down to 2–5% VAF
• Liquid biopsy: 0.1% VAF or lower for MRD and resistance monitoring

Sensitivity thresholds should be established during analytical validation for each specific assay. Not adopted from tool defaults.

Somatic-specific callers

• Mutect2 (GATK): the most widely used paired caller, designed for matched germline filtering.
• Strelka2: high-performance somatic SNV and indel caller tuned for clinical accuracy.
• VarScan2: supports paired and unpaired calling, common for tumor-only applications.
• DRAGEN Somatic: Illumina's hardware-accelerated pipeline, widely used at clinical scale.

No single caller handles every variant type. Gene fusion detection needs dedicated tools (STAR-Fusion, Arriba, FusionCatcher for RNA-based; Delly, Manta for DNA). CNV detection needs separate algorithms. A comprehensive pipeline integrates several specialized callers and reconciles their output.

Artifact filtering

At low VAF, artifact rates can rival true signal. Essential steps include strand-bias filtering (true variants appear on both strands; FFPE C→T artifacts are often strand-specific), population database filtering (gnomAD >0.1% is almost always germline), lab-specific blacklists of recurrent artifact positions, and minimum-depth thresholds appropriate for the assay.

The 2017 AMP/ASCO/CAP guidelines (Li et al., Journal of Molecular Diagnostics) define the standard framework for interpreting and reporting somatic variants. Unlike ACMG germline classification, which focuses on pathogenicity, AMP focuses on actionability: the degree to which a variant influences treatment decisions.

Evidence levels feed tier assignment

• Level A: FDA-approved or standard-of-care guidelines for this tumor type. Examples: BRAF V600E in melanoma, EGFR L858R in NSCLC, HER2 amplification in breast cancer.
• Level B: Strong evidence or expert consensus from large clinical studies, but FDA approval in this tumor type may be absent.
• Level C: Clinical significance in a different tumor type. Relevant for off-label and basket trial eligibility.
• Level D: Preclinical data or limited clinical observations. Biological plausibility, not yet clinical actionability.

The most transformative development in precision oncology over the past decade has been immune checkpoint inhibitor therapy. Three genomic biomarkers now play a central role in identifying which patients are most likely to respond.

As tumor-based somatic NGS expands, a growing issue has emerged: a meaningful fraction of variants identified during tumor profiling are actually germline in origin, and some are clinically significant cancer predisposition variants.

Studies estimate that 4 to 12% of patients undergoing tumor-based molecular profiling carry pathogenic or likely pathogenic germline variants relevant to their cancer, in genes including BRCA1, BRCA2, MLH1, MSH2, TP53, and CDH1. These are often detectable in tumor data because the germline variant is present in all cells (including tumor cells) at ~50% VAF.

When a likely germline pathogenic variant is identified in tumor profiling:

• Inform the ordering oncologist with a clear explanation of the somatic vs germline distinction.
• Offer genetic counseling referral: the patient may have hereditary predisposition with implications for surveillance, risk-reducing interventions, and family member testing.
• Recognize treatment implications: germline BRCA1/2 mutations predict PARP inhibitor sensitivity and may qualify the patient for trials designed for hereditary cancer predisposition.
• Perform a separate germline confirmation test on normal tissue before clinical action is taken on a variant identified in tumor tissue.

A CAP-compliant somatic NGS report must communicate complex molecular findings clearly enough for an oncologist to act on them, accurately enough to support therapeutic decisions, and transparently enough to disclose analytical limitations.

Required content

• Specimen information: tumor type, biopsy site, sample date, tumor purity, DNA yield, sequencing quality
• Methodology: panel name and version, genes covered, sequencing platform, pipeline version, minimum coverage thresholds, reference build, variant callers, tumor-normal or tumor-only
• Tier I and II variants with HGVS, VAF, depth, tier classification, evidence level, matched therapies or trials
• Biomarkers: TMB (mut/Mb), MSI status (stable, indeterminate, high), HRD score where applicable
• Tier III variants reported with uncertainty disclosure
• Negative findings: which genes were analyzed and found negative, coverage gaps or low-depth regions
• Limitations: FFPE artifacts, tumor-only constraints, variant types not detected by this assay
• Therapeutic matching: FDA-approved therapies for Tier I, relevant clinical trials and off-label options for Tier II
• Laboratory director signature and contact

Clinical trial matching

A report identifying actionable variants should actively facilitate trial access. Include relevant ClinicalTrials.gov entries matching the patient's tumor type and identified alterations, so the oncologist can discuss eligibility at the same appointment where molecular results are reviewed.