MSH3 c.1567G>A, p.(Glu523Lys)

Reasoning

BS1 (strong benign) is met. The applicable threshold is the generic SVI default of AF >= 0.01 (1%), whose per-gene derivation is attributed to Whiffin et al. 2017 (PMID:28518168); the supplied calibration explicitly states that per-gene derivation is preferred but that 0.01 is the default when none exists - and none exists for MSH3, since there is no ClinGen CSPEC/VCEP or local framework (generic_acmg_fallback) and therefore no MSH3-specific frequency threshold. The overall all-comers frequency is below 1% (gnomAD v2.1 0.00254891; v4.1 0.0013872), but BS1 is a frequency-versus-disease-expected criterion and must be evaluated on the maximum ancestry-specific frequency: the variant is present at 2.67064% (655/24,526 alleles) in African/African American in gnomAD v2.1 and 2.54492% (1,864/73,244) in African/African American in gnomAD v4.1, with grpmax FAF 0.025364 and 0.0244867 respectively - all approximately 2.5-fold above the 1% threshold. Applying BS1 only to the global average would wrongly treat a variant common in an ancestry that is under-represented in the pooled denominator as rare, which is precisely the situation the population-maximum (popmax/grpmax FAF) form of the Whiffin et al. 2017 derivation was designed to avoid. Concordant support: 7 homozygotes (all in African/African American) in gnomAD v2.1 and 24-25 homozygotes (24 in African/African American) in gnomAD v4.1, carried on both exome and genome platforms across two independent releases - a frequency/genotype burden that is incompatible with a rare, highly penetrant polyposis allele; MSH3-related adenomatous polyposis is rare and described only in a small number of biallelic families. The allele is not a mapping artifact (rs34058399; lcr = false, segdup = false). The variant is also present but unremarkable in gnomAD-Canada v1.0 (34/18,420, 0.185%, no homozygotes), which does not alter the ancestry-specific conclusion. BS1 strength remains strong as specified by ACMG/AMP; this variant is additionally reported as Likely benign/Benign in ClinVar, but that was not used as an input to this frequency determination.

Rule

BS1: strong benign when AF >= 0.01 (generic SVI default; per-gene derivation preferred, Whiffin et al. 2017, PMID:28518168), applied to the highest ancestry-specific frequency.

Evidence considered

• gnomAD v2.1: total AF 0.00254891 (716/280,904 alleles, 7 homozygotes); African/African American AF 0.0267064 = 2.67064% (655/24,526 alleles, 7 homozygotes); grpmax FAF 0.025364.
• gnomAD v4.1: total AF 0.0013872 (2,131/1,536,188 alleles, 25 homozygotes); African/African American AF 0.0254492 = 2.54492% (1,864/73,244 alleles, 24 homozygotes); grpmax FAF 0.0244867.
• BS1 threshold 0.01 (1%) taken verbatim from the supplied published calibration (generic SVI default; per-gene derivation preferred, credited to Whiffin et al. 2017, PMID:28518168). The variant exceeds it on the ancestry-maximum frequency in both gnomAD releases.
• Popmax/grpmax basis: the same allele is below 0.12% in every other ancestry (gnomAD v2.1 European non-Finnish 0.0000855, Admixed American 0.0011963), so the 2.5-2.7% African/African American frequency is a genuinely ancestry-enriched signal rather than a uniformly common allele.
• Homozygote burden: 7 homozygotes (gnomAD v2.1, all African/African American) and 24-25 homozygotes (gnomAD v4.1, 24 African/African American) - relevant because MSH3-related disease is autosomal recessive (biallelic loss of function, attenuated adenomatous polyposis; gene context OMIM 600887), so homozygotes in a population database are directly informative about tolerance.
• ClinGen SVI framework notes from the consulted methodology literature: Sherloc (PMID:28492532) makes population-frequency cutoffs explicit and inheritance-dependent (separate dominant and recessive thresholds) and treats high-frequency evidence as strong benign support; the ACMG/AMP 2015 statement (PMID:25741868) notes that gene- and disease-specific groups should refine criterion weighting, and a CLIA-lab analysis of 11,894 variants (PMID:26467025) reports population-frequency data as the most common driver of reclassification toward benign. None of these sources provides an MSH3-specific threshold, so the generic 1% default was used.
• Excluded explanations: dbSNP rs34058399; region_flag lcr = false and segdup = false; detected in both exome and genome data of both releases, so the frequency is not a mapping/paralog artifact or a single-platform artifact.
• gnomAD-Canada v1.0: 34/18,420 alleles (AF 0.00185, 0 homozygotes) - below the 1% threshold; noted but not decisive.
• MSH3 disease context used to judge whether 2.7% is 'greater than expected for the disorder': biallelic MSH3 variants have been reported in only a small number of families with attenuated adenomatous polyposis (gene-level context in this case), so a 2.5-2.7% carrier allele frequency with multiple population homozygotes is far higher than expected for the disorder.

Still missing

• No MSH3-specific BS1 threshold derived from disease prevalence/penetrance exists (no CSPEC/VCEP); the generic 1% default was used, which the variant exceeds by ~2.5-fold under popmax, so the missing per-gene derivation is unlikely to change the verdict.

Sources

Reasoning

SCOPE: BP1 asks whether this missense variant (p.Glu523Lys) is in a gene for which primarily truncating variants are known to cause disease; it is a gene-level mechanism criterion and was evaluated on gene-level evidence, not on this variant's own classification. FRAMEWORK: no MSH3 ClinGen CSPEC/VCEP specification or local gene-specific framework exists (cspec found = false; gene_vcep_dir = null; framework_mode = generic_acmg), so generic ACMG/AMP BP1 (PMID:25741868) applies at supporting strength (BP1 is a supporting-level benign criterion; no mechanism exists here to upgrade it). EVIDENCE: MSH3-associated disease is caused by loss-of-function/truncating alleles. pvs1_gene_context.json records that a targeted germline literature review supports MSH3 loss of function as the germline disease mechanism (lof_mechanism_supported = true; pvs1_gene_gate = 'eligible'), and its supporting gene-level source set includes PMID:37402566 ('MSH3: a confirmed predisposing gene for adenomatous polyposis'), which confirms biallelic MSH3 pathogenic variants with the EMAST phenotype that hallmarks MSH3 deficiency. Gene-level literature (gene-level mechanism context only - these papers do not mention p.Glu523Lys, and PMID:25741868/26467025 contain no MSH3 data) describes the same picture: Adam et al. 2016 (PMID:27476653, not a source_registry key) identified compound-heterozygous loss-of-function MSH3 germline mutations (c.1148delA, c.2319-1G>A, c.2760delC, c.3001-2A>C) with complete loss of nuclear MSH3 protein and EMAST, defining a recessive adenomatous polyposis subtype. The ClinVar gene-level pathogenic spectrum agrees: among MSH3 records carrying a pathogenic classification there are 319 frameshift and 161 nonsense records versus a single missense record (c.3130G>A p.Gly1044Ser, VCV001476245, single submitter, condition 'not provided'), and ClinVar residue-523 records contain no pathogenic change at all. Gene-level tolerance metrics are consistent with a recessive LoF mechanism (gnomAD pLI approximately 0, oe_mis 1.083). Caveat driving the supporting-only strength: MSH3's disease-causing spectrum is predominantly - not exclusively - truncating, because one pathogenic missense record exists, so BP1 is applied at supporting strength only.

Rule

Generic ACMG/AMP BP1: missense variant in a gene for which primarily truncating variants are known to cause disease. Met at supporting strength - MSH3-associated disease is driven by biallelic loss-of-function/truncating alleles (ClinVar pathogenic MSH3 records: 319 frameshift + 161 nonsense vs 1 missense; MSH3 germline LoF mechanism documented in pvs1_gene_context). Supporting strength only, and no upgrade mechanism exists for BP1.

Evidence considered

• BP1 is a gene-level mechanism criterion: it asks whether truncating variants are the primary cause of disease in MSH3. It was not inferred from this variant's ClinVar classification; the ClinVar data used here are the gene-level spectrum of MSH3 variants classified pathogenic.
• pvs1_gene_context.json: germline_disease_context_found = true, lof_mechanism_supported = true, pvs1_gene_gate = 'eligible', mechanism_rationale stating that a targeted germline literature review identified disease-focused publications supporting MSH3 loss of function as a germline disease mechanism; supporting gene-level sources include PMID:37402566 ('MSH3: a confirmed predisposing gene for adenomatous polyposis', J Med Genet 2023), and its caution is that no official CSPEC/VCEP or local explicit PVS1 framework was available.
• Gene-level literature context (gene-level mechanism only; the variant p.Glu523Lys is not mentioned in these papers, so they are recorded as gene-level context rather than variant-level citations, and PMID:27476653 is not a source_registry key for this case): Adam et al. 2016, Am J Hum Genet (PMID:27476653) found two unrelated individuals with compound-heterozygous loss-of-function MSH3 germline mutations (c.1148delA, c.2319-1G>A, c.2760delC, c.3001-2A>C) with EMAST and complete loss of nuclear MSH3 protein, establishing a recessive subtype of colorectal adenomatous polyposis; Villy et al. 2023 (PMID:37402566) confirms biallelic MSH3 pathogenic variants with EMAST in MSH3-associated polyposis.
• ClinVar gene-level pathogenic spectrum (NCBI E-utilities, MSH3[gene] AND 'clinsig pathogenic'[filter] combined with molecular-consequence filters): 319 frameshift variant records and 161 nonsense records versus a single missense record, c.3130G>A p.Gly1044Ser (VCV001476245, single submitter, condition 'not provided'). Truncating change is therefore the established MSH3 disease mechanism; residue-523 records in ClinVar (p.Glu523Lys benign/likely benign; p.Glu523Val uncertain significance; p.Glu523= synonymous) contain no pathogenic variant at all.
• gnomAD MSH3 gene-constraint metrics (gnomAD API, retrieved for this adjudication): pLI = 5.91e-31 and oe_mis = 1.0827, mis_z = -1.2946 - the gene is not haploinsufficiency-constrained and tolerates missense change, consistent with a recessive loss-of-function mechanism in which heterozygous missense variants are not expected to be pathogenic.
• Framework: no MSH3 CSPEC/VCEP specification or local gene-specific framework was available (cspec found = false, gene_vcep_dir = null, framework_mode = generic_acmg), so generic ACMG/AMP BP1 per Richards et al. 2015 (PMID:25741868) was applied; BP1 is a supporting-strength benign criterion, which is the strength assigned here.

Still missing

• BP1 is applied at supporting strength because the MSH3 ClinVar pathogenic spectrum contains one missense record (c.3130G>A p.Gly1044Ser, single submitter, condition 'not provided'), so the mechanism is predominantly rather than exclusively truncating.
• No MSH3 ClinGen CSPEC/VCEP specification exists to formally codify a truncating-only disease mechanism, and the gene-level LoF literature (e.g. PMID:27476653, PMID:37402566) is gene-level context that does not mention p.Glu523Lys.
• BP1 is a gene-level mechanism criterion applied to a variant whose own evidence for benignity (population frequency, ClinVar consensus) is assessed by the population and other groups; this group's BP1 verdict does not rely on that variant-level evidence.

Sources

Reasoning

SCOPE: NM_002439.5:c.1567G>A is a missense variant (NP_002430.3:p.(Glu523Lys); VEP most_severe_consequence = missense_variant), so BP4 is evaluated from the REVEL missense protein-impact path only; SpliceAI delta scores are not BP4 evidence for a missense variant and are not used. FRAMEWORK: there is no MSH3 CSPEC/VCEP specification and no gene-specific PP3/BP4 lookup for this case (framework_mode = generic_acmg), so the generic ClinGen SVI/Pejaver REVEL calibration applies. EVIDENCE: the local REVEL v1.3 lookup returned 0.228. BP4 requires REVEL <=0.29 (supporting), <=0.183 (moderate) or <=0.016 (strong) (Pejaver et al. 2022, Am J Hum Genet, PMID:36413997). 0.228 is at or below the 0.29 supporting cutoff but above the 0.183 moderate cutoff, so BP4 is met at supporting strength only. The result is not in the indeterminate gray zone (0.29 < score < 0.644), so a benign computational verdict is legitimately reached on the applicable path; conversely the same value sits far below the 0.644 PP3 cutoff, so no opposite-direction evidence is created. The BayesDel score (-0.40543) was not consulted for BP4 because no generic ACMG/AMP calibration for BayesDel exists and no MSH3 VCEP cutoff is available.

