Five-step scale from benign to pathogenic. This variant is likely benign. Select a step to read what it means.
This variantScenario
BenignPathogenic
Likely benign.A likely benign classification under the recorded framework.
In plain language
Gene-level background from the recorded interpretation.
Words with a dotted underline have a short definition. Hover or tap them.
About MSH3
MSH3 encodes a protein that partners with MSH2 to form MutS beta, a key component of the DNA mismatch repair system that recognizes and corrects errors made during DNA replication. It acts as a tumor suppressor: loss of MSH3 function leads to mismatch repair deficiency, an elevated mutation rate, and increased cancer risk, and the gene is frequently altered in mismatch repair-deficient colorectal cancers. Defects in MSH3 are linked to susceptibility to endometrial cancer, and inherited mutations are associated with an autosomal recessive form of familial adenomatous polyposis, with minimal evidence for a role in Lynch syndrome. Tumors with mismatch repair deficiency, including those involving MSH3, tend to respond well to immune checkpoint inhibitor therapies.
LYFE Sciences is an AI system, and it can make mistakes.
Criteria may be applied incorrectly, sources may be misread, and a
confident-looking classification can still be wrong.
Double-check every criterion and its underlying evidence
before relying on any call.
01The reasoning
How it was classified
The recorded framework is Not recorded. The criteria and strengths shown below come from the saved assessment; this page does not recalculate the classification.
Recorded applied criteria
The applied criteria and their strengths are shown as assessed under the stated framework.
No pathogenic criterion marked as applied in the recorded assessment.
Benign criteria applied · 2
BP1Supporting
BP1 · supporting
Met at supporting: MSH3 disease is caused by truncating variants - ClinVar MSH3 P/LP records are frameshift/nonsense/splice versus 1170 missense VUS.
BP4Supporting
BP4 · supporting
Met at supporting strength: REVEL 0.191 is within the <=0.29 BP4 supporting threshold but above the <=0.183 moderate cut-off.
Recorded classification: Likely benign
Likely Benign: BP1 (supporting, missense in a truncating-spectrum gene) plus BP4 (supporting, REVEL 0.191) reach two supporting benign criteria under generic ACMG/AMP 2015.
02The sources
Evidence at a glance
Evidence summaries and captured source records, where available. Missing entries should not be interpreted as negative findings.
gnomAD v4.1
gnomAD v2.1
Population frequency
392 / 1,612,578 alleles
gnomAD v4.1: 392 / 1,612,578 alleles. Recorded AF 0.000243089. Recorded homozygote count 2.
gnomAD v2.1
This variant is present in gnomAD v2.1 (AF= 0.000375266; MAF= 0.03753%, 106/282466 alleles, homozygotes = 0) and has highest observed frequency in the African/African American population (AF= 0.00387503; MAF= 0.38750%, 96/24774 alleles, homozygotes = 0); grpmax FAF= 0.00348064.
gnomAD-Canada
This variant is present in gnomAD-Canada v1.0 (AF= 0.00016286644951140066, 3/18420 alleles, homozygotes = 0).
OncoKB did not identify variant-specific reviewed functional evidence for this variant; gene-level curated context is available for reviewer follow-up. MSH3, a DNA mismatch repair protein, is frequently mutated in colorectal cancers.
ACMG/AMP criteria under Not recorded, with the data and sources behind each one.
28 criteria with recorded statuses: 2 met · 13 not met · 6 not assessed · 7 not applicable. Open any row for the full reasoning, the rule and the evidence behind it.
Group scope: codon/residue/domain criteria (PS1, PM1, PM5, PP2, BP1) for MSH3 NM_002439.5:c.1522A>G, NP_002430.3:p.(Lys508Glu), a missense substitution in the DNA mismatch repair gene MSH3.
No gene-specific specification exists for MSH3: no CSPEC was available (vcep_materials.json: 'no cspec url available'), no local gene framework was supplied, final_classification_framework.json resolved to generic_acmg_fallback, and the domain_tables key is empty. All five criteria were therefore adjudicated against generic ACMG/AMP 2015.
PS1 not met: the same amino-acid change (p.Lys508Glu) is not established as pathogenic (ClinVar concludes 'Conflicting classifications of pathogenicity'; 5 Likely benign, 1 VUS, no P/LP), and no alternative nucleotide change can generate p.Lys508Glu from the AAA (Lys) codon 508.
PM1 not met: no VCEP domain table exists, residue Lys508 sits in MutS domain II (connector region) of MSH3 rather than a critical ATPase/mismatch-recognition domain, it is not a statistically significant hotspot (absent from COSMIC, OncoKB 'Unknown Oncogenic Effect'), and the region is not free of benign variation (gnomAD v4 popmax AF 0.44%, 2 homozygotes).
PM5 not met: the only other codon-508 missense change in ClinVar, c.1523A>G (p.Lys508Arg), is of conflicting significance and not established pathogenic; the remaining codon-508 records are synonymous (c.1524A>G, Likely benign) or a frameshift (c.1524del). The same-residue comparator search skipped by pm5_candidates.json was completed directly against ClinVar here.
PP2 not met: MSH3 is not depleted of benign missense variation (gnomAD mis_z = -1.29, oe_mis = 1.08, pLI approximately 0) and missense change is not an established mechanism of MSH3 disease, whose established pathogenic variants are truncating.
BP1 met at supporting strength: MSH3 disease is caused by primarily truncating, loss-of-function alleles, and this variant is a missense substitution; no escalation of strength is warranted given the recessive biallelic mechanism and the existence of a few singleton missense P/LP MSH3 assertions.
Residue/domain summary: Lys508 (NP_002430.3) lies within MutS domain II (Pfam PF05188, residues 367-521); MSH3's ATPase MutS domain V (PF00488, residues 892-1094) contains the Walker-A motif at 896-903. There is no evidence that codon 508 is a mutation hotspot or a domain critical enough to support PM1, and the residue itself carries common benign population variation.
Five further full texts fetched into this case (PMID:24493721, PMID:24929052, PMID:25394175, PMID:26467025, PMID:28492532) were screened via paper_extracts.json and contain no MSH3- or codon-508-specific content relevant to this group's criteria; none is cited for these criteria.
Governing framework: MSH3 has no ClinGen CSPEC/VCEP or local gene-specific specification (case_summary.json cspec.found = false; vcep_materials.json 'no cspec url available'; final_classification_framework.source = generic_acmg_fallback), so the generic ACMG/AMP 2015 criteria (PMID:25741868) govern this group.
All four criteria assigned to this group (PS2, PM6, PP1, BS4) are genotype/family-based: PS2 and PM6 require a de novo observation in a proband (confirmation of parental non-transmission being the only difference between them), PP1 requires co-segregation in multiple affected relatives, and BS4 requires demonstrated non-segregation in affected relatives.
This case is a variant-level annotation of NM_002439.5:c.1522A>G (NP_002430.3:p.(Lys508Glu)) with no proband, no phenotype, no pedigree, no parental testing and no relative genotypes anywhere in review_manifest.json, prefetch.json, case_summary.json or evidence.json; the MSH3 gene-context record states inheritance = [] and the ClinVar record (VCV000781999, 7 submissions, conflicting classifications, no expert panel) records origin only as germline/unknown/not applicable with no de novo report.
Literature provides no family-level evidence either: all seven retrieved full texts (PMID:24493721, 24929052, 25394175, 25741868, 26467025, 28492532, 28944238) were searched for this variant's notations and none mentions it, so no paper reports this variant in any family. Consequently PS2, PM6, PP1 and BS4 are all returned not_assessed - the required data type (trio genotypes, pedigree, meioses) was never generated or documented - rather than being scored as refuted.
Population context is recorded but not converted into a de novo judgement: the variant is present in gnomAD v2.1 (106/282466 alleles, 96 in African/African American; popmax AF 0.00388) and gnomAD v4.1 (392/1612578 alleles, 2 homozygotes; popmax AF 0.00439), which is consistent with an inherited, relatively common allele rather than a confirmed de novo event, but confirms neither PS2/PM6 nor BS4.
No criterion in this group contributes evidence toward the final classification; a clinical resolution would require obtaining trio/parental genotypes and a family pedigree for a proband carrying this variant.
Functional group (PS3, BS3) contribution to the classification: none. Neither criterion is met because no variant-specific functional study of MSH3 c.1522A>G p.(Lys508Glu) was identified in any retrieved source (OncoKB, ClinVar submissions, the seven fetched full-text papers, or an index-level PubMed search for the variant).
PS3 not met (no assay showing a damaging effect) and BS3 not met (no assay showing preserved function) is the expected symmetric outcome when functional data are entirely absent; the absence of an assay must not be converted into benign functional evidence, and neither criterion carries any strength.
The applicable specification is generic ACMG/AMP 2015: no ClinGen VCEP/CSPEC or gene-specific framework for MSH3 was found for this case, so no gene-specific PS3/BS3 calibration or accepted-assay list was available to grade a hypothetical functional result.
Residual uncertainty is about missing data rather than conflicting data, so no human review flag was raised; if a variant-level MutS-beta mismatch-repair assay for p.Lys508Glu is later published, PS3/BS3 should be re-adjudicated and calibrated against that assay's controls.
Framework: MSH3 has no ClinGen CSPEC and no gene-specific VCEP specification in this case (cspec found=false; vcep_materials.json files_found=[]; final_classification_framework.json framework_complete=false, source=generic_acmg_fallback), so the generic ACMG/AMP 2015 framework (Richards et al., PMID:25741868, with the local generic combination rules file) governs all five criteria in this group.
Numeric-threshold direction check: under the governing framework none of PS4, PP4, PP5, BP5 or BP6 is defined with a numeric threshold or a comparison operator (no >=, <=, >, < is written for any of them; the generic PS4 enrichment requirement and PP4 specificity requirement are qualitative). No inequality was therefore applied or invented, and no criterion was adjudicated by pattern-matching a direction from another criterion; had a VCEP-defined likelihood-ratio-style threshold existed for BP5 it would have required its own explicitly quoted operator, but no such specification exists for MSH3.
PS4 not_met: the only variant-specific enrichment question that could be asked is whether this variant is over-represented in CRC cases. The single MSH3 cohort dataset retrieved (PMID:28944238; 1231 CRC cases, 93 spouse controls) does not contain the variant at all, no case series of the variant was found, and gnomAD v4.1 shows the variant at 392/1,612,578 alleles with 2 homozygotes in unselected population samples, i.e. no observed case enrichment.
PP4 not_assessed: PP4 is a proband-level criterion and no phenotype, family history or clinical indication for the tested individual exists anywhere in this case's evidence; with its input absent it cannot be judged met or not_met, and the ClinVar condition labels are submitter-entered and too broad to substitute for a highly specific phenotype.
