3.4 ACMG/AMP Variant Classification
Key Takeaways
- ACMG/AMP uses a five-tier system—Pathogenic (P), Likely Pathogenic (LP), Variant of Uncertain Significance (VUS), Likely Benign (LB), and Benign (B)—based on combined evidence criteria, not a single test result
- PVS1 (null variant in a gene where loss of function is a known mechanism), PS1–PS4 (strong), PM2 (rarity in population databases), and PP3 (computational/predictive support) are high-yield evidence families counselors must recognize on reports
- Benign evidence (BS/BP codes) can outweigh weak pathogenic signals; classification is a balance of conflicting criteria, not a checklist that always upgrades to Pathogenic
- VUS counseling emphasizes that uncertain significance is not a diagnosis, medical management should follow phenotype and established guidelines rather than the VUS alone, and reclassification over time is expected
- Genetic counselors translate lab classifications into clinical action: cascade testing for P/LP, phenotype-driven care and periodic review for VUS, and reassurance with residual-risk framing for LB/B in the right context
3.4 ACMG/AMP Variant Classification
Quick Answer: Sequence variants are classified in a five-tier ACMG/AMP system—Pathogenic (P), Likely Pathogenic (LP), VUS, Likely Benign (LB), Benign (B)—by combining weighted evidence codes (e.g., PVS1, PS1–4, PM2, PP3, and BS/BP families). Counselors act on P/LP for diagnosis and cascade testing, treat VUS as uncertain (not a diagnosis), and avoid medicalizing LB/B findings.
Domain 3A items test whether you can read a molecular report, explain what the lab’s category means for the patient, and choose the next counseling or testing step. You do not need to recite every evidence code; you need application: which codes move a call toward pathogenic or benign, when a VUS should not drive surgery or prenatal decisions alone, and how to talk about uncertainty without false certainty.
The Five-Tier System
Clinical laboratories interpreting Mendelian sequence variants (germline) typically follow the 2015 ACMG/AMP framework (with gene- and disease-specific refinements from ClinGen and specialty societies). The five categories are designed for clinical reporting, not research curiosity labels.
| Tier | Abbreviation | Clinical meaning for counseling |
|---|---|---|
| Pathogenic | P | Sufficient evidence the variant causes disease in the stated context; supports diagnosis when phenotype fits |
| Likely Pathogenic | LP | High likelihood of pathogenicity (commonly framed as ≥90% certainty in the original framework); often managed like P for cascade testing when phenotype aligns |
| Variant of Uncertain Significance | VUS | Evidence insufficient or conflicting; not a positive diagnostic result |
| Likely Benign | LB | High likelihood the variant is not disease-causing in this context |
| Benign | B | Sufficient evidence the variant is not pathogenic |
Exam trap: Equating “rare” with “pathogenic.” Rarity (often captured conceptually under population-data criteria such as PM2) is only one piece of evidence. Common polymorphisms in gnomAD/population databases frequently support benign classification when allele frequency is too high for the disorder’s prevalence and penetrance model.
Another trap: Treating LP as “maybe” in the same way as VUS. LP is intentionally closer to P for clinical action in many settings (including family variant testing), whereas VUS is explicitly not used alone to establish a molecular diagnosis or to offer relatives predictive testing as if the variant were known disease-causing.
Evidence Codes You Must Recognize Conceptually
Codes are lettered by strength and direction: P = pathogenic-leaning, B = benign-leaning; VS/S/M/P ≈ very strong / strong / moderate / supporting. Specialty rules (e.g., for BRCA1/2, hearing loss, RASopathies) can modify how codes are applied—know that gene-specific guidance can override generic application.
| Code family (examples) | Core idea | Counselor translation |
|---|---|---|
| PVS1 | Predicted null variant (nonsense, frameshift, canonical ±1/2 splice, initiation codon, single- or multi-exon deletion) in a gene where loss of function is a known disease mechanism | Strong push toward pathogenicity—but only if LoF is established for that gene/disease; last-exon or NMD-escape variants may weaken PVS1 |
| PS1 | Same amino acid change as a previously established pathogenic variant (different nucleotide possible) | “Known bad missense at this residue” type support |
| PS2 / PM6 | De novo (maternity/paternity confirmed vs assumed) in a patient with the phenotype and no family history | Powerful in severe pediatric phenotypes; confirm parental relationships when the report claims confirmed de novo |
| PS3 | Well-established functional studies show a damaging effect | Lab or literature assays—not a patient’s clinical severity alone |
| PS4 | Prevalence in affected individuals statistically increased vs controls | More common in case–control / cohort evidence than in a single-family session |
| PM2 | Absent or extremely rare in population databases | Supports pathogenicity only as moderate (or adjusted) evidence—never sufficient alone |
| PP3 | Multiple lines of computational evidence support a deleterious effect (in silico) | Supporting at most in standard use; do not oversell “the computer says pathogenic” |
| BS / BP families | Strong or supporting benign evidence (e.g., high allele frequency, functional studies showing no effect, co-occurrence with a known pathogenic variant in trans for recessive disease, in silico benign predictions) | Can neutralize or outweigh weak pathogenic codes → LB/B or keep as VUS if conflict remains |
How counselors use this on a report: You rarely recalculate the lab’s points from scratch on the CGC exam. You interpret the narrative: “Classified LP based on a predicted null variant in a LoF gene (PVS1-type evidence) plus absence from population databases and de novo occurrence.” Your job is to connect that logic to who should be offered testing next and what not to do with a VUS.
Worked Scenario: Null Variant vs Missense VUS
Scenario A — LP/P leaning null. A child with a classic haploinsufficiency syndrome has a heterozygous frameshift in a gene where LoF is an established mechanism. Parental studies show the variant is de novo (relationships confirmed). Population databases show absence. Expect a P or LP call driven by PVS1-type plus de novo and rarity evidence. Counseling: the molecular result supports the clinical diagnosis; offer cascade testing only if relevant (often limited if de novo and parents negative in blood—still mention germline mosaicism residual risk as in Mendelian teaching).
Scenario B — Missense VUS. An adult with an ambiguous connective-tissue phenotype has a novel heterozygous missense in a large panel gene. Frequency is ultra-rare (PM2-leaning), in silico tools are mixed-to-damaging (PP3-leaning), no functional data, and the variant is inherited from an unaffected parent. Conflicting or insufficient evidence → VUS. Counseling: the finding does not confirm a hereditary syndrome; do not recommend prophylactic surgery or predictive testing of relatives solely because of this VUS; manage based on clinical diagnostic criteria and family history; discuss that labs reclassify over time.
Scenario C — Benign frequency reality check. A “candidate” variant present at 2% allele frequency in a large outbred population database is almost never the cause of a rare highly penetrant Mendelian disease. Expect LB/B via frequency-based benign criteria regardless of a scary-sounding gene name on a large panel.
VUS Counseling Essentials
VUS sessions are high-stakes communication items on the CGC exam.
- Define uncertainty plainly: “The lab does not have enough evidence to say whether this spelling change causes disease.”
- Separate result from care plan: Screening, surgery, and prenatal decisions should follow phenotype, established diagnostic criteria, and guidelines—not the VUS alone.
- Family testing limits: Do not offer relatives “predictive” testing for a VUS as if a negative result clears risk or a positive result confirms diagnosis. Segregation studies may be research-adjacent or lab-directed in limited circumstances; frame carefully.
- Set expectations for change: Variants are reclassified; offer a plan for how the patient will learn of updates (portal, recontact policy, periodic reanalysis—see 3.5).
- Psychosocial honesty: Patients often hear “variant” as “positive.” Check understanding and correct the diagnostic implication.
From Classification to Clinical Action
| Report call | Typical counselor actions |
|---|---|
| P / LP (phenotype fits) | Use as molecular confirmation or strong support; discuss inheritance, cascade testing of relatives for the known familial variant, reproductive options, and management per gene/condition guidelines |
| P / LP (phenotype poor fit) | Pause—consider alternate diagnoses, secondary findings policy, or lab error/misattribution; do not force a mismatched syndrome onto the patient |
| VUS | Phenotype-driven care; document uncertainty; plan for reclassification/reanalysis; avoid cascade “predictive” framing |
| LB / B | Generally not used to explain the patient’s disease; continue diagnostic workup if phenotype remains unexplained |
Secondary findings (e.g., ACMG SF gene list on exome/genome) follow consent and lab policy—classification tiers still apply, but disclosure and actionability add a counseling layer separate from the primary indication.
Master the tiers and the behavioral difference between LP and VUS; that distinction drives most board vignettes more than memorizing the full code catalog.
A laboratory report classifies a heterozygous frameshift in a gene where loss of function is an established disease mechanism as Likely Pathogenic, citing predicted null effect and absence from population databases. The child’s phenotype matches the gene’s disorder. What is the most appropriate counseling emphasis?
Which statement best reflects correct VUS counseling after a multigene panel?
A novel missense variant is ultra-rare in gnomAD (PM2-type population evidence) and has mixed in silico predictions (PP3-type supporting evidence), but it is inherited from an unaffected parent and lacks functional or segregation data. Which classification outcome is most consistent with ACMG/AMP application teaching?
On a report, which evidence concept is most specifically described as a predicted null variant in a gene where loss of function is a known mechanism of disease?