Rule

Generic ACMG/AMP BP4 via the REVEL missense path: REVEL <=0.29 (supporting), <=0.183 (moderate), <=0.016 (strong) per ClinGen SVI REVEL calibration (Pejaver et al. 2022, PMID:36413997). No gene-specific MSH3 VCEP/BP4 override available.

Evidence considered

• Variant scope: NM_002439.5:c.1567G>A is a missense substitution (NP_002430.3:p.(Glu523Lys), VEP most_severe_consequence = missense_variant), so BP4 is taken from the REVEL missense path; SpliceAI is not used as BP4 support for a missense variant.
• No MSH3 CSPEC/VCEP specification or gene-specific PP3/BP4 lookup spreadsheet exists for this case (cspec found = false, gene_vcep_dir = null, framework_mode = generic_acmg), so generic ClinGen SVI REVEL calibration applies.
• Local REVEL v1.3 lookup returned 0.228 for this variant (chr5:80728964 G>A, GRCh38).
• BP4 REVEL thresholds applied verbatim from the published ClinGen SVI calibration: <=0.29 supporting, <=0.183 moderate, <=0.016 strong (Pejaver et al. 2022, Am J Hum Genet, PMID:36413997). 0.228 <= 0.29 and > 0.183, so BP4 is met at supporting strength.
• 0.228 lies outside the indeterminate gray zone between the benign and pathogenic REVEL cutoffs (0.29 < score < 0.644), so the benign verdict rests on the applicable path itself rather than on a cross-path substitution.
• SpliceAI (max delta 0.001) was excluded from BP4: a clean splice prediction is not calibrated evidence that a missense substitution is tolerated.
• BayesDel = -0.40543 was not used: no generic ACMG/AMP strength calibration is established for BayesDel and no MSH3 VCEP cutoff is available.
• PMID:25741868 and PMID:26467025 were reviewed for the generic BP4/computational-evidence framework; PMID:33007869 reports this exact variant in an inflammatory breast cancer cohort but provides no computational or functional prediction data for it, and PMID:28492532 (Sherloc) is a generic classification refinement with no MSH3-specific content.

Sources

Reasoning

SCOPE: NM_002439.5:c.1567G>A (NC_000005.9:g.80024783G>A; GRCh38 chr5:80728964 G>A) is a missense substitution, NP_002430.3:p.(Glu523Lys), so PS1 is evaluable: it asks whether the identical amino-acid change (Glu523Lys, by any nucleotide change) has previously been established as pathogenic. FRAMEWORK: no MSH3 ClinGen CSPEC/VCEP and no local gene-specific framework exists for this case (cspec found = false; gene_vcep_dir = null; framework_mode = generic_acmg; vcep_materials.json domain_tables is empty; no user VCEP files), so the generic ACMG/AMP PS1 definition (Richards et al. 2015, PMID:25741868) governs. EVIDENCE: the only variant carrying the amino-acid change p.Glu523Lys is ClinVar variation ID 780543 / VCV000780543, which is NM_002439.5(MSH3):c.1567G>A (p.Glu523Lys) itself, and it is classified Benign/Likely benign, not pathogenic: 15 submissions (8 laboratories likely benign, 4 benign, 1 benign, 1 uncertain significance) with review status 'criteria provided, multiple submitters, no conflicts'. A codon-level ClinVar search performed for this adjudication (MSH3[gene] AND Glu523[all], NCBI E-utilities) returned only three MSH3 residue-523 records - VCV000780543 (c.1567G>A, p.Glu523Lys, this variant), VCV004829418 (c.1568A>T, p.Glu523Val, uncertain significance, single submitter) and VCV001775448 (c.1569G>A, p.Glu523=, synonymous) - none of which is established pathogenic. No established pathogenic variant encoding p.Glu523Lys therefore exists, and the amino-acid change that PS1 would require has in fact been classified benign/likely benign. PMID:33007869, the only retrieved paper that reports this exact variant, lists it only as a potentially germline finding in a cancer cohort table with no pathogenic assertion and does not establish any alternative nucleotide change yielding p.Glu523Lys.

Rule

Generic ACMG/AMP PS1: same amino-acid change as a previously established pathogenic variant, irrespective of nucleotide change. Not met because no established pathogenic variant produces p.Glu523Lys (the only variant with this amino-acid change is classified Benign/Likely benign). No MSH3 CSPEC/VCEP specification exists to modify this criterion.

Evidence considered

• PS1 requires the same amino-acid change (p.Glu523Lys), regardless of nucleotide change, to have been previously established as pathogenic; the criterion is therefore a question about residue-level precedent, not about this variant's own classification.
• ClinVar VCV000780543 (variation ID 780543) is NM_002439.5(MSH3):c.1567G>A (p.Glu523Lys) and is classified Benign/Likely benign, with 15 submissions (8 clinical laboratories likely benign, 4 benign, 1 benign, 1 uncertain significance) and review status 'criteria provided, multiple submitters, no conflicts' (ClinVar API, last updated 2026-06-27).
• A codon-level ClinVar search run for this adjudication (MSH3[gene] AND Glu523[all], NCBI E-utilities, returning variation IDs 4829418, 1775448 and 780543) showed that the MSH3 residue-523 records are p.Glu523Lys (this variant, Benign/Likely benign), p.Glu523Val (c.1568A>T, uncertain significance, single submitter, evaluated 2026-03-04) and p.Glu523= (c.1569G>A, synonymous); none is an established pathogenic variant producing p.Glu523Lys.
• No retrieved paper establishes p.Glu523Lys as pathogenic: PMID:33007869 (the only source that reports this exact variant) lists it as 'MSH3 (c.1567G > A; p.Glu523Lys) PG' in a targeted-NGS cancer cohort table without a pathogenicity assertion, and the same sample's second MSH3 change, c.1571A>C (p.Asn524Thr), is a different codon and is likewise unclassified.
• Framework note: no MSH3 CSPEC or VCEP/gene-specific specification was retrievable (cspec found = false, gene_vcep_dir = null, framework_mode = generic_acmg), so the generic ACMG/AMP PS1 wording from Richards et al. 2015 (PMID:25741868) was applied rather than any gene-specific modification.

Sources

Reasoning

PS3 requires a well-established in vitro or in vivo functional assay demonstrating a damaging effect on the gene or its protein product (Richards et al. 2015, PMID:25741868). NM_002439.5:c.1567G>A in MSH3 is a missense substitution encoding NP_002430.3:p.(Glu523Lys). No governing MSH3 gene-specific specification exists (no ClinGen CSPEC retrieved; framework_mode = generic_acmg, final_classification_framework found=false), so the generic ACMG/AMP 2015 definition of PS3 applies without any gene-specific VCEP assay/strength calibration. Every source assembled for this case that could carry functional evidence was reviewed. Literature: three of the four fully-reviewed papers are framework or methods documents containing no MSH3 data and no variant data (PMID:25741868, ACMG/AMP standards; PMID:28492532, Sherloc criteria refinement; PMID:26467025, CLIA-laboratory weighted scoring system that explicitly assigns in silico predictions only minor weight). The fourth, PMID:33007869, is the only source in this case that reports the exact variant: it lists MSH3 (c.1567G>A; p.Glu523Lys) in Table 2 for inflammatory breast cancer case IBC15 as a potentially germline (PG, VAF<0.5) finding from a targeted 105-gene panel, alongside a second MSH3 variant in the same sample. That paper provides no functional characterisation of this allele - no MMR-activity, MutS-beta heterodimerisation, expression, stability, or complementation assay, and no functional claim attributed to this variant; its immunohistochemistry was for ER/PR and MDM4/C-MYC, and MSH3 appears otherwise only in cohort-level variant-frequency comparisons. Variant-specific databases: OncoKB reports an unknown oncogenic effect with no variant-specific reviewed functional evidence, and no curated functional annotation for this residue. No somatic hotspot or COSMIC functional annotation exists for this position. ClinVar submissions for this variant were audited and yield no usable criterion-level or expert-panel evidence (0 usable submissions, 0 criterion leads, no expert-panel submission). No functional assay data meeting PS3's requirement - an established assay demonstrating a damaging effect - was identified for this variant, so PS3 is not met.

Rule

generic_acmg_2015:PS3_requires_established_functional_assay_showing_damaging_effect

Evidence considered

• Framework check: no ClinGen CSPEC / VCEP specification is available for MSH3 (cspec found=false; vcep_materials: no cspec url, no gene VCEP files, final_classification_framework found=false), so PS3 is assessed under the generic ACMG/AMP 2015 criterion definition with no gene-specific functional-assay calibration.
• Variant identity and consequence: NM_002439.5:c.1567G>A -> NP_002430.3:p.(Glu523Lys), a single-amino-acid missense substitution (consequence_class = missense; not a canonical splice-consensus, nonsense or frameshift allele), so PS3 would require a protein-level functional assay rather than a nonsense-mediated-decay/splice assay.
• Gene mechanism context: MSH3 acts as a DNA mismatch-repair tumour suppressor through the MutS-beta (MSH2-MSH3) heterodimer, with loss-of-function variants linked to autosomal recessive familial adenomatous polyposis and minimal evidence for Lynch syndrome (gene-level context in the case's MSH3 gene/PVS1 context file). This defines what a PS3-qualifying assay for this gene would need to measure, but is gene-level background only and is not variant-specific evidence.
• PMID:33007869 is the only reviewed source reporting this exact variant. Direct quote: "MSH3 (c.1567G > A; p.Glu523Lys) PG" (Table 2, case IBC15, MMR Genes column). Context: targeted next-generation sequencing panel of 105 cancer-related genes on 21 treatment-naive inflammatory breast cancer biopsies; 'PG' = potentially germline (VAF<0.5). No functional assay, protein-activity measurement, or variant-specific functional interpretation for this allele appears anywhere in the paper; I re-checked the raw full text for MSH3, functional, assay, mismatch repair, immunohistochemistry, MSI and microsatellite terms, and all MSH3 mentions are the Table 2 listings plus a cohort gene-frequency comparison with TCGA.
• PMID:25741868 (ACMG/AMP standards) full text reviewed: contains no MSH3 and no mention of c.1567G>A or p.(Glu523Lys); it defines PS3's evidence type and states that disease-specific groups should refine criterion applicability and weight, which is why the absence of an MSH3 specification leaves only the generic PS3 definition available here.
• PMID:28492532 (Sherloc, semiquantitative ACMG-AMP refinement) full text reviewed: methodology/framework paper with no MSH3 data and no mention of this variant; all worked examples concern other genes, so it contributes nothing variant-specific to PS3.
• PMID:26467025 (standardised clinical-laboratory scoring system) full text reviewed: contains no MSH3 or variant data; its only relevant content is that in silico predictions (SIFT/PolyPhen-2) are assigned minor weight - a point that does not substitute for the functional assay PS3 requires.
• OncoKB variant page (MSH3 E523K): classification 'Unknown Oncogenic Effect', is_variant_specific = false, not_reviewed = true, no citations - i.e. no variant-specific reviewed functional evidence, so no PS3 support from that source.
• ClinVar submissions for this variant audited for criterion-level leads: submissions_total = 0 usable, no expert-panel submission, criterion_leads = [] - no submitter cited functional evidence usable for PS3. Review status is single-submitter, so no expert-panel functional assertion is being inherited either.
• Position-level somatic annotation: no Cancer Hotspots row and no COSMIC entry for this variant, so there is no recurrent-somatic functional annotation at this residue to weigh toward PS3.

Still missing

• No variant-specific functional assay for MSH3 c.1567G>A (p.Glu523Lys) was identified in any assembled source, in either direction; PS3 cannot be satisfied without one.
• MSH3 disease mechanism is loss of function (MutS-beta mismatch-repair competence), so a qualifying PS3 assay for this gene would need to test MMR function or MutS-beta assembly/activity for this missense allele; no such assay exists for this variant.
• The case literature set is limited to four ClinVar-linked PMIDs (three framework/methods papers plus one case report); a targeted search of the MSH3 functional-assay literature for this specific allele was not part of the assembled evidence, so a variant-level functional study that exists outside this set would not have been surfaced.

Sources

Reasoning

FRAMEWORK: PS4 (Richards et al. 2015, generic ACMG/AMP 2015) requires that the prevalence of the variant in affected individuals be significantly increased compared with controls. No MSH3 ClinGen CSPEC/VCEP specification exists for this case (cspec found = false; gene_vcep_dir = null; framework_mode = generic_acmg; final_classification_framework source = generic_acmg_fallback), so the generic qualitative definition governs and no gene-specific PS4 rule is available to apply. NUMERIC THRESHOLD CHECK: no numeric PS4 threshold applies to this case. The governing generic definition is qualitative (a significant increase in affected individuals versus controls) rather than an explicit operator on a stated number, no MSH3 VCEP odds-ratio cutoff exists, and the published calibration set supplied for this case contains no PS4 value. Because no numeric threshold is in force, no >= or <= comparison is written here and this verdict does not rest on any threshold direction. EVIDENCE: no case-control study of this variant exists in any retrieved source. The only variant-level report, PMID:33007869, records c.1567G>A in a single inflammatory breast cancer case (IBC15) detected by targeted NGS of a tumour biopsy and annotated 'PG' (potentially germline, VAF < 0.5); that paper provides no control group, no allele counts in cases versus controls, no odds ratio and no statistical test for this variant, so no enrichment estimate can be derived. Its gene-level statement that MSH3 variants overall were more frequent in its IBC cohort than in TCGA non-IBC stages III-IV concerns MSH3 variants collectively and is not a variant-level affected-versus-control comparison. COUNTER-EVIDENCE: the allele is not rare in the general population - gnomAD v2.1 total AF 0.2549% (716/280,904 alleles, 7 homozygotes), popmax African/African American AF 2.6706%, grpmax FAF 2.5364%; gnomAD v4.1 total AF 0.1387% (2,131/1,536,188 alleles, 25 homozygotes), popmax African/African American AF 2.5449%, grpmax FAF 2.4487% - a profile consistent with a common, non-enriched allele rather than one significantly over-represented in affected individuals. With no case-control comparison available and a substantial population frequency, PS4 is not met at any strength.

Rule

Generic ACMG/AMP 2015 PS4 (strong; Richards et al. 2015, PMID:25741868): 'The prevalence of the variant in affected individuals is significantly increased compared with the prevalence in controls.' Applied qualitatively because no MSH3 CSPEC/VCEP specification and no numeric PS4 cutoff exist for this case.

Evidence considered

• Generic ACMG/AMP PS4 definition applied (no MSH3 VCEP or ClinGen CSPEC specification retrieved; framework_mode = generic_acmg; final_classification_framework source = generic_acmg_fallback): PS4 reflects a significantly increased prevalence of the variant in affected individuals compared with controls, and no MSH3-specific odds-ratio cutoff is available.
• No numeric PS4 threshold is in force for this case: the generic definition is qualitative and the published calibration set supplies no PS4 value, so the verdict is not based on any >= or <= comparison.
• PMID:33007869 (inflammatory breast cancer cohort, targeted NGS of 21 IBC biopsies): the exact variant MSH3 c.1567G>A (p.Glu523Lys) is listed in Table 2 for case IBC15 and annotated 'PG' (potentially germline, VAF < 0.5) alongside a second MSH3 change (c.1571A>C, p.Asn524Thr, also 'PG').
• PMID:33007869 contains no control arm, no case/control allele frequencies, no odds ratio and no p-value for c.1567G>A; a single tumour-panel observation is a case report, not case-control enrichment.
• PMID:33007869's enrichment-type statement is gene-level only (MSH3 variants were more frequent in the IBC cohort than in TCGA non-IBC stages III-IV), concerns MSH3 variants collectively rather than this variant, and supplies no comparison statistic usable for PS4.
• PMID:28492532 (Sherloc) describes PS4 only as 'higher prevalence in affected individuals versus controls' and provides no numeric odds-ratio cutoff applicable to MSH3, confirming that no quantified threshold is available from the retrieved framework literature.
• gnomAD v2.1: AF 0.2549% (716/280,904 alleles), 7 homozygotes, popmax African/African American AF 2.6706%, grpmax FAF 2.5364%.
• gnomAD v4.1: AF 0.1387% (2,131/1,536,188 alleles), 25 homozygotes, popmax African/African American AF 2.5449%, grpmax FAF 2.4487%.
• The high general-population frequency and the presence of multiple homozygotes in gnomAD argue against this allele being a rare pathogenic variant over-represented in affected individuals, so no enrichment signal in the pathogenic direction is present.
• gnomAD frequencies are used here only as PS4 counter-evidence regarding enrichment; their primary evaluation for population-frequency criteria belongs to the population evidence group and no frequency-based benign criterion is claimed in this group.

Still missing

• No case-control study, cohort versus control allele count, odds ratio or statistical test exists for MSH3 c.1567G>A; PS4 cannot be quantified without such data.
• No MSH3 ClinGen CSPEC/VCEP specification exists, so no gene-specific PS4 threshold or prevalence-weighting rule could be applied; the generic qualitative ACMG/AMP definition was used.
• The only individual reported to carry the variant (PMID:33007869, case IBC15) was analysed from a tumour biopsy with unconfirmed germline status and no phenotype-matched control group, so even a crude case:control comparison is impossible.

Sources

Reasoning

SCOPE: PM1 asks whether p.Glu523Lys (MSH3) lies in a mutational hotspot and/or a critical, well-established functional domain without benign variation. FRAMEWORK: there is no MSH3 ClinGen CSPEC/VCEP specification for this case (cspec found = false; gene_vcep_dir = null; framework_mode = generic_acmg), and vcep_materials.json carries an empty domain_tables object ({}), so there is no governing VCEP-approved critical-domain entry for MSH3 to check residue 523 against; the generic ACMG/AMP PM1 definition (PMID:25741868) therefore applies. EVIDENCE, hotspot prong: the CancerHotspots lookup for 'MSH3 E523' returned no result (found = false; residue_significant = false), i.e. residue 523 is not a statistically significant hotspot. EVIDENCE, domain prong: no critical functional domain at residue 523 is documented - the residue lies outside the annotated functional features of MSH3 (UniProt P20585, consulted via the UniProt REST API for this adjudication: annotated features are a disordered EXO1-interaction region 75-297 and a nucleotide-binding site at 896-903; all eight annotated natural variants are dbSNP polymorphisms), and the only pathogenic MSH3 missense in ClinVar (p.Gly1044Ser) sits in the C-terminal ATPase region, not near residue 523. EVIDENCE, 'without benign variation' prong: residue 523 is not depleted of tolerated variation - gnomAD v4.1 reports AF 0.1387% (2131/1536188 alleles, 25 homozygotes; popmax African/African American 2.5449%) and gnomAD v2.1 AF 0.2549% (716/280904 alleles, 7 homozygotes; popmax 2.6706%), and a different missense at the same codon (p.Glu523Val) is present in ClinVar as uncertain significance. Neither the hotspot prong nor the critical-domain-without-benign-variation prong is satisfied, so PM1 is not met at any strength.

Rule

Generic ACMG/AMP PM1: located in a mutational hotspot and/or a critical, well-established functional domain without benign variation. Not met - residue 523 is not a significant hotspot (CancerHotspots no result), is not within any annotated functional domain of MSH3 (UniProt P20585), and the residue tolerates population variation. No MSH3 CSPEC/VCEP domain table exists (vcep_materials.json domain_tables is empty).

Evidence considered

• PM1 requires a mutational hotspot and/or a critical, well-established functional domain without benign variation; both prongs were tested separately for residue 523.
• Framework: cspec found = false, gene_vcep_dir = null, framework_mode = generic_acmg, and vcep_materials.json contains 'domain_tables': {} - there is no VCEP-approved MSH3 domain table (and no local gene-specific framework) that could place residue 523 in an approved critical domain, so the generic ACMG/AMP PM1 definition from PMID:25741868 was applied.
• Hotspot prong: the CancerHotspots query 'MSH3 E523' returned no row (found = false, search_status = no_result, residue_significant = false), so residue 523 is not a statistically significant cancer hotspot. CancerHotspots has no source_registry key in this case and is recorded here as prose.
• Domain prong: UniProt P20585 (DNA mismatch repair protein Msh3, 1137 aa; accessed through the UniProt REST API for this adjudication, no source_registry key) annotates no functional domain, active site or critical residue at Glu523 - the annotated functional features are the EXO1-interaction region 75-297 and a binding site at 896-903, and its eight annotated natural variants (residues 62, 79, 709, 789, 949, 1045, 1054 and the 57-65 repeat) are all dbSNP polymorphisms rather than disease variants.
• Positional context for the domain prong: ClinVar's only pathogenic MSH3 missense record is c.3130G>A p.Gly1044Ser (VCV001476245, single submitter, condition 'not provided'), which lies in the C-terminal ATPase/MutS-V region far from residue 523; the pathogenic MSH3 spectrum is otherwise frameshift/nonsense/splice.
• Benign-variation prong: gnomAD v4.1 reports this residue-523 substitution at AF 0.0013872 (2131/1536188 alleles, 25 homozygotes), popmax African/African American AF 0.0254492 (1864/73244, 24 homozygotes), grpmax FAF 0.0244867; gnomAD v2.1 reports AF 0.00254891 (716/280904 alleles, 7 homozygotes), popmax AFR 0.0267064, grpmax FAF 0.025364. A well-populated, homozygote-bearing position with a second ClinVar missense at the same codon (p.Glu523Val, uncertain significance) is the opposite of the 'without benign variation' pattern PM1 requires.
• Conclusion: neither the hotspot prong nor the critical-domain-without-benign-variation prong is met, so PM1 is not met at any strength.

Still missing

• No MSH3 ClinGen CSPEC/VCEP specification exists, so no authoritative critical-domain or hotspot list is available for this gene; the domain_tables key in vcep_materials.json is present but empty, meaning residue 523 could not be checked against a governing VCEP domain table.
• Critical-domain assignment for MSH3 relied on UniProt P20585 feature annotation and CancerHotspots because no gene-specific specification defines MSH3 functional domains; a reviewer with an MSH3 domain-expert opinion could revisit the domain prong.

Sources

Reasoning

PM2 (rare/absent in population databases, at supporting strength per the supplied SVI-recommended threshold of AF <= 0.0001, PMID:25741868) is clearly not met. The variant is not rare by any measure. gnomAD v2.1 all-comers AF is 0.00254891 (716/280,904 alleles, 7 homozygotes) - about 25-fold above the 0.0001 threshold; gnomAD v4.1 all-comers AF is 0.0013872 (2,131/1,536,188 alleles, 25 homozygotes) - about 14-fold above it. On the highest ancestry-specific frequency the deviation is far larger: 2.67064% (655/24,526) in African/African American in v2.1 and 2.54492% (1,864/73,244) in v4.1, with grpmax FAF 0.025364 and 0.0244867. The exome-only subsets are also far above threshold (v2.1 exome 0.00195976; v4.1 exome 0.00072539), which matters because the non-cancer subsets that would have been preferred under an exome-calibrated VCEP could not be queried - GNOMAD_V2_1_NON_CANCER and GNOMAD_V3_1_NON_CANCER both failed and are treated as missing data, not absence. Because no MSH3 CSPEC/VCEP exists, the all-comers totals are the applicable default. Independent support for not-met: the variant is present in gnomAD-Canada v1.0 at 34/18,420 alleles (0.185%, 0 homozygotes) and has an established dbSNP identifier (rs34058399). PM2 is therefore not met at any strength, including supporting, and no rarity-based pathogenic support can be claimed for this variant.

