ACMG Variant Interpretation

The ACMG/AMP 2015 guidelines (Richards et al.) are the international standard for germline variant interpretation. Updated by ClinGen with refinements for specific gene/disease contexts.

• Pathogenic (P) - ≥2 very strong, or 1 very strong + ≥1 strong evidence
• Likely Pathogenic (LP) - 1 very strong + moderate; or 1 strong + 2–3 moderate; or 2 moderate + 2 supporting
• Variant of Uncertain Significance (VUS) - conflicting or insufficient evidence
• Likely Benign (LB) - 1 strong benign + 1 supporting; or 2 supporting benign
• Benign (B) - ≥2 strong benign; or 1 BA1 standalone

• PVS1 (Very Strong) - null variant (nonsense, frameshift, splice ±1/2, initiation codon) in gene where LOF is disease mechanism
• PS1 (Strong) - same amino acid change as established pathogenic variant
• PS2 (Strong) - de novo (confirmed parentage) in affected child with no family history
• PS3 (Strong) - functional studies showing damaging effect (in vitro, in vivo)
• PS4 (Strong) - prevalence significantly increased in cases vs controls (OR>5)
• PM1 (Moderate) - in mutational hotspot or well-established functional domain
• PM2 (Moderate) - absent from controls in gnomAD (or extremely low frequency)
• PM3 (Moderate) - in trans with pathogenic variant for recessive disease
• PM4 (Moderate) - in-frame deletion/insertion in non-repeat region
• PM5 (Moderate) - novel missense at codon where different missense is pathogenic
• PM6 (Moderate) - assumed de novo (parentage not confirmed)
• PP2 (Supporting) - missense in gene with low benign missense rate
• PP3 (Supporting) - in silico tools support damaging effect (SIFT, PolyPhen, CADD)
• PP4 (Supporting) - phenotype highly specific for gene (e.g., FBN1 for Marfan)
• PP5 (Supporting) - reputable source reports as pathogenic (ClinVar P)

• BA1 (Standalone Benign) - allele frequency >5% in gnomAD (any ancestry) - alone = Benign
• BS1 (Strong Benign) - allele frequency greater than expected for disorder (e.g., >1%)
• BS2 (Strong Benign) - observed in healthy adult with complete penetrance disease
• BS3 (Strong Benign) - functional studies show no damaging effect
• BS4 (Strong Benign) - lack of segregation in affected family members
• BP1 (Supporting Benign) - missense in gene where only LOF variants cause disease
• BP3 (Supporting Benign) - in-frame deletion in repeat region
• BP4 (Supporting Benign) - multiple in silico tools suggest benign
• BP7 (Supporting Benign) - synonymous variant with no predicted splice effect

Step-by-step for a novel missense variant in BRCA2 found in a 40-year-old woman with triple-negative breast cancer:

• 1. gnomAD check: NM_000059.4(BRCA2):c.7480G>A p.(Val2494Met) - gnomAD v4.1 NFE frequency 0.00003 (3 heterozygotes in 130,000 alleles) → PM2 applied (Moderate)
• 2. ClinVar: 2 submissions - both VUS; no star rating conflict; previous assessment inconclusive → PP5 NOT applied (not a reputable classified source)
• 3. Variant type: missense → PVS1 does NOT apply; check if codon previously reported pathogenic
• 4. HGMD: Val2494Ile at same codon reported as Likely Pathogenic by one clinical lab → PM5 applied (novel missense at codon where different change is pathogenic)
• 5. In silico tools: SIFT 0.02 (deleterious), PolyPhen-2 0.98 (probably damaging), CADD 28 (top 0.2%), REVEL 0.78 (>0.75 threshold) → PP3 applied
• 6. Protein domain: Val2494 is within the OB2 fold of the BRCA2 DNA-binding domain - PM1 applied (functional hotspot)
• 7. Family segregation: proband's affected mother tested positive for same variant - 1 affected family member adds PS4 at supporting level (PP1)
• 8. Score tally: PM2 + PM5 + PM1 + PP3 + PP1 = 3 Moderate + 2 Supporting → Likely Pathogenic per Richards 2015 criteria
• 9. Report: "Likely Pathogenic - BRCA2 c.7480G>A p.(Val2494Met). Recommend clinical genetics review and family testing."

• Mistake 1: Applying PVS1 to a missense variant - PVS1 is ONLY for loss-of-function variants (nonsense, frameshift, canonical splice ±1/2, initiation codon deletion)
• Mistake 2: Applying BA1 (>5% gnomAD) to a known disease variant - rs334 (HBB p.Glu7Val) causes sickle cell despite 4% frequency in African populations; BA1 has exceptions for founder mutations
• Mistake 3: Counting PP3 AND PM5 for the same in silico evidence - these use different evidence types; PP3 = computational predictions, PM5 = different pathogenic missense at same codon
• Mistake 4: Claiming PS2 (de novo) without confirming parentage - use PM6 (assumed de novo) unless both parents were tested and confirmed
• Mistake 5: Applying PP5 to a ClinVar submission with 0 stars (conflicting) - PP5 requires reputable source with evidence; single lab submissions without evidence are insufficient
• Mistake 6: Forgetting ClinGen gene-specific guidelines - BRCA1/2, TP53, RASopathy genes all have specific refinements that override the generic ACMG 2015 criteria

A 12-year-old boy with intellectual disability and autism has WES revealing NM_004992.3(MECP2):c.916-2A>G in the canonical splice acceptor of exon 4. This is an X-linked gene (Rett syndrome when pathogenic).

• PVS1: canonical splice-site variant (±1/2 position from exon); MECP2 LOF is the mechanism for Rett syndrome → PVS1 applied (Very Strong)
• PS2: de novo confirmed - father and mother both tested negative → PS2 applied (Strong)
• PM2: absent from gnomAD (0 alleles) → PM2 applied (Moderate)
• PP4: phenotype (intellectual disability + autism in male) is consistent with MECP2-related disorder in males → PP4 applied (Supporting)
• Final classification: PVS1 + PS2 = Pathogenic (per Richards 2015 rules: 1 Very Strong + 1 Strong)
• Clinical impact: diagnosis of MECP2-related severe neonatal encephalopathy; mother tested as carrier (XMeCP2/X); 50% recurrence risk; early intervention services initiated