PP5 not_applicable / BP6 not_applicable: both require an exact-variant ClinGen expert-panel assertion (pathogenic side and benign side respectively). ClinVar VCV000781999 has 7 submissions, 0 from expert panels, review status 1 star with conflicting classifications; the individual laboratory calls (5 Likely benign, 1 VUS, 1 unclassified) and the conflicting aggregate label are excluded from use by this group's rule, so neither criterion can be triggered.
BP5 not_assessed: BP5 requires observing this variant in a case whose disease is explained by an alternate molecular basis, but no source in this case reports any patient carrying c.1522A>G/p.(K508E) and no co-occurring pathogenic variant or tumour MMR/MSI data are available for a carrier, so the criterion has no case to evaluate.
Net effect of this group on the case: no pathogenic-direction evidence (PS4, PP5) and no benign-direction evidence (BP5, BP6) is contributed; PP4 remains open pending proband phenotype data. Deterministic follow-up: the ClinVar expert-panel gate is satisfied for neither PP5 nor BP6, and the PS4 case-control search is empty.
Both allelic criteria assessed here require a second variant with known phase, and no such variant exists in this case: PM3 (recessive, pathogenic variant in trans) and BP2 (pathogenic variant in cis, or in trans in a fully penetrant dominant disorder) are both not met.
MSH3-associated disease is recessive (ClinVar conditions include Familial adenomatous polyposis 4 / MSH3-related disorder; literature-based gene context supports a germline loss-of-function mechanism), so PM3 was applicable and was actively evaluated rather than dismissed as not applicable.
Neither criterion is contradicted; both fail because the required phase-resolved observation was never made. The case is a variant-level curation with no proband genotype, zygosity, or pedigree, and the seven ClinVar submissions evaluate the variant in isolation without criteria or accompanying-variant detail.
No published proband carries c.1522A>G: the extracted full texts (PMID:28944238, PMID:25394175, PMID:28492532, PMID:26467025) do not mention c.1522A>G or p.Lys508Glu, so no literature-based cis/trans configuration could be established.
These two criteria therefore contribute no evidence in either direction to the final classification; the rare (2) homozygotes in gnomAD v4.1 are unphased population observations and were not used to support or refute either criterion.
Variant and scope: NM_002439.5:c.1522A>G (MSH3), NC_000005.9:g.80024738A>G, GRCh38 chr5:80728919 A>G, protein consequence NP_002430.3:p.(Lys508Glu) - a missense substitution. PP3/BP4 for a missense variant are taken from the REVEL path only; the SpliceAI result (max delta 0.002) is not admissible support for either criterion and codes from the splice path were not substituted.
Governing framework: no ClinGen CSPEC specification was found for MSH3 (cspec found=false) and no gene-specific VCEP dir, materials, or PP3/BP4 lookup spreadsheet exists for this case (gene_vcep_dir=null, framework_complete=false, framework_mode=generic_acmg). Generic ACMG/AMP 2015 (PMID:25741868) with the published ClinGen SVI REVEL calibration (Pejaver et al. 2022, PMID:36413997) therefore governs.
PP3: REVEL 0.191 falls below the supporting threshold (>=0.644), so no pathogenic computational evidence; PP3 not met.
BP4: REVEL 0.191 falls at or below the supporting threshold (<=0.29) but above the moderate threshold (<=0.183), so BP4 is met at supporting strength.
BayesDel (-0.230708) was queried but no generic ACMG-strength calibration exists for it, so it is not_available and contributed no weight in either direction.
The two criteria draw on the same single prediction (REVEL 0.191) in opposite directions by threshold, which is the intended use of a calibrated predictor; no prediction was counted twice as independent evidence, and nothing here overlaps BP7 (synonymous-only), which is out of scope for this missense variant.
No population criterion contributes evidence for NM_002439.5:c.1522A>G (MSH3 p.Lys508Glu). The variant is too common for PM2 (total AF 0.000243-0.000375 versus the 0.01% supporting threshold, African/African American popmax 0.44%) but remains well below the benign frequency thresholds (BS1 1% generic default: popmax 0.44%; BA1 5%: popmax 0.44%), so neither PM2 nor BS1/BA1 is met. BS2 is not applicable because MSH3-associated disease is adult-onset, autosomal recessive with unestablished penetrance, failing the criterion's full-penetrance-at-early-age precondition, and the 2 gnomAD v4.1 homozygotes lack colorectal phenotype information. The net population-frequency effect is therefore neutral.
Source selection: no ClinGen VCEP or gene-specific framework exists for MSH3 (generic ACMG/AMP 2015 fallback), so all-comers GNOMAD_V2_1/GNOMAD_V4_1 totals were used as the default; the supplied non-cancer subsets (v2.1 exomes-only grpmax filtering AF 0.3556%; v3.1 genomes-only grpmax filtering AF 0.3351%) were also evaluated and change none of the four calls.
No criteria match.
Met · 2
BP1SupportingMet at supporting: MSH3 disease is caused by truncating variants - ClinVar MSH3 P/LP records are frameshift/nonsense/splice versus 1170 missense VUS.
Details ▾Hide ▴
Reasoning
BP1 applies to a missense variant in a gene for which primarily truncating variants are known to cause disease. MSH3's established pathogenic spectrum is truncating: ClinVar MSH3 pathogenic/likely pathogenic records are frameshift, nonsense and canonical splice variants, and the gene's disease mechanism is biallelic loss of MSH3 function (MSH3 deficiency with EMAST in MSH3-associated adenomatous polyposis). By contrast, missense variation in MSH3 is not established as causal: in a 1200-record sample of ClinVar MSH3 missense records, 1170 are Uncertain significance and 28 conflicting, with only two singleton P/LP missense assertions. This variant is a missense substitution (p.Lys508Glu) in that gene, so BP1 is met at its default supporting strength. Because a small number of singleton missense P/LP MSH3 assertions exist and MSH3 disease is recessive, BP1 is applied only at supporting strength and must not be escalated; no evidence exists that this particular substitution is a hypomorphic loss-of-function allele.
Rule
BP1: missense variant in a gene for which primarily truncating variants are known to cause disease (supporting strength, generic ACMG/AMP 2015).
Evidence considered
ClinVar MSH3 pathogenic/likely pathogenic records are truncating: e.g. c.1790dup (p.Ser598fs), c.1660dup (p.Met554fs), c.1808C>G (p.Ser603Ter), c.1651C>T (p.Gln551Ter), c.910-1G>A, c.793-1G>A - frameshift, nonsense and canonical splice variants only in the established pathogenic set.
ClinVar MSH3 missense records (1200-record sample): 1170 Uncertain significance, 28 conflicting, only 2 singleton P/LP assertions - missense change is not an established disease mechanism for MSH3.
The variant under assessment is a missense substitution, NM_002439.5:c.1522A>G, NP_002430.3:p.(Lys508Glu).
MSH3 germline disease is a biallelic loss-of-function phenotype (MSH3-associated attenuated adenomatous polyposis with EMAST, the hallmark of MSH3 deficiency); pvs1_gene_context.json records lof_mechanism_supported = true for MSH3.
Generic ACMG/AMP (Richards et al. 2015, PMID:25741868) defines BP1 as a missense variant in a gene where primarily truncating variants cause disease, at supporting strength by default.
Still missing
No functional or case-level data establish p.Lys508Glu as a loss-of-function missense allele; BP1 therefore rests on the gene-level pathogenic variant spectrum and is deliberately kept at supporting strength.
BP4SupportingMet at supporting strength: REVEL 0.191 is within the <=0.29 BP4 supporting threshold but above the <=0.183 moderate cut-off.
Details ▾Hide ▴
Reasoning
Scope settled first: NM_002439.5:c.1522A>G is a missense substitution (NP_002430.3:p.(Lys508Glu)), so BP4 is adjudicated on the missense in-silico path (REVEL) only; the SpliceAI score is not admissible as BP4 support for a missense variant. With no ClinGen CSPEC specification or gene-specific VCEP framework for MSH3, the generic ACMG/AMP 2015 fallback and the published ClinGen SVI REVEL calibration apply. The local REVEL v1.3 score is 0.191, which satisfies the BP4 supporting threshold of <=0.29 (Pejaver et al. 2022, PMID:36413997) but does not reach the moderate (<=0.183) or strong (<=0.016) thresholds, so BP4 is met at supporting strength only. The score is not in the indeterminate gray zone (0.29-0.644). This criterion is not double-counted with any splice-prediction or protein-length evidence, and the same REVEL value is the basis of the PP3 not-met call rather than a second, independent assertion. The supporting-strength call is made directly from the calibrated score alone; no other source's classification was used to reach it.
Path selection: missense consequence (NP_002430.3:p.(Lys508Glu)) selects the REVEL missense path for BP4; SpliceAI max delta 0.002 was not used as BP4 support.
Local REVEL v1.3 lookup returned revel_score 0.191 for chrom 5, pos 80728919 (GRCh38), ref A, alt G; found=true.
REVEL thresholds applied verbatim from the ClinGen SVI calibration (Pejaver et al. 2022, Am J Hum Genet, PMID:36413997): BP4 supporting <=0.29, moderate <=0.183, strong <=0.016.
BayesDel score -0.230708 was not used; no established generic ACMG-strength calibration exists for BayesDel, so it is not_available for BP4.
Governing framework: no ClinGen CSPEC entry for MSH3 and no gene-specific VCEP materials or pre-assigned PP3/BP4 lookup code in this case, so generic ACMG/AMP 2015 applies.
Corroborative context only, not the basis of the call: a ClinVar submission for this variant (GeneDx, SCV002558398, Uncertain significance) states that in silico analysis suggests this missense variant does not alter protein structure/function; that submission carries no explicit PP3/BP4 code and was not used to derive the strength.
PS1Not met: p.Lys508Glu is not established as pathogenic - ClinVar shows 5 Likely benign and 1 VUS submission and no pathogenic assertion.
Details ▾Hide ▴
Reasoning
PS1 requires that a DIFFERENT nucleotide change producing the same amino-acid substitution (p.Lys508Glu) has been established as pathogenic. Neither half of that requirement is satisfied. (1) The reference codon 508 is AAA (Lys); the only single-nucleotide substitution that can encode Glu (GAA/GAG) from AAA is c.1522A>G, i.e. this variant itself, so no independent nucleotide change at this codon can produce p.Lys508Glu. (2) The p.Lys508Glu change is not established as pathogenic: ClinVar VCV000781999 (NM_002439.5(MSH3):c.1522A>G) carries the aggregate classification 'Conflicting classifications of pathogenicity' from 7 submissions - 5 laboratories assert Likely benign, 1 asserts Uncertain significance, there are no expert-panel submissions and no pathogenic/likely pathogenic assertion. With no established pathogenic occurrence of this exact amino-acid change, PS1 cannot be met.
Evidence considered
ClinVar VCV000781999 / Variation ID 781999, NM_002439.5(MSH3):c.1522A>G (p.Lys508Glu): overall germline classification 'Conflicting classifications of pathogenicity'; submissions include Ambry Genetics (Likely benign), PreventionGenetics (Likely benign), Sinai Health System (Likely benign), GeneDx (Uncertain significance), Dasa (Likely benign), Quest Diagnostics (likely benign), Labcorp/Invitae (Likely benign); no expert-panel submission and no P/LP assertion.
Codon 508 of NP_002430.3 is AAA (Lys); deriving Glu (GAA or GAG) requires the c.1522A>G change itself, so no alternate nucleotide change exists that yields p.Lys508Glu - PS1's premise (a different nucleotide change with the same amino-acid change) is structurally impossible here.
PS3Not met: no assay has tested p.(Lys508Glu), and OncoKB reports no variant-specific functional evidence for this change.
Details ▾Hide ▴
Reasoning
PS3 (Richards et al. 2015, PMID:25741868) requires a well-established functional study demonstrating a damaging effect on the gene or gene product, and any strength above Supporting requires that the assay be validated with defined positive and negative controls in the disease-relevant readout. MSH3 c.1522A>G (NP_002430.3:p.(Lys508Glu)) is a missense change in a DNA mismatch repair gene whose product forms the MutS-beta heterodimer with MSH2, so the biologically relevant assay platform would be a variant-level MutS-beta-dependent mismatch-repair or mismatch-binding assay. No such study exists for this variant: variant-specific functional evidence was absent from every retrieved source. OncoKB's K508E entry reports 'Unknown Oncogenic Effect' with no variant-specific reviewed functional evidence and no variant-level citations. The ClinVar record (VCV000781999, 7 submissions, no expert panel) contains no submission that reports a functional criterion, and the two submissions with a validated PMID trail cite only the ACMG framework (25741868), a 36-gene panel-sequencing cohort (28944238) and a Lynch-syndrome modifier cohort (37088804) - none of which assays this variant. All seven full-text papers fetched for this case were examined for a p.Lys508Glu assay (searched terms MSH3, c.1522A>G, p.(K508E), K508E): only PMID:28944238 names MSH3, and only at gene level, using REVEL-based in-silico tiering rather than an assay. An index-level PubMed search for 'MSH3 AND K508E' and 'MSH3 AND 1522A>G' returned zero records, and broader MSH3 functional-assay searches returned only studies unrelated to this residue (22787155, GAA-TTC repeat expansion; 11506498, bladder-cancer MMR activity). Mismatch-repair-deficiency readouts such as MSI or IHC loss are not variant-specific and cannot substitute for a variant-level assay. With no functional result to calibrate, PS3 cannot be met at any strength; no strength is assigned.
Rule
PS3 requires a well-established functional assay showing a damaging effect on MSH3; requirement unsatisfied (no variant-level functional data), strength left null.
Evidence considered
PS3 definition applied: well-established functional study showing a damaging effect on the gene or gene product (Richards et al. 2015, PMID:25741868); no such study was identified for MSH3 c.1522A>G.
OncoKB K508E entry returned 'Unknown Oncogenic Effect' with no variant-specific reviewed functional evidence and no variant-level citations (only gene-level context: MSH3 is a mismatch repair protein frequently mutated in colorectal cancers).
ClinVar VCV000781999 (NM_002439.5(MSH3):c.1522A>G (p.Lys508Glu)): 7 submissions, 0 expert-panel submissions, 0 submissions reporting a functional criterion in their evidence signals; the submissions with validated PMID trails cite 25741868, 28944238 and 37088804, none of which reports an assay of this variant.
All seven full-text papers supplied for this case were checked for a p.Lys508Glu functional result (searched terms included MSH3, c.1522A>G, p.(K508E), p.Lys508Glu, K508E, NP_002430.3:p.(K508E)); only PMID:28944238 mentions MSH3, at gene level only (36-gene targeted sequencing of 1231 colorectal cancer cases, with missense variants tiered by REVEL >=0.5 and counsellor review), and reports no assay of p.Lys508Glu.
Direct PubMed index search (NCBI E-utilities esearch) for 'MSH3 AND K508E' and 'MSH3 AND 1522A>G' returned 0 records; a broader MSH3 functional-assay search returned only studies unrelated to this residue (PMID:22787155, GAA-TTC repeat expansion in FRDA cells; PMID:11506498, mismatch repair activity in bladder cancer lines), neither of which assays p.Lys508Glu.
Biological relevance considered for assay selection: MSH3 encodes the MutS-beta (MSH2-MSH3) mismatch repair heterodimer and acts as a tumour suppressor, so a disease-relevant PS3 assay would need to measure MutS-beta-dependent mismatch repair or mismatch binding for the variant protein with assay controls; no such result is available.
Population-level observations (e.g. 2 homozygotes in gnomAD v4.1) are not functional evidence and cannot support PS3.
Still missing
No variant-specific functional study of MSH3 c.1522A>G p.(Lys508Glu) exists in any retrieved source; a validated in vitro MutS-beta-dependent mismatch-repair (or mismatch-binding) assay with defined positive and negative controls and a disease-relevant readout would be required before PS3 could be met.
Mismatch-repair-deficiency observations (MSI-high status, IHC loss of MMR proteins) are tumour- or patient-level and not variant-specific, so they cannot substitute for a variant-level assay under PS3.
No MSH3 ClinGen VCEP/CSPEC or gene-specific specification was found (generic ACMG/AMP 2015 fallback), so no gene-specific PS3 strength calibration (Supporting/Moderate/Strong) or assay-acceptability criteria are available to grade any future functional result.
PS4Not met: no case-control enrichment exists, as the variant is absent from the 1231-case CRC cohort and present in gnomAD v4.1 at 392/1,612,578 alleles.
Details ▾Hide ▴
Reasoning
Governing framework: no ClinGen CSPEC or gene-specific VCEP specification exists for MSH3 (cspec found=false; vcep_materials.json files_found=[]; final_classification_framework.json framework_complete=false, source=generic_acmg_fallback), so generic ACMG/AMP 2015 (Richards et al., PMID:25741868) governs. Generic PS4 requires that the prevalence of the variant in affected individuals be significantly increased compared with controls. No numeric threshold or comparison operator is defined for PS4 by this governing framework, so there is no >=/<= (or >/<) inequality to evaluate for this criterion; the numeric-direction check is vacuous here and no threshold was invented (screen_PS4.json also records screened=0, passed=0). On the available evidence the criterion is not met: (1) the only case-control-type dataset retrieved for MSH3 is the case/control CRC sequencing study PMID:28944238 (1231 CRC cases across five subgroups - FCCTX n=153, unselected n=548, young onset n=333, pMMR n=68, dMMR n=129 - plus 93 unaffected spouse controls), which does not contain this variant at all: targeted searches of its full text for c.1522A>G, p.(K508E), p.Lys508Glu and K508E returned no match, and its MSH3 findings (raised carrier counts driven by p.Ser490Tyr, p.His827Arg, p.Leu911Trp) concern different residues; (2) no case report or case series naming this exact variant was found in the retrieved literature, so there is no affected-individual count to compare with controls; (3) the population data point the other way - gnomAD v4.1 reports 392/1,612,578 alleles (AF 0.000243), including 329 alleles and 2 homozygotes in the African/African American subpopulation (grpmax FAF 0.003997), i.e. the variant is seen at appreciable, homozygous-observed frequency in unselected population samples of a common cancer, which is not the pattern expected of a variant enriched in CRC cases. A ClinVar submission for the exact variant (GeneDx, SCV002558398) asserts the variant was 'observed in individuals with colorectal cancer in published literature (pmid: 28944238)', but that paper does not mention the variant, so this cannot be used as PS4 support. Direction of the numeric check for this criterion as written by the governing source: no comparison operator is specified, and no threshold applies.
Rule
generic_acmg_2015 PS4: variant prevalence in affected individuals significantly increased versus controls; no numeric threshold or comparison operator defined by the governing framework for MSH3 (no VCEP)
Evidence considered
Framework selection: MSH3 has no ClinGen CSPEC and no gene-specific VCEP specification in this case (cspec found=false; vcep_materials.json note 'no cspec url available', files_found=[]), so generic ACMG/AMP 2015 (PMID:25741868) was applied; PS4 is defined there as increased prevalence of the variant in affected individuals versus controls, with no numeric cut-off and no stated comparison operator - therefore no threshold inequality (>=, <=, >, <) exists to apply for this criterion in this case.
PMID:28944238 is a germline targeted-sequencing study of 36 CRC susceptibility genes in 1231 CRC cases (five subgroups) plus 93 unaffected spouse controls; its own variant-level content concerns MSH3 p.Ser490Tyr, p.His827Arg and p.Leu911Trp, not codon 508. Direct text searches of its full text for 'c.1522A>G', 'p.(K508E)', 'p.Lys508Glu', 'K508E', '1522' and '508Glu' produced no match, confirming the exact variant is not reported in that cohort and that no affected-versus-control count for it exists there.
gnomAD v4.1 for chr5-80728919-A-G (source_registry key gnomad_v4): 392/1,612,578 alleles (AF 0.000243), 2 homozygotes, popmax African/African American AF 0.004388 (329/74,984 alleles, 2 homozygotes), grpmax FAF 0.003997 - unselected-population presence including homozygotes, rather than case enrichment.
ClinVar submission audit (clinvar_submission_audit.json, source_registry key clinvar): 7 submissions, 6 usable, 0 expert-panel submissions, aggregate label 'Conflicting classifications of pathogenicity' (1 star) - a non-expert, non-enrichment source that cannot support PS4. One submission's claim that the variant was observed in CRC cases in PMID:28944238 is not borne out by that paper's text.
screen_PS4.json records screened=0 and passed=0 candidate case-observation records for PS4; literature_pass.json/ pmid_triage.json list PMID:37088804 and PMID:33516529 as PS4 leads, of which neither supplied variant-specific data (37088804 abstract checked directly via PubMed: no MSH3/1522/K508 mention; 33516529 full text elsewhere in this workspace: no MSH3 mention).
Still missing
No proband-level case count for this exact variant exists in the retrieved evidence; a variant-specific case-control or affected-vs-control enrichment study would be required to assess PS4 positively.
PMID:37088804 (Vibert et al. 2023, 'Lynch syndrome: influence of additional susceptibility variants on cancer risk'), cited by ClinVar submission SCV004221042 for this exact variant, was not retrieved as full text in this case. Its title/abstract were checked directly through PubMed and contain no mention of MSH3, c.1522, or K508, so it supplies no PS4 evidence; its full text remained unexamined.
PMID:33451724 (SGO endometrial cancer evidence-based review, part II), listed as a ClinVar-cited PS4/PP5 lead, was retrieved only as a screenshot (no text extraction available in this case), so it could not be searched for the variant.
PMID:33516529 (SGO endometrial cancer review, part I), also listed as a PS4/PP5 lead, was not retrieved in this case; the cached full text available elsewhere in this workspace contains no mention of MSH3, c.1522, or K508.
PM1Not met: residue 508 lies in MSH3 MutS domain II but is no hotspot and is not free of benign variation (gnomAD popmax AF 0.44%, 2 homozygotes).
Details ▾Hide ▴
Reasoning
No governing VCEP specification, CSPEC, or local framework exists for MSH3 (vcep_materials.json reports no CSPEC url and its domain_tables key is empty), so PM1 was assessed by generic ACMG/AMP. PM1 requires a mutational hotspot or a critical, well-established functional domain WITHOUT benign variation. Residue Lys508 is not a hotspot: the variant does not lie in a statistically significant hotspot, it is absent from COSMIC somatic data, and OncoKB assigns 'Unknown Oncogenic Effect' with no variant-specific evidence. Domain-wise, Lys508 falls in MutS domain II (connector region, Pfam PF05188, residues 367-521) of the DNA mismatch repair protein, between MutS domain I (230-341) and MutS domain III (540-838), with the ATPase MutS domain V / Walker-A motif at 892-1094; the domain-table precondition for treating any MSH3 region as a PM1 critical domain is not met, and in any case the 'absence of benign variation' condition fails at this residue - gnomAD v4 shows common population variation (popmax AF 0.44% in African/African American, 2 homozygotes) and gnomAD v2 popmax AF 0.39%, the neighbouring residue change p.Lys508Arg is of conflicting ClinVar significance, and the synonymous c.1524A>G (p.Lys508=) is Likely benign. PM1 is not met.
Evidence considered
gnomAD v4.1: this variant is present (AF 0.000243; 392/1612578 alleles; 2 homozygotes) with popmax AF 0.43876% in African/African American; gnomAD v2.1: AF 0.000375 (106/282466), popmax 0.38750% in African/African American - appreciable benign population variation in this region and at this residue.
Hotspot screen: 'This variant does not lie in a statistically significant hotspot' (no hotspot row found for MSH3 K508); COSMIC: variant not previously reported in somatic cancers.
OncoKB: 'Unknown Oncogenic Effect' with no variant-specific reviewed functional evidence; only gene-level context (MSH3, a DNA mismatch repair protein, frequently mutated in colorectal cancers).
ClinVar codon-508 neighbourhood: NM_002439.5:c.1523A>G (p.Lys508Arg) = 'Conflicting classifications of pathogenicity'; NM_002439.5:c.1524A>G (p.Lys508=) = Likely benign - the residue does not show the pathogenic-variant clustering expected of a hotspot.
Domain mapping (no VCEP domain table available; vcep_materials.json domain_tables is empty): Pfam/InterPro on UniProt P20585 places Lys508 inside MutS domain II (PF05188, 367-521); MutS domain I = 230-341 (PF01624), MutS domain III = 540-838 (PF05192), MutS domain V/ATPase = 892-1094 (PF00488), with the Walker-A binding site at 896-903. MSH3 ATPase/mismatch-binding domains are the functionally critical regions; the connector domain II is not an established critical/hotspot region, and this residue carries benign population variation.
Still missing
No MSH3-specific VCEP/CSPEC or curated domain table was available; domain boundaries in this rationale come from Pfam/InterPro annotation of UniProt P20585 rather than a governing specification.
PM2Not met: total allele frequency 0.024-0.038% in gnomAD v4.1/v2.1 exceeds the 0.01% PM2 rarity threshold, with the African/African American popmax at 0.44%.
Details ▾Hide ▴
Reasoning
PM2 (supporting; applied threshold: allele frequency <= 0.0001, i.e. 0.01%) is not met. The variant is far too common to be treated as rare. In the default all-comers sources it is present at 106/282,466 alleles in gnomAD v2.1 (AF 0.000375 = 0.0375%, approximately 3.8-fold above the threshold) and 392/1,612,578 alleles in gnomAD v4.1 (AF 0.000243 = 0.0243%, approximately 2.4-fold above), with highest subpopulation frequencies of 0.4388% (gnomAD v4.1 African/African American) and 0.3875% (gnomAD v2.1 African/African American) - roughly 40-fold above the threshold. It is also observed in the homozygous state in gnomAD v4.1 (2 homozygotes). The alternative sources do not rescue rarity either: gnomAD v2.1 non-cancer exomes 74/236,548 = 0.000313, gnomAD-Canada v1.0 3/18,420 = 0.000163, and gnomAD v3.1 non-cancer genomes 170/147,900 = 0.001149 (the highest of all sources). For an autosomal recessive gene this pattern is expected, since pathogenic MSH3 alleles persist at appreciable carrier frequency in the general population, so allele rarity cannot be claimed for this variant. No MSH3-specific PM2 rule exists; the generic 0.01% supporting threshold is exceeded on every source examined, including the most permissive one, so PM2 is not met.
Rule
generic_acmg: PM2 supporting when allele frequency <= 0.0001
Evidence considered
gnomAD v4.1 all-comers: total 392/1,612,578 alleles, AF 0.000243 (MAF 0.0243%), 2 homozygotes; exome 220/1,460,304 = 0.000151, genome 172/152,274 = 0.001130; African/African American popmax 329/74,984 = 0.4388%.
gnomAD v2.1 all-comers: total 106/282,466 alleles, AF 0.000375 (MAF 0.0375%), 0 homozygotes; exome 79/251,068 = 0.000315, genome 27/31,398 = 0.000860; African/African American popmax 96/24,774 = 0.3875%.
gnomAD v3.1 non-cancer (genomes only): 170/147,900 = 0.001149 total AF, 0 homozygotes.
gnomAD-Canada v1.0 (HostSeq genomes): 3/18,420 alleles, AF 0.000163, 0 homozygotes.
Threshold applied verbatim as supplied data: PM2 supporting AF <= 0.0001 (SVI recommendation to downgrade PM2 to Supporting; PMID:25741868).
Because no VCEP or gene-specific specification governs MSH3, the all-comers GNOMAD_V2_1/GNOMAD_V4_1 totals were the default source; none of the non-cancer subsets yields a frequency at or below the 0.01% threshold, so the call is robust to source selection.
No paper in the pre-computed full-text extractions mentions this variant or reports a frequency for it, so no publication contributes evidence to this criterion.
Still missing
No MSH3-specific PM2 rule or gene-specific rarity cutoff exists; the generic 0.01% supporting threshold (PMID:25741868) was applied.
The variant is present in the homozygous state in gnomAD v4.1, but no phenotype data are available for those individuals, so this observation cannot be used to argue absence of recessive pathogenicity.
PM3Not met: no pathogenic variant was observed in trans with MSH3 c.1522A>G, and no phase-resolved proband data supports an in-trans configuration.
Details ▾Hide ▴
Reasoning
PM3 (Richards et al. 2015, PMID:25741868) is met only when a variant is detected in trans with a pathogenic variant, evaluated for recessive disorders. MSH3-associated disease is recessive: ClinVar files this exact variant under conditions that include 'Familial adenomatous polyposis 4' and 'MSH3-related disorder', and the gene-level germline context collected for MSH3 supports a loss-of-function germline disease mechanism, so PM3 is structurally applicable here and was assessed rather than dismissed. No trans observation exists for NM_002439.5:c.1522A>G (p.Lys508Glu). The seven ClinVar submissions (six classified likely benign, one uncertain significance; no expert-panel submission) each classify the single variant in isolation, none lists criteria, and none reports a second MSH3 variant, a phase assignment, or any cis/trans configuration. This case contains no affected-proband genotype, zygosity, or pedigree data at all, so there is no proband in whom trans status could have been determined, and no second pathogenic or likely pathogenic MSH3 variant is reported anywhere in the collected sources. None of the extracted full texts (PMID:28944238, a 36-gene targeted CRC panel in which MSH3 appears only at gene level; PMID:25394175, an MMR/CMMRD referral guideline whose recessive gene list is MLH1/MSH2/MSH6/PMS2 and does not include MSH3; PMID:28492532 and PMID:26467025, classification-methodology papers) names this variant or p.Lys508Glu, so no published proband provides a trans partner. The two homozygotes reported in gnomAD v4.1 (AF 0.000243, 392/1,612,578 alleles; AFR 2 homozygotes) reflect same-variant homozygosity in a population database, not a distinct pathogenic allele on the opposite chromosome, and therefore cannot satisfy PM3. Accordingly PM3 is not met: no pathogenic variant in trans with c.1522A>G has been observed. The limiting factor is missing phase/proband data rather than contradicting evidence, which is recorded below as a gap.
Rule
PM3: recessive disorder - variant detected in trans with a pathogenic variant (Richards et al. 2015, PMID:25741868)
Evidence considered
PM3 rule text (Richards et al. 2015, PMID:25741868): the criterion is defined for recessive disorders and requires the variant to be detected in trans with a pathogenic variant, i.e. a second, independently pathogenic allele on the opposite chromosome with confirmed phase.
Inheritance mode is recessive, so PM3 is applicable rather than moot: ClinVar (Variation ID 781999, conditions supplied via the ClinVar API) lists 'Familial adenomatous polyposis 4' and 'MSH3-related disorder', and the gene-level germline disease context assembled for MSH3 (literature-based, loss-of-function mechanism supported) indicates biallelic MSH3 disease. No CSPEC exists for MSH3, so the generic ACMG/AMP 2015 rule was used.
ClinVar submissions for the exact variant (7 total: 5 clinical laboratories likely benign, 1 laboratory likely benign, 1 laboratory uncertain significance; 0 expert-panel submissions): every submission evaluates the variant in isolation, none supplies assertion criteria naming a second MSH3 allele, and none states that the variant was observed in trans with a pathogenic variant. The three validated PMIDs attached to those submissions (25741868, 28944238, 37088804) include no in-trans observation for this variant.
Case-level data are absent: the case is a variant-level curation (input NM_002439.5:c.1522A>G) with no proband, phenopacket, zygosity call, pedigree, or phase-resolved genotype, so there is no individual in whom c.1522A>G could be shown to be in trans with a pathogenic variant.
gnomAD v4.1 (392/1,612,578 alleles, AF 0.000243; 2 homozygotes, both in the African/African American subset at AF 0.0044): same-variant homozygosity at population level is not a distinct pathogenic variant in trans and does not provide PM3 support.
Full-text extractions checked for a trans-partner report: PMID:28944238 (MSH3 only gene-level, three recurrent MSH3 variants named, p.Lys508Glu not mentioned), PMID:25394175 (recessive CMMRD gene list MLH1/MSH2/MSH6/PMS2 excludes MSH3), PMID:28492532 and PMID:26467025 (methodology only, variant not mentioned). No proband carrying this variant, and therefore no phase information, is reported in any of them.
Still missing
No proband-level data of any kind in this case: no affected individual, no pedigree, no zygosity and no phase information, so trans configuration with c.1522A>G could not be examined directly.
No second pathogenic or likely pathogenic MSH3 variant is reported in any consulted source (ClinVar submissions, gnomAD, or the extracted full texts); a targeted search for a reported in-trans MSH3 partner allele would be required to move PM3 off not_met.
No CSPEC/VCEP specification exists for MSH3, so no gene-specific PM3 weighting, recessive-disease scope, or ClinGen SVI PM3 points-based refinement could be applied; generic ACMG/AMP 2015 semantics were used.
PM5Not met: the only other codon-508 missense variant, p.Lys508Arg, has conflicting ClinVar classifications and is not established pathogenic.
Details ▾Hide ▴
Reasoning
PM5 requires a different missense change at the same residue (Lys508) to be established as pathogenic. ClinVar contains only two other codon-508 variants: NM_002439.5:c.1523A>G (p.Lys508Arg, VCV000639736), which carries the aggregate classification 'Conflicting classifications of pathogenicity' and therefore is not established as pathogenic, and the synonymous NM_002439.5:c.1524A>G (p.Lys508=, VCV002785889), which is Likely benign and is not a missense comparator at all. Searches for further codon-508 missense alleles (p.Lys508Asn, p.Lys508Met, p.Lys508Gln) returned no ClinVar records; the only other codon-508 record, c.1524del (p.Lys508fs), is a frameshift, not a missense change. The same-residue comparator search that pm5_candidates.json skipped was completed directly against ClinVar for this assessment, so no pathogenic same-residue comparator exists and PM5 is not met.
Evidence considered
ClinVar VCV000639736, NM_002439.5(MSH3):c.1523A>G (p.Lys508Arg): 'Conflicting classifications of pathogenicity' (criteria provided, conflicting classifications) - not an established pathogenic same-residue comparator.
ClinVar VCV002785889, NM_002439.5(MSH3):c.1524A>G (p.Lys508=): Likely benign - synonymous, not a missense comparator.
ClinVar searches within MSH3 for p.Lys508Asn, p.Lys508Met and p.Lys508Gln returned zero records; the only additional codon-508 record is c.1524del (p.Lys508fs), a frameshift rather than a missense change.
pm5_candidates.json reported found=false and skipped candidate harvesting ('Unable to confirm classic same-residue PM5 semantics safely'); that comparator search was performed directly against ClinVar here, so the gap is closed rather than left open.
PP2Not met: MSH3 shows no missense constraint (gnomAD oe_mis 1.08, mis_z -1.29) and its established pathogenic variants are truncating, not missense.
Details ▾Hide ▴
Reasoning
PP2 has two preconditions: the gene must show a low rate of benign missense variation (missense-constrained), and missense changes must be a common, established mechanism of disease. MSH3 satisfies neither. Gene-level gnomAD constraint shows no depletion of missense variation at all (mis_z = -1.29; observed/expected missense = 1.08, i.e. slightly more missense variation than expected under neutrality) and the gene is loss-of-function tolerant (pLI approximately 0, oe_lof = 0.86), a pattern typical of a recessive biallelic disease gene rather than missense intolerance. Consistent with this, benign/uncertain missense variation is abundant in MSH3: in a 1200-record sample of ClinVar MSH3 missense records, 1170 are Uncertain significance and 28 conflicting, with only two singleton P/LP missense assertions, while the established pathogenic spectrum is truncating (frameshift, nonsense, canonical splice). MSH3-associated disease is a biallelic MSH3-deficiency (loss-of-function) phenotype, so missense substitution is not an established mechanism of disease for this gene. PP2 is not met.
Evidence considered
gnomAD (GRCh38) MSH3 gene constraint: mis_z = -1.29, oe_mis = 1.08 (observed/expected missense > 1, i.e. no depletion of missense variation), pLI approximately 0, oe_lof = 0.86.
ClinVar MSH3 missense records: a 1200-record sample contained 1170 'Uncertain significance', 28 'Conflicting classifications of pathogenicity', 1 Pathogenic and 1 Likely pathogenic (both single-submitter assertions, e.g. p.Gly1044Ser and p.Ser773Cys) - the overwhelming majority of MSH3 missense variation is of uncertain or benign significance.
ClinVar MSH3 pathogenic/likely pathogenic records are truncating (frameshift, nonsense, canonical splice), establishing loss of function, not missense change, as the gene's disease mechanism.
pvs1_gene_context.json: MSH3 germline disease context found from literature with lof_mechanism_supported = true ('disease-focused publications supporting MSH3 loss of function as a germline disease mechanism'), consistent with a recessive loss-of-function mechanism rather than a missense mechanism.
PP3Not met: REVEL 0.191 for this missense variant is below the >=0.644 PP3 supporting threshold.
Details ▾Hide ▴
Reasoning
Scope settled first: NM_002439.5:c.1522A>G in MSH3 is a missense substitution, predicted to encode NP_002430.3:p.(Lys508Glu), so it falls on the missense in-silico path (REVEL) and the splice path is not used for this criterion. No ClinGen CSPEC specification and no gene-specific VCEP framework exist for MSH3 (cspec found=false; framework_complete=false; framework_mode=generic_acmg), so the generic ACMG/AMP 2015 fallback with the published ClinGen SVI REVEL calibration applies. The local REVEL v1.3 lookup returned a score of 0.191 for the variant's GRCh38 coordinates (chr5:80728919 A>G, chromosome 5, ref A, alt G), which is below the PP3 supporting threshold of >=0.644 and therefore also below the moderate (>=0.773) and strong (>=0.932) thresholds (Pejaver et al. 2022, PMID:36413997). The BayesDel score in the evidence (-0.230708) was not used because no generic ACMG-strength calibration for BayesDel is established; the SpliceAI max delta (0.002) was not used because splice prediction is not the applicable path for a missense variant. PP3 is therefore not met.
Path selection: missense consequence (NP_002430.3:p.(Lys508Glu)) selects the REVEL missense path for PP3/BP4; the SpliceAI path was not used for either criterion.
Local REVEL v1.3 lookup returned revel_score 0.191 for chrom 5, pos 80728919 (GRCh38), ref A, alt G, matching the case's normalized GRCh38 coordinates; found=true.
REVEL thresholds applied verbatim from the ClinGen SVI calibration (Pejaver et al. 2022, Am J Hum Genet, PMID:36413997): PP3 supporting >=0.644, moderate >=0.773, strong >=0.932.
BayesDel score -0.230708 present but not used: no generic ACMG-strength PP3/BP4 calibration exists for BayesDel.
Governing framework: no ClinGen CSPEC entry for MSH3 and no gene-specific VCEP materials or PP3/BP4 lookup table in this case (vcep_materials.json found=false, framework_complete=false), so generic ACMG/AMP 2015 applies.
SpliceAI lookup for this variant reported max delta score 0.002 (DS_AG 0.002, all other deltas 0.0); this was deliberately not carried into PP3/BP4 because the variant is missense.
BA1Not met: the highest frequency at usable sample size is 0.44% (gnomAD v4.1 African/African American), roughly tenfold below the 5% stand-alone benign threshold.
Details ▾Hide ▴
Reasoning
BA1 (stand-alone benign; applied threshold: highest observed population frequency >= 5%) is not met. MSH3 has no ClinGen VCEP and no gene-specific framework was available for this case (framework fallback = generic ACMG/AMP 2015), so the all-comers gnomAD v2.1 and v4.1 totals are the default source; the non-cancer subsets were also inspected because they were supplied. The highest frequency in any ancestry group with a usable sample size is the African/African American subset: gnomAD v4.1 329/74,984 alleles = 0.4388% (joint grpmax filtering AF 0.3997%) and gnomAD v2.1 96/24,774 = 0.3875% (grpmax FAF 0.3481%). Both are roughly an order of magnitude below 5%. The alternative sources are no higher: gnomAD v2.1 non-cancer exomes grpmax filtering AF 0.3556%, gnomAD v3.1 non-cancer genomes grpmax filtering AF 0.3351%, gnomAD-Canada v1.0 overall AF 0.0163%. The only subpopulation frequencies above 1% are single-allele calls in very small reference panels (HGDP:Yoruba 1/38 = 2.63%; 1000G ACB_XY 2/90 = 2.22%; 1000G ESN_XY 2/112 = 1.79%; HGDP:Yoruba_XX 1/16 = 6.25%), with allele numbers as low as 16-38; these are small-sample ascertainment artifacts rather than evidence of a common benign allele and were not used as the popmax for threshold comparison. The variant is also present in the homozygous state in gnomAD v4.1 (2 individuals, African/African American), which is a homozygote observation to be weighed separately rather than a stand-alone benign frequency argument. BA1 is therefore not met.
Rule
generic_acmg: BA1 stand-alone when highest observed population frequency (popmax / grpmax filtering AF) >= 0.05
Evidence considered
gnomAD v4.1 all-comers: 392/1,612,578 alleles, AF 0.000243 (MAF 0.0243%), 2 homozygotes; African/African American popmax 329/74,984 = 0.4388%; joint grpmax filtering AF 0.3997%; sex-split African/African American afr_xx 193/41,458 = 0.4655%.
gnomAD v2.1 non-cancer (exomes only): 74/236,548 = 0.000313 exome AF, 0 homozygotes, grpmax filtering AF 0.3556% (African/African American); genome component of that subset is the unfiltered 27/31,398.
gnomAD v3.1 non-cancer (genomes only, 170/147,900 = 0.1149% total AF, 0 homozygotes, grpmax filtering AF 0.3351%): its highest raw subpopulation frequencies come from tiny 1000 Genomes/HGDP panels (HGDP:Yoruba 1/38, 1000G ACB_XY 2/90), not from a population cohort of adequate size.
gnomAD-Canada v1.0 (HostSeq genomes): 3/18,420 alleles, AF 0.000163, 0 homozygotes.
Threshold applied verbatim as supplied data: BA1 stand-alone AF >= 0.05 (ACMG/AMP 2015, PMID:25741868).
Population source choice: because no VCEP or gene-specific specification governs MSH3, the all-comers GNOMAD_V2_1/GNOMAD_V4_1 entries were used as the default; the non-cancer subsets were checked as an alternative and give a lower or equal frequency, so the BA1 call is unchanged by source selection.
No paper in the pre-computed full-text extractions reports this variant or any variant-specific population frequency (checked PMID:24493721, PMID:28492532, PMID:26467025, PMID:25394175, PMID:24929052, PMID:28944238, PMID:25741868), so no publication contributes variant-level frequency evidence to this criterion.
Still missing
No MSH3-specific or VCEP-defined BA1 cutoff exists; the generic 5% stand-alone default (PMID:25741868) was applied.
MSH3-associated disease prevalence and penetrance are not established, so a prevalence-derived BA1 could not be computed.
BS1Not met: the highest frequency at usable sample size is 0.44% (gnomAD v4.1 African/African American), about 2.3-fold below the generic 1% BS1 threshold.
Details ▾Hide ▴
Reasoning
BS1 (strong benign; applied threshold: highest observed population frequency >= 1%) is not met. MSH3 has no VCEP or gene-specific specification, so the generic ClinGen-SVI default of 1% (a per-gene prevalence-derived cutoff is preferred but none exists for MSH3, Whiffin et al. 2017, PMID:28518168) was applied to the all-comers gnomAD v2.1/v4.1 totals. The highest frequency at a usable sample size is again the African/African American subset: gnomAD v4.1 329/74,984 = 0.4388% (joint grpmax filtering AF 0.3997%) and gnomAD v2.1 96/24,774 = 0.3875% (grpmax FAF 0.3481%) - about 2 to 2.6-fold below 1%, with every other sizable ancestry group below 0.06% (gnomAD v4.1: Admixed American 0.0300%, remaining 0.0576%; gnomAD v2.1: Admixed American 0.0141%, Non-Finnish European 0.0025%). The non-cancer subsets agree: gnomAD v2.1 non-cancer exomes 74/236,548 = 0.0313% overall with grpmax filtering AF 0.3556%, and gnomAD v3.1 non-cancer genomes 170/147,900 = 0.1149% with grpmax filtering AF 0.3351%. The allele is heterozygous in 390 gnomAD v4.1 alleles and homozygous in 2 African/African American individuals (gnomAD v4.1); gnomAD v2.1, the non-cancer subsets and gnomAD-Canada v1.0 report no homozygotes. No ancestry group of adequate sample size reaches 1%, and the frequencies exceeding 1% in gnomAD v3.1 non-cancer come only from single-allele calls in 16-112 allele reference panels (HGDP:Yoruba 1/38, 1000G ACB_XY 2/90), which are sampling artifacts and not a defensible popmax. BS1 is therefore not met, although the 0.44% popmax is within the same order of magnitude as the generic 1% cutoff.
Rule
generic_acmg: BS1 strong when highest observed population frequency (popmax / grpmax filtering AF) >= 0.01
Evidence considered
gnomAD v4.1 all-comers highest subpopulation frequency: African/African American 329/74,984 = 0.4388% (joint grpmax filtering AF 0.3997%); next highest sizable groups Admixed American 18/60,026 = 0.0300% and remaining 36/62,452 = 0.0576%.
gnomAD v2.1 all-comers highest subpopulation frequency: African/African American 96/24,774 = 0.3875% (grpmax FAF 0.3481%); Admixed American 5/35,414 = 0.0141%; Non-Finnish European 3/118,074 = 0.0025%.
gnomAD v2.1 non-cancer (exomes only): 74/236,548 = 0.000313 (0.0313%) overall, 0 homozygotes, grpmax filtering AF 0.3556% in the African/African American subset.
gnomAD v3.1 non-cancer (genomes only): 170/147,900 = 0.001149 (0.1149%) overall, 0 homozygotes, grpmax filtering AF 0.3351%; individual 1000 Genomes/HGDP subcohort values reach 1-2.6% but on allele numbers of 16-112, so they were not used for the threshold comparison.
Threshold applied verbatim as supplied data: BS1 strong AF >= 0.01 (generic SVI default; per-gene derivation preferred, Whiffin et al. 2017, PMID:28518168).
Homozygote data: 2 homozygotes in gnomAD v4.1 African/African American exomes (1 afr_xx, 1 afr_xy); 0 in gnomAD v2.1, 0 in both non-cancer subsets and 0 in gnomAD-Canada v1.0.
No paper in the pre-computed full-text extractions reports this variant or a variant-specific population frequency, so no publication contributes frequency evidence to this criterion.
Still missing
No MSH3-specific BS1 cutoff and no per-gene prevalence derivation exist (the preferred Whiffin et al. 2017, PMID:28518168, approach requires a disease prevalence estimate that is not available for MSH3-associated disease); the generic 1% default was applied.
The variant sits at 0.44% popmax, below but not far below 1%, so a disease-model-adjusted (autosomal recessive) cutoff would be the deciding input and cannot be computed from the available evidence.
BS3Not met: no functional assay of p.(Lys508Glu) exists, so preserved MSH3 function has never been experimentally demonstrated.
Details ▾Hide ▴
Reasoning
BS3 (Richards et al. 2015, PMID:25741868) requires a well-established functional study demonstrating no damaging effect on protein function, for example a validated variant-level mismatch-repair or mismatch-binding assay in which MSH3 p.(Lys508Glu) behaves like wild-type with adequate assay controls. No such study exists for this variant. OncoKB reports no variant-specific reviewed functional evidence for K508E, no ClinVar submission for VCV000781999 reports a functional criterion or a benign functional result, and none of the seven full-text papers examined contains any MSH3 assay data - the only paper naming MSH3 (PMID:28944238) treats it at gene level in a panel-sequencing cohort and classifies variants by in-silico prediction, not experimentally. Searches of the PubMed index for the variant itself returned zero records, and the MSH3 functional-assay literature retrieved concerns other residues and other phenotypes (repeat expansion, bladder cancer MMR activity). Because the absence of a functional assay is not the same as a functional assay showing preserved function, BS3 cannot be met at any strength and no benign functional evidence is contributed to the classification.
Rule
BS3 requires a well-established functional assay showing no damaging effect on MSH3 function; requirement unsatisfied (no variant-level functional data), strength left null.
Evidence considered
BS3 definition applied: well-established functional study showing no damaging effect on protein function (Richards et al. 2015, PMID:25741868); no such study was identified for MSH3 c.1522A>G.
OncoKB K508E entry: 'Unknown Oncogenic Effect', no variant-specific reviewed functional evidence, no variant-level citations - i.e. no functional result indicating preserved MSH3 function.
ClinVar VCV000781999: 7 submissions (5 benign-leaning, 1 uncertain, 1 no classification), 0 expert-panel submissions, no submission reporting a functional criterion or a benign functional assay; benign-leaning submissions rest on population/case reasoning, not on function.
All seven full-text papers supplied for this case were checked for variant-level functional results (searched terms: MSH3, c.1522A>G, p.(K508E), p.Lys508Glu, K508E); six do not mention MSH3 at all, and PMID:28944238 covers MSH3 only at gene level with in-silico (REVEL) tiering - no functional assay for p.Lys508Glu is reported anywhere.
Direct PubMed index search (NCBI E-utilities esearch) for 'MSH3 AND K508E' and 'MSH3 AND 1522A>G' returned 0 records; broader MSH3 functional-assay searches returned only unrelated studies (PMID:22787155 repeat expansion; PMID:11506498 bladder-cancer MMR activity) that do not assay p.Lys508Glu.
Biological relevance considered: because MSH3 functions as the MSH2-MSH3 (MutS-beta) mismatch repair heterodimer, a benign-function (BS3) claim would require a variant-level MutS-beta repair/mismatch-binding assay with positive and negative controls; none is available.
Boundary noted and not used here: population observations such as homozygotes in gnomAD v4.1 belong to population-frequency criteria and are not functional evidence for BS3.
Still missing
No variant-level functional study of MSH3 c.1522A>G p.(Lys508Glu) was identified, so there is no assay demonstrating retained or wild-type-like MSH3 function; a validated MutS-beta mismatch-repair/mismatch-binding assay with established controls showing normal activity would be required for BS3.
Mismatch-repair-deficiency status (MSI, IHC) is not variant-specific and, when normal, would not by itself establish normal function of this allele under BS3.
No MSH3-specific VCEP/CSPEC exists (generic ACMG/AMP 2015 fallback), so BS3 strength calibration and assay-acceptability standards for this gene are not defined.
BP2Not met: neither a cis nor a trans pathogenic variant was observed with MSH3 c.1522A>G, and no phase-resolved partner allele is reported anywhere.
Details ▾Hide ▴
Reasoning
BP2 (Richards et al. 2015, PMID:25741868, as refined by the Sherloc criteria refinements in PMID:28492532) is met when the variant is observed in trans with a pathogenic variant for a fully penetrant dominant gene/disorder, or observed in cis with a pathogenic variant under any inheritance pattern. Both arms require a second, independently pathogenic variant whose phase relative to c.1522A>G is known. No such partner variant is documented. The seven ClinVar submissions for this exact variant (six likely benign, one uncertain significance, no expert panel) describe the variant alone and record no criteria, no phase statement, and no accompanying variant; the case itself carries no proband genotype, zygosity, or pedigree data; and none of the extracted full texts (PMID:28944238, PMID:25394175, PMID:28492532, PMID:26467025) names c.1522A>G or p.Lys508Glu, so no published individual places it in cis or in trans with a pathogenic allele. The only allele-level observations available are population frequencies (gnomAD v2.1 AF 0.000375; gnomAD v4.1 AF 0.000243 with 2 homozygotes), which are unphased and irrelevant to a cis/trans determination. Because c.1522A>G is the only variant under consideration, neither a cis nor a trans configuration with a pathogenic partner can be established, and BP2 is not met. This reflects the absence of the phase-resolved observation BP2 requires, not positive evidence that the variant co-occurs with a pathogenic allele.
Rule
BP2: observed in trans with a pathogenic variant for a fully penetrant dominant gene/disorder, or in cis with a pathogenic variant in any inheritance pattern (Richards et al. 2015, PMID:25741868)
Evidence considered
BP2 rule text (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 - both clauses require a second pathogenic variant with known phase relative to the variant under assessment.
ClinVar submissions for the exact variant (Variation ID 781999; 7 submissions, 6 likely benign and 1 uncertain significance, no expert panel, review status 'criteria provided, conflicting classifications'): each submission classifies c.1522A>G alone, none provides criteria text or an accompanying MSH3 variant, and none reports a cis or trans configuration.
Case-level data are absent: the case is a variant-level curation with no proband, zygosity, pedigree, or phasing information, so there is no individual in whom a cis or trans partner allele could have been observed.
Full-text extractions checked for a cis/trans observation: PMID:28944238 (targeted 36-gene CRC panel; MSH3 handled at gene level only, p.Lys508Glu never named), PMID:25394175 (MMR/CMMRD referral guideline; MSH3 not among the listed recessive MMR genes), PMID:28492532 (Sherloc criteria refinement; MSH3 and the variant not mentioned). None offers a phase-resolved observation involving this variant.
Unphased population observations only: gnomAD v2.1 (106/282,466 alleles, AF 0.000375, 0 homozygotes) and gnomAD v4.1 (392/1,612,578 alleles, AF 0.000243, 2 homozygotes). Population allele counts carry no phase information about a pathogenic partner allele and cannot satisfy BP2.
Still missing
No phase-resolved genotype is available for this variant in any source; the cis/trans configuration relative to a pathogenic MSH3 allele has never been determined, so BP2 rests on absence of the required observation.
MSH3 has no CSPEC/VCEP specification, so the gene-specific scope of the cis/trans clauses (including whether the dominant-gene clause can apply at all to this recessive disease gene) could not be confirmed against an authoritative framework; generic ACMG/AMP 2015 was used.
ClinVar submissions provide no criteria or accompanying-variant detail for this variant, so a submission-level cis/trans statement could not be retrieved even from records that are otherwise usable for classification.
PS2Not assessed: no parental-testing data exist, so the confirmed de novo PS2 requires (PMID:25741868) is undeterminable; gnomAD v2.1 shows 106 alleles and v4.1 two homozygotes.
Details ▾Hide ▴
Reasoning
PS2 (Richards et al. 2015, PMID:25741868) requires a de novo occurrence of the variant with BOTH maternity and paternity confirmed in a patient with the disease and no family history. No MSH3 ClinGen CSPEC/VCEP or local gene framework exists for this gene (case_summary cspec.found = false; vcep_materials.json: 'no cspec url available'; final_classification_framework.source = generic_acmg_fallback; framework_mode = generic_acmg), so the generic ACMG/AMP 2015 definition governs. This case is a variant-level annotation of NM_002439.5:c.1522A>G (NP_002430.3:p.(Lys508Glu)) with no proband-level clinical record, no pedigree and no parental genotype data in any source: review_manifest.json, prefetch.json, case_summary.json and evidence.json contain only variant normalisation and database lookups (gnomAD, ClinVar, OncoKB, COSMIC, CancerHotspots, SpliceAI, REVEL, BayesDel) plus literature triage, and the MSH3 gene-context record reports inheritance = [] (no disease inheritance model stated). All seven retrieved full-text papers were searched for c.1522A>G, p.(K508E), Lys508Glu and MSH3 and none reports this variant (paper_extracts.json: variant_mentioned = false in every entry), so no de novo occurrence is documented in the literature. The ClinVar record VCV000781999 (7 submissions: GeneDx, Ambry, PreventionGenetics, Sinai Health System, Rogan Lab, Quest Diagnostics, Labcorp) records origin only as germline, unknown, or not applicable and never reports a de novo origin. Population data also argue against a confirmed-de-novo interpretation: the variant is present in gnomAD v2.1 in 106/282466 alleles (96 of them in the African/African American population) and in gnomAD v4.1 in 392/1612578 alleles including 2 homozygotes. Because the trio/parental testing PS2 requires was neither performed nor documented, PS2 cannot be confirmed or excluded and is returned not_assessed rather than scored.
Rule
ACMG/AMP 2015 PS2 (PMID:25741868): de novo (maternity and paternity confirmed) in a patient with the disease and no family history; no MSH3 VCEP/CSPEC, generic fallback applied
Evidence considered
No MSH3-specific ClinGen CSPEC or VCEP specification exists (case_summary.json cspec.found = false; vcep_materials.json 'no cspec url available'), so the generic ACMG/AMP 2015 PS2 definition was applied.
The case contains no proband-level information: review_manifest.json, prefetch.json, case_summary.json and evidence.json hold only variant normalisation plus gnomAD/ClinVar/OncoKB/COSMIC/CancerHotspots/SpliceAI/REVEL/BayesDel lookups and literature triage - no phenotype, family history, pedigree, parental testing, or inheritance-model field.
MSH3 gene-context record (pvs1_gene_context) lists inheritance = [], i.e. no disease inheritance model is stated for this gene, and its supporting literature is gene-level only.
All seven retrieved full-text papers were searched for 'c.1522A>G', 'p.(K508E)', 'Lys508Glu', 'K508E', 'NP_002430.3:p.(K508E)' and 'MSH3'; none reports this variant (paper_extracts.json, variant_mentioned = false in each), so no published de novo occurrence exists.
ClinVar VCV000781999 (7 submissions, 1-star, conflicting classifications, no expert panel) lists origin only as germline, unknown, or not applicable; no submission reports a de novo origin.
Population observation weighs against a de novo pathogenic allele: gnomAD v2.1 106/282466 alleles (AF 0.000375; 96 in African/African American) and gnomAD v4.1 392/1612578 alleles (AF 0.000243) with 2 homozygotes.
PS2 was evaluated against the published ACMG/AMP 2015 definition (PMID:25741868); no other threshold or calibration applies to this criterion.
Still missing
No proband disease status, no pedigree and no trio/parental genotype data are present in this case, so PS2 can neither be met nor excluded.
PMID:37088804 (Lynch syndrome: influence of additional susceptibility variants on cancer risk), cited for this variant by a ClinVar submitter, was never retrieved in full text, so any de novo or family data it contains is unverified.
PMID:33451724 exists in the publications directory only as a page image with no text extract, so it could not be searched for this variant.
PM6Not assessed: no parental testing or de novo observation is documented, so PM6's assumed-de-novo premise (PMID:25741868) cannot be evaluated; the variant recurs in gnomAD (106 v2.1 alleles).
Details ▾Hide ▴
Reasoning
PM6 (Richards et al. 2015, PMID:25741868) requires an assumed de novo occurrence - an observed de novo in a patient with the disease but WITHOUT confirmation of paternity and maternity. No MSH3 ClinGen CSPEC/VCEP or local gene framework exists (case_summary cspec.found = false; vcep_materials.json 'no cspec url available'; generic_acmg fallback), so the generic ACMG/AMP 2015 definition governs. Even at this weaker standard PM6 cannot be evaluated: this case is a variant-level annotation of NM_002439.5:c.1522A>G (p.(Lys508Glu)) with no proband, no parental testing status and no family data anywhere in review_manifest.json, prefetch.json, case_summary.json or evidence.json, and no retrieved paper reports the variant at all (paper_extracts.json: variant_mentioned = false in every one of the seven full texts). Because PM6 still requires an actual de novo observation, the complete absence of proband/parent information means the criterion is not assessable; indirect arguments (the variant recurs in population controls - gnomAD v2.1 106 alleles, v4.1 392 alleles with 2 homozygotes) are context against a de novo pathogenic allele but are not a substitute for an observed de novo event, and they are not used to call PM6 in either direction.
Rule
ACMG/AMP 2015 PM6 (PMID:25741868): assumed de novo, without confirmation of paternity and maternity; no MSH3 VCEP/CSPEC, generic fallback applied
Evidence considered
No MSH3 CSPEC/VCEP or local framework exists (case_summary.json cspec.found = false; vcep_materials.json 'no cspec url available'), so generic ACMG/AMP 2015 PM6 was applied.
No proband, parental-testing status, or family data is recorded in any case file; review_manifest.json, prefetch.json, case_summary.json and evidence.json contain only variant normalisation and database lookups.
All seven retrieved full texts were searched for this variant's notations and for MSH3; none reports the variant or a de novo occurrence (paper_extracts.json).
ClinVar VCV000781999 lists 7 submissions with origin germline, unknown, or not applicable; none reports a de novo origin and no expert-panel assertion exists.
The only indirect de novo-related dataset is population frequency: gnomAD v2.1 106/282466 alleles (96 in African/African American) and gnomAD v4.1 392/1612578 alleles with 2 homozygotes, indicating the allele is inherited in the general population.
PM6 was evaluated against the published ACMG/AMP 2015 definition (PMID:25741868); no other threshold applies.
Still missing
No proband-level record and no statement of whether parents were tested, so PM6's 'assumed de novo' premise cannot be evaluated.
PMID:37088804, cited for this variant by a ClinVar submitter, was not retrieved in full text and could not be checked for de novo or family data.
PMID:33451724 has no text extract (page image only) and could not be searched for this variant.
PP1Not assessed: no pedigree or relative genotypes exist and no paper reports this variant in a family, so PP1 co-segregation (PMID:25741868) cannot be evaluated.
Details ▾Hide ▴
Reasoning
PP1 (Richards et al. 2015, PMID:25741868) requires co-segregation of the variant with disease in multiple affected family members in a gene definitively known to cause the disease (Sherloc refinement, PMID:28492532, likewise conditions this criterion on family-based segregation evidence). No MSH3 ClinGen CSPEC/VCEP or local gene framework exists (case_summary cspec.found = false; vcep_materials.json 'no cspec url available'), so the generic ACMG/AMP 2015 definition governs; MSH3 is also not established as a definitively disease-causing gene for a dominant condition - the gene-level context record states inheritance = [] and the strongest MSH3 literature is for biallelic/attenuated adenomatous polyposis. No segregation evidence exists to evaluate in this case: no pedigree, family members or genotypes of relatives are recorded anywhere in review_manifest.json, prefetch.json, case_summary.json or evidence.json, and none of the seven retrieved full-text papers reports NM_002439.5:c.1522A>G (p.(Lys508Glu)) in any family (paper_extracts.json: variant_mentioned = false throughout). Without any family carrying the variant and any segregation count or LOD score, PP1 is not assessable rather than met or refuted.
Rule
ACMG/AMP 2015 PP1 (PMID:25741868): co-segregation with disease in multiple affected family members; no MSH3 VCEP/CSPEC, generic fallback applied
Evidence considered
No MSH3 CSPEC/VCEP or local framework exists (case_summary.json cspec.found = false; vcep_materials.json 'no cspec url available'), so generic ACMG/AMP 2015 PP1 was applied.
The case files contain no pedigree, no family members, no relative genotypes and no segregation statistic: review_manifest.json, prefetch.json, case_summary.json and evidence.json hold only variant normalisation and database lookups.
PMID:28944238 (36-gene colorectal cancer panel; 1231 cases) was searched for this variant and for family/segregation data: c.1522A>G and p.K508E are absent from the text and tables, and MSH3 appears only at gene level (12 predicted-damaging Tier 2 variants, driven by p.Ser490Tyr, p.His827Arg, p.Leu911Trp) with no segregation analysis for p.K508E.
PMID:26467025 and PMID:28492532 were reviewed for how family segregation is weighted generically (segregation/LOD is one of five weighted evidence types in the former; the latter refines the segregation criteria) but neither reports MSH3 family data.
The three clinical guideline papers retrieved (PMID:24493721, PMID:24929052, PMID:25394175) discuss cancer family history and Lynch-syndrome referral in general terms and name MLH1/MSH2/MSH6/PMS2/EPCAM, not MSH3, so they provide no segregation evidence for this variant.
ClinVar VCV000781999 (7 submissions, conflicting classifications, no expert panel) records no family segregation data and no provenance for the variant in relatives.
PP1 was evaluated against the published ACMG/AMP 2015 definition (PMID:25741868); no other threshold applies.
Still missing
No pedigree, no genotypes from affected or unaffected relatives, and no segregation count or LOD score are available, so PP1 cannot be evaluated.
MSH3 has no ClinGen CSPEC/VCEP and no stated inheritance model (pvs1_gene_context inheritance = []), so the disease context in which co-segregation would be judged is itself unspecified.
PMID:37088804 (Lynch syndrome: influence of additional susceptibility variants on cancer risk), cited for this variant by a ClinVar submitter, was not retrieved in full text and could not be checked for family data.
PMID:33451724 has no text extract (page image only) and could not be searched for this variant.
PP4Not assessed: no proband phenotype or family history was recorded for this case, so PP4's requirement for a highly specific single-gene disease phenotype cannot be tested.
Details ▾Hide ▴
Reasoning
Framework: no ClinGen CSPEC or gene-specific VCEP specification exists for MSH3, so generic ACMG/AMP 2015 (PMID:25741868) governs. Generic PP4 requires that the patient's phenotype or family history be highly specific for a disease with a single genetic etiology; it is a patient-level criterion and its only input is the proband's clinical phenotype/family history. No proband phenotype, HPO term set, family history, age at diagnosis, tumour MMR/MSI status or clinical indication is recorded anywhere in the evidence assembled for this case (no phenotype field is present in the case files, and none of the retrieved sources reports a patient carrying this variant), so the criterion's input is absent and PP4 cannot be evaluated: status not_assessed rather than not_met. PP4 carries no numeric threshold or comparison operator in the governing framework, so no >=/<= inequality applies to it. For completeness, the ClinVar condition labels attached to VCV000781999 (cervical cancer, endometrial carcinoma, familial adenomatous polyposis 4, hereditary cancer-predisposing syndrome, MSH3-related disorder, 'not provided') are submitter-entered, non-specific and, in several submissions, empty; they are not a proband phenotype and do not establish a phenotype 'highly specific for a disease with a single genetic etiology'. No MSH3 phenotype-specificity framework exists to substitute. The one PP4-flavoured signal in the case files is a non-expert ClinVar submission (GeneDx SCV002558398) that applied PP4 citing colorectal cancer in the literature and REVEL-based in-silico reasoning; that is a laboratory assertion about other individuals, not this patient's phenotype, and is therefore not usable.
Rule
generic_acmg_2015 PP4: patient phenotype or family history highly specific for a disease with a single genetic etiology; requires proband phenotype input (absent in this case)
Evidence considered
Framework selection: no CSPEC/VCEP for MSH3 (cspec found=false; vcep_materials.json files_found=[]), so generic ACMG/AMP 2015 (PMID:25741868) was applied; PP4 there is defined as the patient's phenotype or family history being highly specific for a disease with a single genetic etiology, with no numeric threshold or comparison operator (no inequality to apply).
Case files contain no phenotype information for the tested individual: no HPO terms, no clinical indication, no family history and no proband-level record are present in the case evidence, and no retrieved source reports a patient carrying c.1522A>G/p.(K508E).
ClinVar VCV000781999 (clinvar): condition labels across submissions are 'Cervical cancer', 'Endometrial carcinoma', 'Familial adenomatous polyposis 4', 'Hereditary cancer-predisposing syndrome', 'MSH3-related disorder' and 'not provided'; several submissions list no condition at all. These are submitter-entered labels, not a documented proband phenotype, and are too broad to satisfy a 'highly specific' phenotype requirement.
ClinVar submission SCV002558398 (GeneDx) is the only record in this case referencing PP4; it cites 'observed in individuals with colorectal cancer in published literature (pmid: 28944238)' - that paper does not mention this variant - so the signal is a non-expert laboratory assertion, not patient phenotype evidence.
No MSH3 gene-specific phenotype/specificity framework was available in this case (vcep_materials.json, final_classification_framework.json framework_complete=false).
Still missing
Need the proband's clinical phenotype and family history (cancer type(s), ages at diagnosis, tumour MMR/MSI or IHC status, Amsterdam/Bethesda-type criteria, multi-generational family history) to determine whether the presentation is highly specific for a single-gene MSH3-associated disorder.
Need to establish whether MSH3 has an accepted, phenotype-specific disease association (e.g. a biallelic/CMMRD-type presentation) rather than a low-penetrance modifier role; no gene-specific framework for MSH3 was available in this case.
PMID:33451724 (retrieved as screenshot only) and PMID:33516529 (not retrieved in this case) were ClinVar-cited PP4/PS4 leads that could not be searched for variant- or phenotype-level content.
BS4Not assessed: no pedigree or relative genotypes exist to demonstrate non-segregation, and absence of family data is not BS4 evidence (PMID:25741868).
Details ▾Hide ▴
Reasoning
BS4 (Richards et al. 2015, PMID:25741868) is benign strong-supporting evidence defined as lack of segregation in affected members of a family - it requires a family in which affected individuals do not carry the variant (and, in the usual formulation, an appropriate number of informative meioses). No MSH3 ClinGen CSPEC/VCEP or local gene framework exists (case_summary cspec.found = false; vcep_materials.json 'no cspec url available'), so the generic ACMG/AMP 2015 definition governs. This case contains no family data of any kind: no pedigree, no affected or unaffected relative genotypes and no meioses are recorded in review_manifest.json, prefetch.json, case_summary.json or evidence.json, and no retrieved full text reports NM_002439.5:c.1522A>G (p.(Lys508Glu)) in any family (paper_extracts.json: variant_mentioned = false throughout). Absence of a pedigree is not itself non-segregation evidence, so BS4 is not met; because the data type required for the criterion was never generated, it is returned not_assessed rather than asserted as benign evidence.
Rule
ACMG/AMP 2015 BS4 (PMID:25741868): lack of segregation in affected members of a family; no MSH3 VCEP/CSPEC, generic fallback applied
Evidence considered
No MSH3 CSPEC/VCEP or local framework exists (case_summary.json cspec.found = false; vcep_materials.json 'no cspec url available'), so generic ACMG/AMP 2015 BS4 was applied.
No family data of any kind is present: review_manifest.json, prefetch.json, case_summary.json and evidence.json contain only variant normalisation and database lookups, with no pedigree, relative genotypes, or meioses.
All seven retrieved full texts were searched for this variant (c.1522A>G, p.(K508E), Lys508Glu, K508E, MSH3); none reports the variant in any family or reports segregation outcomes for it (paper_extracts.json).
PMID:28944238, the only retrieved cohort study of MSH3 in colorectal cancer cases, reports MSH3 only at gene level and provides no segregation or non-segregation data for p.(K508E).
ClinVar VCV000781999 (7 submissions, conflicting classifications, no expert panel) provides no family segregation or non-segregation evidence; its origins (germline, unknown, not applicable) do not indicate non-segregation.
BS4 was evaluated against the published ACMG/AMP 2015 definition (PMID:25741868); no other threshold applies.
Still missing
No pedigree, no affected-relative genotypes and no informative meioses are available, so lack of segregation (BS4) cannot be demonstrated.
PMID:37088804, cited for this variant by a ClinVar submitter, was not retrieved in full text and could not be checked for non-segregation data.
PMID:33451724 has no text extract (page image only) and could not be searched for this variant.
BP5Not assessed: no case carrying this variant was reported anywhere in the evidence, so the alternate-molecular-basis observation BP5 requires could not be evaluated.
Details ▾Hide ▴
Reasoning
Framework: no ClinGen CSPEC or gene-specific VCEP specification exists for MSH3, so generic ACMG/AMP 2015 (PMID:25741868) governs. Generic BP5 requires that the variant be found in a case with an alternate molecular basis for the disease - i.e. it is a case-level criterion needing an affected individual carrying this variant in whom a different (typically pathogenic) variant in another gene explains the phenotype. No such case exists in the evidence assembled for this variant: no source retrieved in this case reports any patient carrying c.1522A>G/p.(K508E) at all (PMID:28944238, the one cohort study of MSH3 in CRC cases, does not contain the variant; the ClinVar records are laboratory classifications without case-level detail such as co-occurring pathogenic variants or tumour MMR/MSI results). Because there is no case to test, BP5 cannot be evaluated and is not_assessed rather than not_met. BP5 under this governing framework has no numeric threshold and no comparison operator, so no >=/<= inequality (and no likelihood-ratio-style reversed comparison) applies. Related but non-qualifying observation: PMID:28944238 does describe a CRC case carrying two heterozygous variants (APC p.His2045fs plus MSH3 p.Gln74fs), but that involves a different MSH3 variant (p.Gln74fs) in a different person and cannot be transferred to c.1522A>G.
Rule
generic_acmg_2015 BP5: variant found in a case with an alternate molecular basis for disease; requires case-level observation of this variant (absent in this case)
Evidence considered
Framework selection: no CSPEC/VCEP for MSH3, so generic ACMG/AMP 2015 (PMID:25741868) governs; BP5 is defined there as the variant being found in a case with an alternate molecular basis for disease, with no numeric threshold or comparison operator (no inequality to apply and no likelihood-ratio-style reversed comparison defined for MSH3).
No retrieved source in this case reports any patient carrying c.1522A>G/p.(K508E): full-text searches of PMID:28944238 for c.1522A>G, p.(K508E), p.Lys508Glu, K508E and '508Glu' returned no match, and the ClinVar record (VCV000781999) contains classifications only, with no case-level co-occurring-variant or tumour-test detail and no proband information.
PMID:28944238 describes one CRC case carrying both APC p.His2045fs and MSH3 p.Gln74fs - a different MSH3 variant and different individual, so it does not satisfy BP5 for c.1522A>G.
ClinVar submission audit (clinvar): 7 clinical-laboratory submissions, none with an alternate-molecular-basis observation for this variant or for a proband; 2 submissions had usable PMID trails (SCV006057929 -> PMID:25741868; SCV004221042 -> PMID:37088804, PMID:28944238), neither of which reports a case carrying this variant.
Still missing
Need case-level data for this exact variant: co-occurring (likely) pathogenic variants in established CRC/MMR genes (MLH1, MSH2, MSH6, PMS2, APC, MUTYH, etc.) and/or tumour MMR-IHC/MSI results in carriers, to test BP5.
No proband-level record (phenotype, tumour studies, additional variants) was available for this case; the ClinVar submissions are laboratory classifications without case detail.
PMID:33451724 (screenshot only in this case) and PMID:33516529 (not retrieved in this case) could not be searched for case-level content.
Not applicable · 7· recorded as not applicableShow reasons ▾Hide reasons ▴PVS1N/APVS1 (ClinGen SVI PVS1 recommendations, PMC6185798, pvs1_generic_framework) applies only to null variant classes — nonsense, frameshift, canonical ±1/2 splice-consensus, translation-initiation-codon loss, and stop-loss variants — each expected to trigger nonsense-mediated decay or otherwise abolish normal protein function. NM_002439.5:c.1522A>G in MSH3 is a missense substitution predicted to produce NP_002430.3:p.(Lys508Glu). As a result, no null-variant mechanism (nonsense-mediated decay, truncation, or canonical splice disruption) is triggered by this variant class, so PVS1's structural prerequisite is not met by this variant class. This determination was made deterministically from the variant's consequence class, prior to LLM adjudication, because no structured VCEP criteria-combination framework governs this case (framework_complete=false); the generic ACMG/AMP 2015 fallback applies.
Details ▾Hide ▴
PM4N/APM4 (Richards et al. 2015, PMID:25741868) applies to protein length changes: in-frame insertions/deletions in a non-repetitive region, or stop-loss variants that extend the protein past the reference stop codon. NM_002439.5:c.1522A>G in MSH3 is a missense substitution predicted to produce NP_002430.3:p.(Lys508Glu). As a result, no change in protein length occurs, so there is nothing for this criterion to evaluate, so PM4's structural prerequisite is not met by this variant class. This determination was made deterministically from the variant's consequence class, prior to LLM adjudication, because no structured VCEP criteria-combination framework governs this case (framework_complete=false); the generic ACMG/AMP 2015 fallback applies.
Details ▾Hide ