Skip to contents

Variant classification

Source: vignettes/variant_classification.Rmd
variant_classification.Rmd

Oncogenicity

Somatic aberrations (SNV/InDels) are evaluated for oncogenicity through an implementation of standard operating procedures proposed by ClinGen/CGC/VICC (Horak et al. 2022). Here, various properties of the variants and genes affected are assigned specific scores according to several criteria, both negative and positive, pending on whether the properties support an oncogenic or benign variant type. These scores are in turn aggregated towards an overall oncogenicity score.

Note that all properties/criteria provided in the SOP’s are not readily implemented in PCGR, specifically the ones requiring manual curation or expert review (i.e. experimental oncogenic variant evidence, requiring support from in vitro or in vivo functional studies (criteria OS2)). Also, the current source for existing oncogenicity classifications (ClinVar), which is needed for implementation of criteria OS1, OM3, is also considerably limited. Considering the current limitations, some oncogenic variants are likely to be missed and classified with uncertain significance (VUS) by PCGR. We highlight conventional ClinVar classifications (i.e. with respect to pathogenicity) alongside the current oncogenicity classifications, this may add value to the interpretation for uncertain cases. Furthermore, taking into the account the nature of the current implementation, we have adopted slightly different score thresholds for variant classifications to those proposed originally by (Horak et al. 2022). We are working to further improve the oncogenicity classification in PCGR, and welcome feedback on this matter.

Note also that for somatic copy number aberrations, we showcase potential oncogenic events as proto-oncogenes subject to amplifications (where level of amplification is configurable by the user), as well as tumor suppressor genes subject to homozygous deletions.

The following criteria/codes are currently used for variant oncogenicity classification in PCGR (key resources/tools used for implementation indicated in parentheses):

  • ONCG_OVS1 - Null variant - predicted as LoF - in bona fide tumor suppressor gene (VEP;CGC;CancerMine)
  • ONCG_OVS1_A - Null variant - annotated by OncoKB as Loss-of-function - in bona fide tumor suppressor gene (OncoKB)
  • ONCG_OVS1_B - Null variant - annotated by OncoKB as Likely Loss-of-function - in bona fide tumor suppressor gene (OncoKB)
  • ONCG_OS1 - Same amino acid change as previously established oncogenic variant - regardless of nucleotide change (ClinVar)
  • ONCG_OS2_A - Well established in vitro/in vivo functional studies (OncoKB-curated) show oncogenic effect of variant (OncoKB)
  • ONCG_OS2_B - Well established in vitro/in vivo functional studies (OncoKB-curated) show likely oncogenic effect of variant (OncoKB)
  • ONCG_OS3 - Located in a mutation hotspot with >= 50 samples with variant at AA position, >= 10 samples with same AA change (cancerhotspots.org)
  • ONCG_OM1 - Presumably critical site of functional domain (CIViC)
  • ONCG_OM2 - Protein length changes from in-frame dels/ins in known oncogene/tumor suppressor genes or stop-loss variants in a tumor suppressor gene (VEP;CGC;CancerMine)
  • ONCG_OM3 - Missense variant at an amino acid residue where a different missense variant determined to be oncogenic (using this standard) has been documented (ClinVar)
  • ONCG_OM4 - Located in a mutation hotspot with < 50 samples with variant at AA position, >= 10 samples with same AA change (cancerhotspots.org)
  • ONCG_OP1 - Multiple lines of computational evidence support of a damaging variant effect on the gene or gene product (dbNSFP)
  • ONCG_OP3 - Located in a mutation hotspot with < 10 samples with the same amino acid change (cancerhotspots.org)
  • ONCG_OP4 - Absent from controls (gnomAD) / very low MAF (any five major gnomAD subpopulations) (gnomAD)
  • ONCG_SBVS1 - Very high MAF (any five major gnomAD subpopulations) (gnomAD)
  • ONCG_SBS1 - High MAF (any five major gnomAD subpopulations) (gnomAD)
  • ONCG_SBS2_A - Well established in vitro/in vivo functional studies (OncoKB-curated) show neutral effect of variant (OncoKB)
  • ONCG_SBS2_B - Well established in vitro/in vivo functional studies (OncoKB-curated) show likely neutral effect of variant (OncoKB)
  • ONCG_SBP1 - Multiple lines of computational evidence support a benign variant effect on the gene or gene product (dbNSFP)
  • ONCG_SBP2 - Silent and intronic changes outside of the consensus splice site (VEP)

Actionability

Clinical actionability assessment of somatic SNVs/InDels, gene copy number aberrations, and RNA fusions found in the tumor sample implements recommendation guidelines by AMP/ASCO/CAP (Li et al. 2017). Specifically, different levels of actionability are implemented in the following manner:

  • Tier I: Variants of strong clinical significance — aberrations linked to predictive, prognostic, or diagnostic biomarkers in the CIViC database and the Cancer Biomarkers Database (and optionally OncoKB) that show
  • Tier II: Variants of potential clinical significance — aberrations linked to predictive, prognostic, or diagnostic biomarkers in the CIViC database and the Cancer Biomarkers Database (and optionally OncoKB) that show either
  • Tier III: Variants of unknown/uncertain clinical significance — Tier I and II assignment is alteration-type agnostic (driven solely by biomarker evidence strength and tumor-type matching). Tier III, however, is assigned through alteration-type specific routes:
    1. Biomarker-linked, weak evidence only (all alteration types) — variants linked to known biomarkers but only through weak clinical evidence (CIViC C/D/E; OncoKB therapeutic level 4) in a non-matching tumor type or pan-cancer (Any) context
    2. No biomarker link — alteration-type specific criteria:
      • SNVs/InDels: coding variants in oncogenes or tumor suppressor genes with a population allele frequency below 0.001 (gnomAD), not linked to any known biomarker in underlying databases
      • Copy number aberrations: amplified oncogenes, or deleted (homozygous, hemizygous, or heterozygous) tumor suppressor genes, yet not linked to any known biomarker in underlying databases
      • RNA fusions: fusions where at least one partner gene (5’ or 3’) is an oncogene and where there is a record in the Mitelman database, yet not linked to any known biomarker in underlying databases

In PCGR, we skip the classification of variants into the AMP/ASCO/CAP-specified Tier IV (benign/likely benign variants), but rather take a more cautious approach. Specifically, for SNVs/indels that do not fall into tier I, II, or III, we classify them into Tier V: Other coding variants, which includes protein-coding variants in non-cancer related genes, as well as Tier VI: Other non-coding variants, which includes synonymous variants, intronic variants, and other variants in non-coding regions.

Tier I/II — alteration-type agnostic biomarker matching

Tier I and II assignment applies identically to all somatic alteration types (SNVs/InDels, copy number aberrations, and RNA fusions). Assignment depends on whether the user has specified a primary tumor site (tumor-type specific mode) or the generic Any site (tumor-type agnostic mode):

Tumor-type specific (a primary tumor site is specified, e.g. Lung, Breast):

Biomarker site Evidence strength Tier
Matching site or pan-cancer (Any) Strong (CIViC A/B; OncoKB therapeutic 1/2/3A/3B) I
Non-matching tumor-specific site Strong (CIViC A/B; OncoKB therapeutic 1/2/3A/3B) II
Matching site Weak (CIViC C/D/E; OncoKB therapeutic 4) II
Non-matching tumor-specific site or pan-cancer (Any) Weak (CIViC C/D/E; OncoKB therapeutic 4) III

Tumor-type agnostic (primary site set to Any):

Biomarker site Evidence strength Tier
Pan-cancer (Any) Strong (CIViC A/B; OncoKB therapeutic 1/2/3A/3B) I
Non-pan-cancer (any tumor-specific site) Strong (CIViC A/B; OncoKB therapeutic 1/2/3A/3B) II
Pan-cancer (Any) or any tumor-specific site Weak (CIViC C/D/E; OncoKB therapeutic 4) III

Pan-cancer strong evidence is treated as equivalent to a matching-site strong item and always promotes a variant to Tier I, regardless of query mode. Pan-cancer weak evidence contributes to Tier III in both modes.

Tier III — alteration-type specific routes

Unlike Tier I/II, Tier III assignment is alteration-type specific. Two distinct routes lead to Tier III and it is important to distinguish between them:

Route 1 — Biomarker-linked, weak evidence only (all alteration types)

Variants that are linked to known biomarkers but only through weak clinical evidence (CIViC C/D/E; OncoKB therapeutic level 4) in non-matching tumor types or pan-cancer contexts. These variants have some clinical signal but insufficient evidence to reach Tier II. This route applies equally to SNVs/InDels, copy number aberrations, and RNA fusions.

Route 2 — No biomarker link (alteration-type specific criteria)

Variants not linked to any known biomarker in the underlying databases, but flagged as potentially significant based on gene/alteration properties:

Alteration type Tier III criterion
SNVs/InDels Coding variant in an oncogene or tumor suppressor gene and population allele frequency < 0.001 (gnomAD)
Copy number aberrations Amplified oncogene, or deleted (homozygous, hemizygous, or heterozygous) tumor suppressor gene
RNA fusions At least one fusion partner gene (5’ or 3’) is an oncogene and at least one record for the fusion event in the Mitelman database

Evidence items: tier-defining vs. additional

Each biomarker evidence item associated with a variant is classified as either tier-defining or additional support:

  • Tier-defining — the item directly justifies the variant’s assigned tier (e.g. a matching-site strong item on a Tier I variant, or a weak item on a Tier III variant)
  • Additional — the item provides supporting context but did not drive the tier assignment (e.g. a non-matching-site strong item on a Tier I variant, or pan-cancer weak evidence on a Tier I or II variant)

References

Horak, Peter, Malachi Griffith, Arpad M Danos, et al. 2022. “Standards for the Classification of Pathogenicity of Somatic Variants in Cancer (Oncogenicity): Joint Recommendations of Clinical Genome Resource (ClinGen), Cancer Genomics Consortium (CGC), and Variant Interpretation for Cancer Consortium (VICC).” Genet. Med., January. https://dx.doi.org/10.1016/j.gim.2022.01.001.
Li, Marilyn M, Michael Datto, Eric J Duncavage, et al. 2017. “Standards and Guidelines for the Interpretation and Reporting of Sequence Variants in Cancer: A Joint Consensus Recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists.” J. Mol. Diagn. 19 (1): 4–23. https://doi.org/10.1016/j.jmoldx.2016.10.002.