Rule

PM2: supporting pathogenic when AF <= 0.0001 (SVI-recommended supporting-strength threshold, PMID:25741868); not met.

Evidence considered

• PM2 threshold taken verbatim from the supplied published calibration: supporting strength at AF <= 0.0001 (ClinGen SVI recommendation to downgrade PM2 to supporting, PMID:25741868).
• gnomAD v2.1: total AF 0.00254891 (716/280,904 alleles, 7 homozygotes), exome AF 0.00195976 (489/249,520), genome AF 0.00723298 (227/31,384) - approximately 20-70-fold above the 0.0001 threshold depending on subset.
• gnomAD v4.1: total AF 0.0013872 (2,131/1,536,188 alleles, 25 homozygotes), exome AF 0.00072539 (1,004/1,384,088), genome AF 0.00740960 (1,127/152,100) - all far above 0.0001.
• Highest ancestry-specific frequency: African/African American 2.67064% (655/24,526) in gnomAD v2.1 and 2.54492% (1,864/73,244) in gnomAD v4.1; grpmax FAF 0.025364 and 0.0244867. Even the rarest ancestry groups are not absent (e.g. Admixed American 42/35,108 in v2.1).
• gnomAD-Canada v1.0 (HostSeq genomes): 34/18,420 alleles, AF 0.0018458, 0 homozygotes, rs34058399 - also well above 0.0001.
• Non-cancer subsets unavailable: GNOMAD_V2_1_NON_CANCER and GNOMAD_V3_1_NON_CANCER both returned query_failed with frequency UNKNOWN, explicitly treated as missing data rather than evidence of absence; with no MSH3 VCEP specifying a non-cancer or exome-only source, the all-comers totals are the default, and the exome-only figures are above the threshold as well.
• Data-consistency note: the case evidence brief summarizes gnomAD-Canada as 'Absent from gnomAD-Canada v1.0' while the direct query record reports 34/18,420 alleles; the direct query is the primary record, and PM2 is not met under either reading.

Sources

Reasoning

PM3 (generic ACMG/AMP 2015) requires the variant to be detected in trans with a pathogenic variant in a recessive disorder. MSH3-associated disease is autosomal recessive: the established phenotype is biallelic MSH3-associated attenuated adenomatous polyposis (MSH3 partners with MSH2 to form MutS beta; germline MSH3 disease is reported only for biallelic carriers, with minimal evidence for a dominant or Lynch-syndrome role). No MSH3 VCEP/CSPEC specification exists, so the generic criterion applies (vcep_materials.json: found=false, generic_acmg_fallback). The only report of this exact variant, c.1567G>A p.(Glu523Lys), is in an inflammatory breast cancer sample (case IBC15) analysed by targeted NGS of a tumour biopsy, where it is flagged 'PG' (potentially germline, VAF < 0.5) alongside a second MSH3 missense change, c.1571A>C p.(Asn524Thr), also flagged PG; the paper explicitly states that constitutional/germline confirmation was not performed, so the phase (cis vs trans) of the two MSH3 alleles is undetermined. No pathogenic (or likely pathogenic) MSH3 variant in trans was identified in that or any other reported case, and the accompanying variant p.Asn524Thr carries no pathogenic assertion. ClinVar (Variation ID 780543) aggregates only classification-level submissions (8x Likely benign, 4x Benign, 1x VUS, 1x benign) with no expert-panel review and no genotype/phase information, and the submission audit yielded zero usable submissions, zero criterion leads and zero validated PMIDs. Because no proband-level phased genotype pairing this allele with a pathogenic MSH3 variant exists, the in-trans requirement is not satisfied. PM3 is not met.

Rule

ACMG/AMP 2015 PM3 (moderate; Richards et al. 2015, PMID:25741868): 'For recessive disorders, detected in trans with a pathogenic variant.' No gene-specific MSH3 specification exists, so the generic criterion is applied unchanged.

Evidence considered

• Generic ACMG/AMP PM3 definition applied (no MSH3 VCEP or ClinGen CSPEC specification retrieved; framework mode generic_acmg).
• PMID:33007869 (inflammatory breast cancer cohort, targeted NGS of 21 IBC biopsies): the exact variant MSH3 c.1567G>A (p.Glu523Lys) is listed in Table 2 for case IBC15 and annotated 'PG' (potentially germline, VAF < 0.5); the same sample carries a second MSH3 variant, c.1571A>C (p.Asn524Thr), also annotated 'PG'.
• PMID:33007869 states no germline confirmation was performed and does not report phase for the two co-occurring MSH3 variants, so no trans configuration can be established; the tumour context also prevents exclusion of a somatic origin.
• PMID:33007869 reports no pathogenic MSH3 allele in case IBC15 (the co-occurring p.Asn524Thr change carries no pathogenic assertion), so the essential PM3 comparator - a pathogenic variant in trans - is absent.
• ClinVar Variation ID 780543 carries only classification-level submissions (Likely benign x8 laboratories, Benign x4, Uncertain significance x1, benign x1), review status 'criteria provided, single submitter', no expert-panel review, and no genotype or phase information usable for PM3.
• The ClinVar submission audit for this variant returned zero submissions usable for classification and zero criterion leads, i.e. no submitter-reported trans configuration.
• MSH3 germline disease is biallelic/autosomal recessive (MutS beta mismatch-repair deficiency causing attenuated adenomatous polyposis); monogenic dominant inheritance is not established, so PM3 rests entirely on identifying a pathogenic allele in trans in a recessive proband, which no source provides.

Still missing

• No proband-level phased genotype is available; no case of biallelic MSH3 disease carrying c.1567G>A with a pathogenic second allele has been reported, and no parental/family phase testing exists.
• The single variant-level report (PMID:33007869, case IBC15) is from tumour-panel sequencing; germline status was not confirmed, so cis/trans configuration relative to the co-occurring MSH3 c.1571A>C (p.Asn524Thr) cannot be determined.
• No MSH3-specific ClinGen CSPEC/VCEP specification exists, so PM3 was assessed on the generic ACMG/AMP definition alone.

Sources

Reasoning

SCOPE: PM5 requires a different missense change at the same residue (MSH3 codon 523) to be established as pathogenic; p.Glu523Lys itself is the variant under assessment and cannot serve as its own comparator. FRAMEWORK: no MSH3 CSPEC/VCEP specification or local gene-specific framework exists (cspec found = false; gene_vcep_dir = null; framework_mode = generic_acmg), so the generic ACMG/AMP PM5 definition (PMID:25741868) applies. EVIDENCE: the case's own PM5 harvesting artifact found nothing - pm5_candidates.json reports found = false, total_candidates_seen = 0, eligible_for_classic_pm5_search = false, with the stated reason that classic same-residue PM5 semantics could not be confirmed safely and candidate harvesting was skipped, and its recommendation level is 'not_applicable'. Because no comparator lead was produced, a codon-level ClinVar check was run directly for this adjudication (MSH3[gene] AND Glu523[all]): the only non-synonymous MSH3 residues-523 variants are p.Glu523Lys (this variant, Benign/Likely benign, 15 submissions) and p.Glu523Val (c.1568A>T, VCV004829418, uncertain significance, single submitter, 2026-03-04); there is no established pathogenic change at residue 523. The nearest reported non-synonymous change is c.1571A>C p.Asn524Thr (PMID:33007869, case IBC15), which affects a different codon (residue 524) and is itself reported only as a potentially germline finding without classification, so it cannot support PM5. PMID:26467025's rule that a variant differing from a pathogenic variant in the same codon counts only as minor evidence (+0.5) presupposes a pathogenic variant at that codon, which does not exist here. PM5 is therefore not met.

Rule

Generic ACMG/AMP PM5: missense change at an amino-acid residue where a different, pathogenic missense change has been seen. Not met - no established pathogenic missense exists at MSH3 residue 523 (p.Glu523Val = uncertain significance; p.Glu523Lys = this variant, Benign/Likely benign). No MSH3 CSPEC/VCEP specification exists to modify this criterion.

Evidence considered

• PM5 requires a different missense change at the same amino-acid residue to have been determined pathogenic; p.Glu523Lys is the variant under assessment and cannot act as its own PM5 comparator.
• pm5_candidates.json for this case: found = false, pm5_mode = 'unknown', eligible_for_classic_pm5_search = false, total_candidates_seen = 0, candidates = [], recommendation level = 'not_applicable', with the reason 'Unable to confirm classic same-residue PM5 semantics safely; skipping automatic candidate harvesting' - no PM5 comparator lead exists in the case artifacts.
• Direct codon-level ClinVar check performed for this adjudication (MSH3[gene] AND Glu523[all], NCBI E-utilities): residue 523 carries p.Glu523Lys (c.1567G>A, this variant, Benign/Likely benign, 15 submissions), p.Glu523Val (c.1568A>T, VCV004829418, uncertain significance, single submitter, 2026-03-04) and the synonymous p.Glu523= (c.1569G>A, VCV001775448). There is no established pathogenic missense at residue 523, so the required comparator does not exist.
• PMID:33007869 reports the only adjacent-codon non-synonymous change in this case's literature, MSH3 c.1571A>C (p.Asn524Thr), in the same sample (IBC15) that carries c.1567G>A; it is residue 524, not residue 523, and is listed only as a potentially germline variant (VAF <0.5) with no classification or functional data, so it is not PM5 support (a different residue cannot satisfy PM5, and it is not established pathogenic).
• PMID:26467025 was consulted for PM5 semantics: it treats a variant that changes the amino acid in a codon containing pathogenic variants as minor evidence (+0.5) only when the amino acid differs from the pathogenic variant's. That rule presupposes a pathogenic variant at the same codon, and no such variant exists at MSH3 codon 523 here; PMID:26467025 contains no MSH3 data and does not mention c.1567G>A.
• Framework: no MSH3 CSPEC/VCEP or local gene-specific framework was available (cspec found = false, gene_vcep_dir = null, framework_mode = generic_acmg), so generic ACMG/AMP PM5 per Richards et al. 2015 (PMID:25741868) was applied.

Still missing

• The automated PM5 candidate harvest did not run (pm5_candidates.json: eligible_for_classic_pm5_search = false, 0 candidates), so the comparator set rests on the ClinVar codon-523 check performed for this adjudication rather than on a systematic residue-level literature search; a dedicated residue-523 literature search was not performed.
• No MSH3 CSPEC/VCEP exists to define gene-specific PM5 applicability.

Sources

Reasoning

SCOPE: PP2 requires both that the gene has a low rate of benign missense variation and that missense variants are a common mechanism of disease; p.Glu523Lys is a missense variant in MSH3, so both prongs were tested. FRAMEWORK: no MSH3 ClinGen CSPEC/VCEP specification or local gene-specific framework exists (cspec found = false; gene_vcep_dir = null; framework_mode = generic_acmg), so the generic ACMG/AMP PP2 definition (PMID:25741868) applies. EVIDENCE, prong 1 (missense tolerance): gnomAD gene-constraint metrics for MSH3 (retrieved from the gnomAD API for this adjudication) show observed missense variants 1459 versus 1347.6 expected, oe_mis = 1.083, mis_z = -1.2946 and pLI approximately 0 (5.9e-31). Missense variation is therefore not depleted in MSH3 - the gene tolerates missense change, which directly contradicts the 'low rate of benign missense variation' requirement (rather than being close to any constrained range, mis_z is negative). EVIDENCE, prong 2 (missense as a disease mechanism): the MSH3-associated disease mechanism is loss of function. ClinVar records for MSH3 carrying a pathogenic classification are overwhelmingly truncating - 319 frameshift and 161 nonsense records versus a single missense record (p.Gly1044Ser, single submitter, condition 'not provided') - and pvs1_gene_context.json documents germline MSH3 loss-of-function as the supported disease mechanism from a targeted germline literature review. MSH3-associated adenomatous polyposis is a recessive disorder driven by loss-of-function alleles, not a missense-driven disease. Neither prong is satisfied, so PP2 is not met.

Rule

Generic ACMG/AMP PP2: missense variant in a gene with a low rate of benign missense variation in which missense variants are a common mechanism of disease. Not met - MSH3 missense variation is not depleted (gnomAD oe_mis 1.083, mis_z -1.295, pLI ~0) and the established MSH3 pathogenic spectrum is truncating (ClinVar: 319 frameshift + 161 nonsense vs 1 missense).

Evidence considered

• PP2 requires (a) a gene with a low rate of benign missense variation and (b) missense variants being a common mechanism of disease; both prongs were evaluated for MSH3.
• Framework: cspec found = false, gene_vcep_dir = null, framework_mode = generic_acmg; generic ACMG/AMP PP2 per Richards et al. 2015 (PMID:25741868) was applied because no MSH3-specific specification exists.
• Prong 1 (gnomAD gene constraint for MSH3, retrieved from the gnomAD API for this adjudication): observed missense 1459 vs expected 1347.6, oe_mis = 1.0827, oe_mis_upper = 1.1305, mis_z = -1.2946, pLI = 5.91e-31. Missense variation is not depleted; the negative mis_z means MSH3 tolerates missense substitution, so the gene does not have a low rate of benign missense variation.
• Prong 2 (gene-level pathogenic spectrum, ClinVar via NCBI E-utilities, count of MSH3 records carrying a pathogenic classification): 319 frameshift variant records, 161 nonsense records and 1 missense record (c.3130G>A p.Gly1044Ser, VCV001476245, single submitter, condition 'not provided'). Missense change is therefore not a common established MSH3 disease mechanism.
• pvs1_gene_context.json documents that a targeted germline literature review supports MSH3 loss of function as the germline disease mechanism (lof_mechanism_supported = true; pvs1_gene_gate = 'eligible'), with the caution that no official CSPEC/VCEP or local explicit PVS1 framework exists; MSH3-associated adenomatous polyposis is recessive and loss-of-function driven.
• Conclusion: both PP2 prongs fail (missense is tolerated and missense is not a common disease mechanism), so PP2 is not met at any strength.

Still missing

• No MSH3 ClinGen CSPEC/VCEP specification exists to define gene-specific PP2 applicability, so the verdict rests on gnomAD missense-constraint metrics and the ClinVar gene-level pathogenic spectrum (gene-level evidence, not variant-level).

Sources

Reasoning

SCOPE: NM_002439.5:c.1567G>A (NC_000005.9:g.80024783G>A; GRCh38 chr5:80728964 G>A) is a missense substitution, NP_002430.3:p.(Glu523Lys), with VEP most_severe_consequence = missense_variant. It is not a frameshift, nonsense or other truncating/protein-length-changing variant, and it is not a canonical +/-1,2 splice-consensus change, so PP3/BP4 for this variant are taken from the missense protein-impact path (REVEL) only; the SpliceAI result is out of scope for this consequence type and is not used. FRAMEWORK: no MSH3 ClinGen CSPEC/VCEP specification and no gene-specific PP3/BP4 lookup spreadsheet exists for this case (cspec found = false; gene_vcep_dir = null; framework_mode = generic_acmg; no domain tables, no user VCEP files, no index guidance), so the generic ClinGen SVI/Pejaver REVEL calibration applies. EVIDENCE: the local REVEL v1.3 lookup returned 0.228 for this variant. PP3 requires REVEL >=0.644 (supporting), >=0.773 (moderate) or >=0.932 (strong) (Pejaver et al. 2022, Am J Hum Genet, PMID:36413997); 0.228 is below the lowest (supporting) threshold, so PP3 is not met at any strength. The SpliceAI result (max delta 0.001; DS_AL 0.001, all other deltas 0.000, variant_consequence = sequence_variant) is not evidence for or against PP3 for a missense variant because SpliceAI is calibrated for splice disruption, not for tolerance of an amino-acid substitution. The BayesDel score (-0.40543) was not used: no generic ACMG/AMP strength calibration is established for BayesDel and no MSH3 VCEP cutoff is available, so it is not_available for PP3/BP4.

Rule

Generic ACMG/AMP PP3 via the REVEL missense path: REVEL >=0.644 (supporting), >=0.773 (moderate), >=0.932 (strong) per ClinGen SVI REVEL calibration (Pejaver et al. 2022, PMID:36413997). No gene-specific MSH3 VCEP/PP3 override available.

Evidence considered

• Variant scope: NM_002439.5:c.1567G>A is a missense substitution, NP_002430.3:p.(Glu523Lys); VEP most_severe_consequence = missense_variant, and the variant is not a canonical +/-1,2 splice-consensus change, so the REVEL missense path governs PP3 and BP4 for this variant.
• No MSH3 CSPEC or VCEP specification and no gene-specific PP3/BP4 lookup table is available for this case (cspec found = false, gene_vcep_dir = null, framework_mode = generic_acmg, final_classification_framework source = generic_acmg_fallback), so generic ACMG/AMP computational-evidence calibration applies.
• Local REVEL v1.3 lookup returned a score of 0.228 for chr5:80728964 G>A (GRCh38), matching NM_002439.5:c.1567G>A / NP_002430.3:p.(Glu523Lys).
• PP3 REVEL thresholds applied verbatim from the published ClinGen SVI calibration: >=0.644 supporting, >=0.773 moderate, >=0.932 strong (Pejaver et al. 2022, Am J Hum Genet, PMID:36413997). REVEL 0.228 is below all three, so PP3 is not met at any strength.
• SpliceAI was deliberately not used for PP3: the recorded SpliceAI max delta was 0.001 (DS_AL 0.001, DS_AG/DS_DG/DS_DL 0.000, variant_consequence = sequence_variant). A splice predictor says nothing about whether an amino-acid substitution is tolerated, so this score neither supports nor refutes PP3 for this missense variant.
• BayesDel returned -0.40543 for this variant but was not used: no generic ACMG/AMP strength calibration is established for BayesDel and no MSH3 VCEP specifies a cutoff, so the score is not_available for PP3/BP4 (thresholds must come from a named publication or the governing specification).
• The in-silico relevance of the retrieved literature was checked: PMID:25741868 establishes the generic PP3/BP4 computational-evidence criteria, PMID:26467025 benchmarks SIFT/PolyPhen-2 as low-weight in silico evidence, and neither PMID:33007869 (the only paper that reports this exact variant) nor PMID:28492532 provides any computational prediction for p.(Glu523Lys).

Sources

Reasoning

FRAMEWORK: PP5 (Richards et al. 2015, generic ACMG/AMP 2015) is credited only for an exact-variant ClinVar assertion by an expert panel that is Pathogenic or Likely pathogenic; ordinary laboratory submissions, aggregate ClinVar labels and non-expert assertions must never trigger it. No MSH3 ClinGen CSPEC/VCEP specification exists for this case (vcep_materials.json found = false; framework_mode = generic_acmg; final_classification_framework source = generic_acmg_fallback), so no gene-specific PP5 override applies. EVIDENCE: ClinVar Variation ID 780543 (VCV000780543; NM_002439.5(MSH3):c.1567G>A p.Glu523Lys) records zero expert-panel submissions - source_data.expert_panels is an empty object and no submitter is a ClinGen/VCEP expert panel. The record's API review status is 'criteria provided, multiple submitters, no conflicts', i.e. 2 stars rather than the 3-star expert-panel review level. All 15 submissions come from 15 clinical laboratories (Sinai Health System; University of Chicago; Fulgent Genetics; GeneDx; ARUP Laboratories; Ambry Genetics; PreventionGenetics; Rigshospitalet; Quest Diagnostics; Labcorp/Invitae; Breakthrough Genomics; King Faisal Specialist Hospital; Mayo Clinic Laboratories; Dasa; Myriad Genetics). Their assertions are eight Likely benign, four Benign, one Uncertain significance and one benign: there is no Pathogenic or Likely pathogenic assertion of any kind, whether expert or laboratory-derived. The ClinVar submission audit for this variant independently returned zero expert-panel submissions and zero criterion leads. Because no exact-variant expert-panel Pathogenic/Likely pathogenic assertion exists, PP5 is not met.

Rule

PP5 applied under the expert-panel-only rule for this group plus the generic ACMG/AMP 2015 definition (supporting; Richards et al. 2015, PMID:25741868): only an exact-variant ClinVar expert-panel Pathogenic/Likely pathogenic classification supports PP5; aggregate ClinVar labels and ordinary laboratory submissions do not.

Evidence considered

• Governing rule applied: PP5 requires an exact-variant ClinVar expert-panel classification of Pathogenic or Likely pathogenic; ordinary laboratory submissions, aggregate ClinVar labels and non-expert assertions must never trigger PP5.
• ClinVar VCV000780543 (Variation ID 780543) for NM_002439.5(MSH3):c.1567G>A (p.Glu523Lys): source_data.expert_panels = {} (no expert-panel submissions) and API review status 'criteria provided, multiple submitters, no conflicts' (2 stars, not the 3-star expert-panel level).
• All 15 ClinVar submissions are from clinical laboratories (8x Likely benign, 4x Benign, 1x Uncertain significance, 1x benign); none is a Pathogenic or Likely pathogenic assertion, so no expert-panel pathogenic assertion exists to credit under PP5.
• The ClinVar submission audit for this case returned zero expert-panel submissions, zero submissions usable for classification and zero criterion leads, confirming the absence of an expert-panel assertion.
• No MSH3 CSPEC/VCEP specification was retrieved (vcep_materials.json found = false, framework_mode = generic_acmg), so the generic PP5 definition is applied and no gene-specific PP5 override exists.

Still missing

• No ClinVar expert-panel (ClinGen/VCEP) submission exists for MSH3 c.1567G>A, so the only evidence class that can trigger PP5 is absent; nothing further can be retrieved to satisfy the criterion.
• No MSH3 ClinGen CSPEC/VCEP specification exists, so no gene-specific PP5 rule or pre-scored PP5 code could be applied.

Sources

Reasoning

BA1 is the ACMG/AMP stand-alone benign allele-frequency criterion (AF >= 5%, PMID:25741868). No ClinGen CSPEC/VCEP or local gene-specific framework exists for MSH3 (cspec.found = false; final_classification_framework.source = 'generic_acmg_fallback'), so generic thresholds and the all-comers gnomAD totals (GNOMAD_V2_1, GNOMAD_V4_1) are the applicable default; the specified non-cancer subsets could not be used (both queries failed) and are not required absent an MSH3 VCEP. gnomAD v2.1 all-comers: AF 0.00254891 (716/280,904 alleles; 7 homozygotes), with the highest ancestry-specific frequency in African/African American 0.0267064 (2.6706%, 655/24,526 alleles; 7 homozygotes) and grpmax FAF 0.025364. gnomAD v4.1 all-comers: AF 0.0013872 (2,131/1,536,188; 25 homozygotes), African/African American 0.0254492 (2.5449%, 1,864/73,244; 24 homozygotes), grpmax FAF 0.0244867. The most stratified subgroup available in v4.1 (AFR_XY, 884/32,866) is 0.026897. Every measure - total AF, popmax AF and grpmax FAF, in both gnomAD releases - is below the 5% stand-alone threshold, most by roughly half. Even under the autosomal recessive model described for MSH3 (biallelic loss of function associated with attenuated adenomatous polyposis), a 2.5-2.7% allele frequency does not reach the 5% stand-alone cutoff. The frequency is not explicable by a technical threshold or mapping artifact: the allele is rs34058399, carries no LCR or segdup flag, and is detected in both exome (489/249,520 in v2.1) and genome (227/31,384 in v2.1) data in two independent gnomAD releases.

Rule

BA1: stand-alone benign when AF >= 0.05 (PMID:25741868); applied to the highest ancestry-specific frequency (grpmax/popmax).

Evidence considered

• gnomAD v2.1 all-comers (GRCh37, 5-80024783-G-A): total AF 0.00254891, 716/280,904 alleles, 7 homozygotes; exome AF 0.00195976 (489/249,520; 5 homozygotes) and genome AF 0.00723298 (227/31,384; 2 homozygotes).
• gnomAD v2.1 highest ancestry group: African/African American AF 0.0267064 (2.67064%, 655/24,526 alleles, 7 homozygotes); grpmax FAF 0.025364; no other ancestry exceeds 0.12% (Admixed American 0.0011963; European non-Finnish 0.0000855).
• gnomAD v4.1 all-comers (GRCh38, chr5-80728964-G-A): total AF 0.0013872, 2,131/1,536,188 alleles, 25 homozygotes; exome 0.00072539 (1,004/1,384,088; 9 homozygotes) and genome 0.00740960 (1,127/152,100; 16 homozygotes).
• gnomAD v4.1 highest ancestry group: African/African American AF 0.0254492 (2.54492%, 1,864/73,244 alleles, 24 homozygotes); grpmax FAF 0.0244867; the AFR_XY stratum reaches 0.026897 (884/32,866).
• BA1 threshold applied verbatim from the supplied published calibration: stand-alone benign at AF >= 0.05 (ACMG/AMP 2015, PMID:25741868), evaluated on the highest ancestry-specific frequency because a stand-alone frequency call must not be masked by a low global average.
• No MSH3 CSPEC/VCEP or local gene framework was available (cspec found = false; framework source 'generic_acmg_fallback', gene_vcep_dir null), so no gene-specific BA1 threshold exists and the generic 5% value governs.
• Non-cancer subsets were unavailable for BA1: GNOMAD_V2_1_NON_CANCER and GNOMAD_V3_1_NON_CANCER both returned query_failed with frequency UNKNOWN (missing data, not absence); with no MSH3 VCEP naming a non-cancer source, the all-comers totals remain the correct default.
• Threshold/liftover artifacts excluded as an explanation: the variant is dbSNP rs34058399 with region_flag lcr = false and segdup = false, concordantly detected across exome and genome data in both gnomAD releases, so its frequency is not a mapping or paralog artifact.
• The recessive inheritance context for MSH3 (biallelic loss of function associated with adenomatous polyposis; gene-level context OMIM 600887) does not raise the stand-alone threshold to the point of reaching 2.7%.

Sources

Reasoning

BS2 (strong benign) is not met, although the underlying population data are informative in the same direction as BS1. BS2 as defined in ACMG/AMP 2015 (PMID:25741868) requires observation in a healthy adult individual for a recessive (homozygous), dominant (heterozygous) or X-linked (hemizygous) disorder 'with full penetrance expected at an early age'. The gnomAD data do show homozygotes: 7 in v2.1 (all African/African American) and 24-25 in v4.1 (24 African/African American), plus 700+ heterozygous adults of both sexes (AFR_XX 337/13,346 and AFR_XY 318/11,180 in v2.1). Three reasons prevent calling BS2 met at strong strength here. (1) The disorder context does not satisfy the criterion's penetrance clause: MSH3-related disease is an autosomal recessive, adult-onset, incompletely penetrant attenuated adenomatous polyposis (gene-level context; biallelic MSH3 reported in a small number of families), not a fully penetrant early-onset condition, so a homozygous adult in a database cannot be equated with a definitively unaffected adult. (2) The population source is all-comers: the gnomAD totals include cancer-affected cohorts (which is why dedicated non-cancer subsets exist), and both non-cancer queries for this variant - GNOMAD_V2_1_NON_CANCER and GNOMAD_V3_1_NON_CANCER - returned query_failed with frequency UNKNOWN, so the unaffected status of the homozygotes cannot be independently verified. (3) Adenomatous polyposis is clinically silent and detected only endoscopically, so absence of a recorded phenotype in a population database does not establish health for this specific condition. The frequency-based benign evidence for this variant is therefore better captured by BS1 (met, strong, at popmax 2.67%) and, at the criterion level, BS2 is recorded as not met rather than applied at reduced strength, since ACMG/AMP does not define a supporting-strength BS2. Human review is flagged because a laboratory using a permissive homozygote-in-population convention for recessive genes could reasonably apply BS2 here.

Rule

BS2: strong benign for observation in a healthy adult individual with a recessive (homozygous), dominant (heterozygous) or X-linked disorder with full penetrance expected at an early age (PMID:25741868); the healthy-adult and full-penetrance conditions are not satisfied here.

Evidence considered

• Homozygotes observed in gnomAD v2.1: 7 total (5 exome, 2 genome), all 7 in African/African American; heterozygous carriers are numerous and include both sexes (AFR_XX 337/13,346; AFR_XY 318/11,180).
• Homozygotes observed in gnomAD v4.1: 25 total (9 exome, 16 genome), 24 in African/African American (AFR_XX 8; AFR_XY 16); 'remaining' populations contribute 1.
• BS2 wording applied verbatim from ACMG/AMP 2015 (PMID:25741868): observation in a healthy adult individual for a recessive (homozygous), dominant (heterozygous) or X-linked (hemizygous) disorder with full penetrance expected at an early age.
• Disease/penetrance context: MSH3 germline disease is autosomal recessive - biallelic MSH3 variants associated with attenuated adenomatous polyposis (gene-level context, OMIM 600887) - with adult onset and incomplete penetrance, so the criterion's early-onset/full-penetrance premise is not satisfied and database homozygotes cannot be assumed unaffected.
• Adenomatous polyposis is endoscopically detected and clinically silent, so absence of phenotype annotation in a population cohort does not establish the healthy-adult status BS2 requires.
• Population source limitation: the gnomAD all-comers totals include cancer cohorts, and both GNOMAD_V2_1_NON_CANCER and GNOMAD_V3_1_NON_CANCER queries failed (frequency UNKNOWN, treated as missing data), so unaffected status of the homozygotes could not be confirmed from the non-cancer subsets.
• The variant is also reported in ClinVar as Likely benign/Benign by most submitters (ClinVar variation 780543), but no ClinVar submission was used as an input, and ClinVar classification is not evidence for this criterion.

Still missing

• Homozygote genotypes in gnomAD are not phenotype- or age-annotated, and the non-cancer subsets that would exclude cancer cohorts both failed, so the 'healthy adult' premise of BS2 cannot be confirmed for the 7 (v2.1) / 24-25 (v4.1) homozygotes.

Sources

Reasoning

BS3 requires well-established in vitro or in vivo functional studies showing no damaging effect on the protein or gene product (Richards et al. 2015, PMID:25741868). No MSH3 VCEP or gene-specific specification governs this case (no ClinGen CSPEC; framework_mode = generic_acmg), so the generic ACMG/AMP 2015 definition applies. For the same reason as PS3, no functional assay of MSH3 NM_002439.5:c.1567G>A (NP_002430.3:p.Glu523Lys) exists in the assembled evidence: the only source reporting the variant, PMID:33007869, records it as a Table 2 targeted-panel finding in inflammatory breast cancer case IBC15 ('MSH3 (c.1567G > A; p.Glu523Lys) PG') with no assay, no MMR-competence or MutS-beta assembly measurement, and no functional claim for this allele; the remaining reviewed papers (PMID:25741868, PMID:28492532, PMID:26467025) are framework or methods documents with no MSH3 or variant data. Variant-specific resources are equally empty in the benign direction: OncoKB reports an unknown oncogenic effect with no variant-specific reviewed functional evidence, and the ClinVar submission audit returned no usable submissions or criterion leads, so no submitter functional assertion of a non-damaging effect can be inherited. Because the criterion is affirmative - it demands an established assay demonstrating preserved function - the absence of any qualifying functional study means BS3 is not met. The variant-level benign signal in this case comes from population data (gnomAD allele frequencies and homozygote observations), which is the population group's evidentiary route and is explicitly not a functional-assay result, so it cannot be used to meet BS3. Note also that the group's scope is a single missense allele in a recessively acting mismatch-repair gene, for which the relevant (and still unperformed) assay would be a mismatch-repair/MutS-beta functional test.

Rule

generic_acmg_2015:BS3_requires_established_functional_assay_showing_no_damaging_effect

Evidence considered

• Framework check: no ClinGen CSPEC or gene-specific VCEP specification exists for MSH3 (cspec found=false; no VCEP files, no SVI guidance; final_classification_framework found=false, generic_acmg fallback), so BS3 is assessed under the generic ACMG/AMP 2015 criterion definition with no gene-specific assay or strength calibration.
• Assay-availability review for the benign direction: no in vitro or in vivo functional study of NM_002439.5:c.1567G>A (p.Glu523Lys) was found in any assembled source; no MMR-activity assay, MutS-beta (MSH2-MSH3) assembly or stability assay, expression assay, or cell-based complementation assay reports this variant, so there is no assay result - damaging or non-damaging - to weigh.
• PMID:33007869 (the only source reporting this variant) provides the relevant negative finding. Direct quote: "MSH3 (c.1567G > A; p.Glu523Lys) PG" (Table 2, case IBC15, MMR Genes column). Context: tNGS 105-gene panel on 21 treatment-naive inflammatory breast cancer biopsies; the variant is annotated only as potentially germline (VAF<0.5). The paper reports no functional or protein-level assay of this allele; its immunohistochemistry covered ER/PR and MDM4/C-MYC, and no MSI/mismatch-repair functional result is attributed to this variant, so it cannot support BS3.
• PMID:25741868 full text reviewed: defines BS3 ('well-established functional studies show no damaging effect') and contains no MSH3 data and no mention of c.1567G>A or p.(Glu523Lys); the generic BS3 requirement of an established, validated assay is therefore unmet here.
• PMID:28492532 full text reviewed: Sherloc framework paper, no MSH3 data, no mention of this variant - no bearing on a BS3 functional determination beyond confirming that no gene-specific functional scoring path applies.
• PMID:26467025 full text reviewed: no MSH3 or variant data; its treatment of in silico tools as minor-weight evidence is not a functional assay result and cannot meet BS3.
• OncoKB variant page (MSH3 E523K): 'Unknown Oncogenic Effect', no variant-specific reviewed functional evidence and no citations - neither damaging nor non-damaging functional evidence is curated.
• ClinVar submission audit for this variant: 0 usable submissions, 0 criterion leads, no expert-panel submission - no submitter functional assertion of preserved function is available to support BS3.
• Population observations (gnomAD allele frequency and homozygotes) were noted only to explain why BS3 is not silently satisfied by a benign-leaning data set; they are not functional evidence and are not used as BS3 support.
• Gene-mechanism context: MSH3 functions as a mismatch-repair tumour suppressor via MutS-beta, with loss-of-function alleles associated with autosomal recessive adenomatous polyposis; a BS3 claim for a missense allele in this gene would require a validated MMR/MutS-beta functional assay showing normal function, which does not exist for this variant.

Still missing

• No variant-specific functional study demonstrating preserved (non-damaging) function for MSH3 c.1567G>A (p.Glu523Lys) exists in the assembled evidence, so BS3 cannot be met from functional data.
• No benign-direction functional assay exists for this allele; only population-frequency data point benign, and that evidence belongs to the population criterion, not BS3.
• MSH3 has no ClinGen CSPEC/VCEP-specified functional assay or BS3 strength calibration, so even a future assay would have to be evaluated against the generic ACMG/AMP BS3 requirements (assay validation, controls, relevance to the gene's loss-of-function mechanism).
• The assembled literature was limited to four ClinVar-linked PMIDs, three of which are framework/methods papers; a dedicated search for MSH3 functional studies of this allele was not part of the assembled evidence.

Sources

Reasoning

BP2 (generic ACMG/AMP 2015) is met either when the variant is observed in trans with a pathogenic variant for a fully penetrant dominant gene/disorder, or when it is observed in cis with a pathogenic variant in any inheritance pattern. Neither arm can be satisfied here. MSH3 has no established fully penetrant dominant disease; germline MSH3 disease is biallelic and autosomal recessive (attenuated adenomatous polyposis, MutS beta deficiency), so the first arm is inapplicable on gene/disease grounds. For the second (cis) arm, a pathogenic variant would have to be documented on the same allele as c.1567G>A; no pathogenic MSH3 variant was identified in any sample reported to carry this variant. The only variant-level report (PMID:33007869, IBC case IBC15) lists a co-occurring MSH3 missense change, c.1571A>C p.(Asn524Thr), that is not pathogenic, is annotated only as 'potentially germline' from tumour-panel data, and whose phase relative to c.1567G>A was not determined; the paper states germline/constitutional confirmation was not performed. ClinVar Variation ID 780543 provides no expert-panel review, no genotype data and no pathogenic MSH3 allele in cis. With no pathogenic variant in cis or in trans, BP2 is not met.

Rule

ACMG/AMP 2015 BP2 (supporting; Richards et al. 2015, PMID:25741868): 'Observed in trans with a pathogenic variant for a fully penetrant dominant gene/disorder or observed in cis with a pathogenic variant in any inheritance pattern.' No gene-specific MSH3 specification exists, so the generic criterion is applied unchanged.

Evidence considered

• Generic ACMG/AMP BP2 definition applied (no MSH3 VCEP or ClinGen CSPEC specification retrieved; framework mode generic_acmg).
• Arm 1 (trans with a pathogenic variant in a fully penetrant dominant gene/disorder) is inapplicable: MSH3 germline disease is biallelic and autosomal recessive (attenuated adenomatous polyposis via MutS beta loss of function), with minimal evidence for a Lynch-syndrome/dominant role.
• Arm 2 (in cis with a pathogenic variant, any inheritance pattern) is not satisfied: no pathogenic MSH3 variant is reported in any individual carrying c.1567G>A.
• PMID:33007869 (case IBC15) reports c.1567G>A alongside MSH3 c.1571A>C (p.Asn524Thr); both are annotated only 'PG' (potentially germline, VAF < 0.5) from tumour-biopsy targeted NGS, the second change is not classified pathogenic, and the paper states germline/constitutional alterations were not investigated - so neither cis nor trans configuration can be established.
• ClinVar Variation ID 780543 provides only classification-level submissions (predominantly Likely benign/Benign, single-submitter status, no expert panel) and no genotype, phase or in-cis pathogenic MSH3 allele information.

Still missing

• No phase-resolved genotype exists for the only reported carriers of c.1567G>A; cis configuration with a pathogenic MSH3 allele has never been tested or reported.
• The single variant-level report is tumour-panel based with no germline confirmation, so even the co-occurring MSH3 c.1571A>C change cannot be assigned to the same (cis) allele with confidence.
• No MSH3-specific ClinGen CSPEC/VCEP specification exists, so BP2 was assessed on the generic ACMG/AMP definition alone.

Sources

Reasoning

FRAMEWORK: BP5 (Richards et al. 2015, generic ACMG/AMP 2015) applies when the variant is found in a case for which an alternate molecular basis for the disease is established - i.e. the individual was ascertained for a phenotype that is explained by a different variant or gene. No MSH3 ClinGen CSPEC/VCEP specification exists for this case (cspec found = false; framework_mode = generic_acmg), so the generic definition governs. EVIDENCE: the only individual reported to carry NM_002439.5:c.1567G>A is case IBC15 of PMID:33007869, an inflammatory breast cancer biopsy analysed with a 105-gene targeted tumour panel, in which the variant is annotated 'PG' (potentially germline, VAF < 0.5). That case carries additional alterations - PALB2 c.43G>T (p.Glu15Ter) with copy-number LOH in the homologous-recombination column, and a second MSH3 missense change c.1571A>C (p.Asn524Thr) - but the paper presents these only as tumour-panel variant calls: it asserts no causal relationship for any of them, provides no clinical attribution of the case's disease to another gene or variant, and states that germline/constitutional confirmation was not performed, so the co-occurring events cannot even be assigned to the germline. No source reports a proband in whom c.1567G>A was identified during a Mendelian diagnostic workup whose phenotype was subsequently attributed to a different molecular cause; ClinVar VCV000780543 contributes only classification-level submissions with no case or genotype information. Because BP5 requires a documented case with an established alternate molecular basis and no such case exists in the retrieved evidence, BP5 is not met. The co-occurring PALB2 c.43G>T p.(Glu15Ter) with copy-number LOH in IBC15 is flagged for human review as the only BP5-adjacent observation, but a somatic tumour finding without an asserted causal role does not establish the criterion.

Rule

Generic ACMG/AMP 2015 BP5 (supporting; Richards et al. 2015, PMID:25741868): 'Variant found in a case with an alternate molecular basis for disease.' No MSH3 CSPEC/VCEP specification exists, so the generic criterion is applied unchanged.

Evidence considered

• Generic ACMG/AMP BP5 definition applied (no MSH3 VCEP or ClinGen CSPEC specification retrieved; framework_mode = generic_acmg): the variant is found in a case with an alternate molecular basis for disease.
• PMID:33007869 is the only source reporting the exact variant, in case IBC15 (inflammatory breast cancer) detected by a 105-gene targeted tumour-panel assay and annotated 'PG' (potentially germline, VAF < 0.5).
• Case IBC15 carries additional variants, including PALB2 c.43G>T (p.Glu15Ter) with copy-number LOH and a second MSH3 missense change c.1571A>C (p.Asn524Thr), but the paper presents them as tumour-panel calls without asserting that any of them explains the individual's disease, and it states that germline/constitutional alterations were not investigated.
• The co-occurring PALB2 truncation with copy-number LOH is a tumour-level observation, so it cannot be treated as an established germline alternate molecular basis for a Mendelian phenotype in this case.
• ClinVar VCV000780543 contains only classification-level laboratory submissions with no case information, so it supplies no case with an alternate molecular basis.
• No source describes a proband ascertained for an MSH3-related phenotype in whom c.1567G>A was found and the disease attributed to a different variant or gene; the required observation for BP5 is therefore absent.

Still missing

• No proband-level clinical information exists for the only reported carrier (IBC15, PMID:33007869): the sample is a tumour biopsy, germline status was not confirmed, and no phenotype workup or attribution of disease to another gene is reported.
• Case IBC15 does carry PALB2 c.43G>T (p.Glu15Ter) with copy-number LOH, but the paper does not state that this explains the individual's disease and cnLOH indicates a tumour event; human review is needed if BP5 is to be considered on that basis.
• No MSH3 CSPEC/VCEP specification exists to define what would count as an alternate molecular basis for MSH3-related disease.

Sources

Reasoning

FRAMEWORK: BP6 (Richards et al. 2015, generic ACMG/AMP 2015) is credited only for an exact-variant ClinVar assertion by an expert panel that is Benign or Likely benign; aggregate ClinVar labels and non-expert laboratory submissions must never trigger it, however concordant. No MSH3 ClinGen CSPEC/VCEP specification exists for this case (vcep_materials.json found = false; framework_mode = generic_acmg), so no gene-specific BP6 override applies. EVIDENCE: the ClinVar record for VCV000780543 (Variation ID 780543; NM_002439.5(MSH3):c.1567G>A p.Glu523Lys) shows an aggregate classification of 'Benign/Likely benign' with review status 'criteria provided, multiple submitters, no conflicts' (2 stars; last updated 2026-06-27; 15 submissions from 15 clinical laboratories). That consensus is entirely non-expert: source_data.expert_panels is an empty object, no submitter is a ClinGen/VCEP expert panel, and the review level is 2-star rather than the 3-star expert-panel level this criterion requires. Although 14 of 15 laboratory submissions call the variant Benign or Likely benign (eight Likely benign, four Benign, one benign, one Uncertain significance), an aggregate of ordinary laboratory assertions is exactly the evidence class that must not trigger BP6. The submission audit likewise identified zero expert-panel submissions. BP6 is therefore not met; the concordant but non-expert benign consensus is recorded as a human-review flag rather than scored as evidence.

Rule

BP6 applied under the expert-panel-only rule for this group plus the generic ACMG/AMP 2015 definition (supporting; Richards et al. 2015, PMID:25741868): only an exact-variant ClinVar expert-panel Benign/Likely benign classification supports BP6; aggregate ClinVar labels and ordinary laboratory submissions do not.

Evidence considered

• Governing rule applied: BP6 requires an exact-variant ClinVar expert-panel classification of Benign or Likely benign; aggregate ClinVar labels and non-expert assertions must never trigger BP6.
• ClinVar VCV000780543 for NM_002439.5(MSH3):c.1567G>A (p.Glu523Lys): aggregate classification 'Benign/Likely benign', API review status 'criteria provided, multiple submitters, no conflicts' (2 stars, last updated 2026-06-27), 15 submissions from 15 clinical laboratories.
• source_data.expert_panels = {} for this record and no submitter is a ClinGen/VCEP expert panel; the review level is 2-star, not the 3-star expert-panel level required for BP6.
• Submission breakdown: 8x Likely benign, 4x Benign, 1x Uncertain significance, 1x benign, all from ordinary clinical laboratories (Sinai Health System; University of Chicago; Fulgent Genetics; GeneDx; ARUP Laboratories; Ambry Genetics; PreventionGenetics; Rigshospitalet; Quest Diagnostics; Labcorp/Invitae; Breakthrough Genomics; King Faisal Specialist Hospital; Mayo Clinic Laboratories; Dasa; Myriad Genetics).
• The ClinVar submission audit for this case returned zero expert-panel submissions, confirming that the benign consensus is not expert-panel derived and must not be scored under BP6.
• No MSH3 CSPEC/VCEP specification was retrieved (vcep_materials.json found = false, framework_mode = generic_acmg), so the generic BP6 definition plus the expert-panel-only rule governs.

Still missing

• No ClinVar expert-panel (ClinGen/VCEP) benign assertion exists for MSH3 c.1567G>A, so the only evidence class that can trigger BP6 is absent.
• Fourteen of fifteen laboratory submissions classify the variant Benign or Likely benign with no conflicts; this concordant non-expert signal cannot be scored under the expert-panel-only rule and is left for human review.
• No MSH3 ClinGen CSPEC/VCEP specification exists, so no gene-specific BP6 rule or pre-scored BP6 code could be applied.

Sources

Reasoning

PS2 (de novo occurrence, with maternity and paternity confirmed, in a patient with the disease and no family history) requires four things this record does not contain: a proband, a phenotype consistent with the gene-disease relationship, confirmed biological parentage, and absence of family history. The case input is variant-only (NM_002439.5:c.1567G>A, NP_002430.3:p.(Glu523Lys) in MSH3); the consolidated record contains no proband, no clinical phenotype, no statement of parental testing, and no pedigree. The only variant-specific case-level report located (PMID:33007869, inflammatory breast cancer cohort) lists this variant in a single sample, case IBC15, annotated 'PG' (potentially germline, VAF<0.5); no parents were tested and no trio or pedigree was reported, so de novo status for this variant is simply unknown rather than absent. No MSH3-specific VCEP/CSPEC specification exists in the reviewed materials (cspec found=false, gene_vcep_dir=null), so the generic ACMG/AMP 2015 criteria (PMID:25741868) govern this case, and under those criteria PS2 is strength-calibrated on confirmed de novo occurrence in an affected proband - a prerequisite that is entirely unevidenced here. For context, the variant is not rare in population controls (gnomAD v2.1 AF 0.25489%, 716/280904 alleles with 7 homozygotes; gnomAD v4.1 AF 0.13872%, 2131/1536188 alleles with 25 homozygotes; gnomAD-Canada AF 0.1846%, 34/18420 alleles), which is difficult to reconcile with a highly penetrant de novo pathogenic allele and is a material caveat should a de novo claim ever be advanced for this variant.

Evidence considered

• Case record is variant-only: no proband identifier, no phenotype/HPO terms, and no family-history field are present in any of the consolidated case files.
• PMID:33007869 (targeted NGS of 105 cancer genes in 21 inflammatory breast cancer cases) reports MSH3 c.1567G>A p.Glu523Lys in case IBC15 flagged 'PG' (potentially germline, VAF<0.5); no parental testing, trio, or pedigree was performed, so the variant's de novo status could not be established.
• No MSH3-specific VCEP/CSPEC criteria specification is available (cspec found=false, gene_vcep_dir=null), so the generic ACMG/AMP 2015 definition of PS2 (de novo with confirmed maternity and paternity in an affected patient with no family history) applies as written.
• Searches of the case record for proband, de novo, parental, segregation, phenotype, family history, and meioses returned no case-level clinical content, confirming the gap is missing data rather than a negative finding.
• Population-frequency context checked to test whether a de novo pathogenic claim is even plausible: gnomAD v2.1 total AF 0.25489% (716/280904 alleles, 7 homozygotes), gnomAD v4.1 AF 0.13872% (2131/1536188 alleles, 25 homozygotes), gnomAD-Canada v1.0 AF 0.1846% (34/18420 alleles, 0 homozygotes).

Still missing

• No proband or clinical phenotype recorded for this case.
• No parental genotype results (maternity/paternity confirmation) - de novo status unknown.
• No family history or pedigree recorded.
• No gene-specific (VCEP/CSPEC) MSH3 specification exists; generic ACMG/AMP 2015 requirements for confirmed de novo evidence cannot be satisfied from the available data.

Sources

Reasoning

PM6 (assumed de novo, but without confirmation of maternity and paternity) requires at least an assertion of a de novo occurrence that has not been parentally confirmed. Nothing in this record asserts that NM_002439.5:c.1567G>A arose de novo: there is no proband, no unaffected-parent testing, no trio, and no statement of parentage one way or the other. The single case-level report of this variant (PMID:33007869, case IBC15) describes a potentially germline finding from tumour/normal sequencing of an inflammatory breast cancer sample, with no parental samples analysed and no family history given; that is a cancer-cohort observation, not a transmitted-versus-de-novo observation. No MSH3-specific VCEP/CSPEC specification exists (cspec found=false), so generic ACMG/AMP 2015 (PMID:25741868) governs, under which PM6 still presupposes a de novo event in an affected patient. Because no such event is even claimed, the criterion cannot be assessed from this evidence; the gap is the complete absence of family/parental data rather than a negative result. Weakens any future de novo argument: the variant is present in gnomAD with multiple homozygotes (v2.1 AF 0.25489%, 7 homozygotes; v4.1 AF 0.13872%, 25 homozygotes).

Evidence considered

• No field in the consolidated case record states or implies a de novo origin for NM_002439.5:c.1567G>A; the case provides only the variant, with no proband, parental, or pedigree information.
• PMID:33007869 reports this exact variant once, in inflammatory breast cancer case IBC15, as a potentially germline variant (VAF<0.5) from a 105-gene targeted NGS panel; the study tested tumour material from 21 treatment-naive patients and performed no parental genotyping.
• PMID:28492532 (Sherloc refinement of ACMG/AMP) was reviewed for any expanded de novo scoring that might apply; it is a generic framework paper with no MSH3 or variant-specific data.
• No MSH3 VCEP/CSPEC specification is available (cspec found=false, gene_vcep_dir=null), so generic ACMG/AMP 2015 PM6 (assumed de novo without confirmation of maternity and paternity) applies as written.
• Population-frequency context: gnomAD v2.1 total AF 0.25489% (716/280904 alleles, 7 homozygotes); gnomAD v4.1 AF 0.13872% (2131/1536188 alleles, 25 homozygotes); gnomAD-Canada v1.0 AF 0.1846% (34/18420 alleles).

Still missing

• No proband or affected individual described for this case.
• No parental samples or trio data - absence of confirmation is not itself an assumption of de novo status.
• No family history or pedigree recorded.
• No MSH3-specific VCEP/CSPEC specification available to substitute gene-specific de novo rules.

Sources

Reasoning

PP1 requires co-segregation of the variant with disease across multiple affected family members in a gene known to cause the disease. This case supplies no pedigree, no affected relatives, no meioses counted, and no phenotype, so the co-segregation premise cannot be tested at all. The literature reviewed for this case contains no segregation study for MSH3 c.1567G>A: the only variant-specific report (PMID:33007869, case IBC15) is a single tumour sample from a cancer cohort with no family study, and the remaining retrieved papers are generic methodology documents (PMID:25741868 ACMG/AMP 2015, PMID:28492532 Sherloc, PMID:26467025 laboratory scoring system) that contain no MSH3 variant data. No MSH3-specific VCEP/CSPEC specification exists (cspec found=false, gene_vcep_dir=null), so generic ACMG/AMP 2015 governs. PP1 is therefore not assessed: the evidence needed (an informative pedigree with multiple affected carriers) is absent rather than contradicted. Two contextual factors would make a future co-segregation claim hard to substantiate: the variant is common in controls (gnomAD v2.1 AF 0.25489%; v4.1 AF 0.13872%) with 7 and 25 homozygotes respectively, and ClinVar contains only non-expert single-submitter assertions for it, which cannot substitute for family data.

Evidence considered

• The consolidated case record contains no pedigree, no affected family members, and no meioses; PP1's requirement for co-segregation in multiple affected relatives cannot be evaluated.
• PMID:33007869 (the only retrieved source that mentions this exact variant) reports it in a single inflammatory breast cancer sample (case IBC15) with no family members genotyped and no segregation analysis.
• PMID:25741868 (ACMG/AMP 2015) defines PP1 as supporting co-segregation evidence and notes that gene/disease-specific groups should define applicability and weight; PMID:28492532 (Sherloc) refines segregation evidence weighting but contains no MSH3 data.
• Literature triage for this case surfaced no segregation-themed papers; the high-priority set comprised case-observation/functional themes only, with none reporting family studies of this variant.
• Context checked against population data: gnomAD v2.1 AF 0.25489% (716/280904 alleles, 7 homozygotes) and gnomAD v4.1 AF 0.13872% (2131/1536188 alleles, 25 homozygotes), which would be difficult to reconcile with a rare co-segregating pathogenic allele.
• ClinVar (ClinVarID 780543) holds only non-expert single-submitter assertions for this variant; these final classifications were not used to infer PP1 in either direction.

Still missing

• No pedigree and no affected relatives reported for this case.
• No number of informative meioses available to evaluate co-segregation.
• No segregation study for this variant identified in the literature reviewed.
• No MSH3 VCEP/CSPEC specification defining gene-specific co-segregation requirements.

Sources

Reasoning

FRAMEWORK: PP4 (Richards et al. 2015, generic ACMG/AMP 2015) requires that the patient's phenotype or family history be highly specific for a disease with a single genetic etiology. No MSH3 ClinGen CSPEC/VCEP specification exists for this case (cspec found = false; gene_vcep_dir = null; framework_mode = generic_acmg), so only the generic criterion is available. EVIDENCE AVAILABLE: this case contains no proband-level clinical information - no phenotype description, no HPO terms, no age at onset, no family history, and no stated indication for testing - so neither PP4 arm (phenotype specificity or family history) can be evaluated for the individual in whom c.1567G>A was found. The only phenotype-adjacent data are the ClinVar condition tags attached to VCV000780543 (Endometrial carcinoma, Familial adenomatous polyposis 4, Hereditary cancer-predisposing syndrome, MSH3-related disorder, 'not provided', 'not specified'), which are generic panel-testing condition labels aggregated across 15 independent laboratories rather than a proband-specific phenotype, and which are themselves heterogeneous and partly non-specific (a recessive polyposis syndrome listed together with broad cancer-predisposition and unspecified labels) - the opposite of a phenotype highly specific for a single-gene disorder. The only individual reported to carry the variant (PMID:33007869, case IBC15) is an inflammatory breast cancer tumour-panel sample; the reported context is a somatic tumour diagnosis, MSH3-related disease is biallelic/attenuated adenomatous polyposis rather than breast cancer, and the paper reports no clinical phenotype detail for that individual. Because the decisive input for PP4 (the affected individual's phenotype and family history) is absent rather than merely negative, PP4 is recorded as not assessed; a proband phenotype with HPO terms, or a documented family history, is required to adjudicate it.

Rule

Generic ACMG/AMP 2015 PP4 (supporting; Richards et al. 2015, PMID:25741868): 'Patient's phenotype or family history is highly specific for a disease with a single genetic etiology.' No MSH3 CSPEC/VCEP specification exists; criterion not assessable because no proband phenotype or family history is available.

Evidence considered

• Generic ACMG/AMP PP4 definition applied (no MSH3 VCEP or ClinGen CSPEC specification retrieved; framework_mode = generic_acmg): PP4 requires the patient's phenotype or family history to be highly specific for a disease with a single genetic etiology.
• This case contains no proband phenotype record: no clinical description, HPO terms, age at onset, testing indication, or family history is present in the case evidence, so PP4 has no assessable input.
• ClinVar VCV000780543 condition tags (Endometrial carcinoma; Familial adenomatous polyposis 4; Hereditary cancer-predisposing syndrome; MSH3-related disorder; 'not provided'; 'not specified') are aggregate panel-testing labels from 15 independent clinical laboratories, not a proband-specific phenotype, and are heterogeneous rather than specific to one disease.
• PMID:33007869, the only paper reporting the exact variant, places it in an inflammatory breast cancer tumour-panel sample (case IBC15) with no clinical phenotype detail for that individual; the reported disease context does not match the recessive attenuated-polyposis phenotype attributed to MSH3, and provides no phenotype-specificity evidence for PP4.

Still missing

• No proband-level phenotype, HPO terms, age at onset, or testing indication is present in this case, so phenotype specificity cannot be judged.
• No family history is available (no pedigree, no affected relatives), so the family-history arm of PP4 cannot be evaluated.
• The ClinVar condition list for VCV000780543 reflects multi-gene panel testing indications across 15 laboratories and is not proband-specific; it cannot substitute for the patient's own phenotype.
• No MSH3 CSPEC/VCEP specification exists to define what phenotype would count as highly specific for MSH3-related disease.

Sources

Reasoning

BS4 requires lack of segregation of the variant with disease in affected family members, i.e. an informative pedigree in which affected relatives who carry the variant are absent or affected non-carriers are observed. No pedigree and no family members are described anywhere in this case record, so there is no family in which non-segregation could be demonstrated and no meioses to count. The only variant-specific source retrieved (PMID:33007869, case IBC15) tested a single tumour sample and genotyped no relatives. The remaining reviewed papers are generic methodology documents (PMID:25741868, PMID:28492532, PMID:26467025) with no MSH3 or variant-level family data. No MSH3-specific VCEP/CSPEC specification exists (cspec found=false, gene_vcep_dir=null), so generic ACMG/AMP 2015 governs. Non-segregation was explicitly not inferred from the ClinVar benign/likely-benign submissions for this variant (ClinVarID 780543), because a final classification by another submitter is not segregation evidence and cannot be used to satisfy BS4; the criterion is therefore not assessed, with the missing pedigree as the identified gap.

Evidence considered

• No pedigree, family members, or segregation data appear anywhere in the consolidated case record for NM_002439.5:c.1567G>A, so absence of co-segregation cannot be established.
• PMID:25741868 (ACMG/AMP 2015) defines BS4 as lack of segregation in affected family members; PMID:28492532 (Sherloc) refines segregation evidence weighting, but neither contains MSH3 or variant-specific data.
• PMID:33007869, the only retrieved source naming this variant, reports it in one inflammatory breast cancer sample (case IBC15) with no relatives genotyped and no segregation analysis.
• ClinVar submissions for this variant (ClinVarID 780543; majority benign/likely benign from non-expert single submitters) were deliberately not converted into BS4 support, since another submitter's final classification is not segregation evidence.
• No MSH3-specific VCEP/CSPEC specification is available (cspec found=false, gene_vcep_dir=null), so generic ACMG/AMP 2015 governs this criterion.

Still missing

• No pedigree or family members available, so non-segregation cannot be demonstrated.
• No carrier/non-carrier phenotype status recorded for any relative.
• No BS4-directed segregation data in any retrieved publication.
• No MSH3 VCEP/CSPEC specification defining gene-specific non-segregation requirements.

Sources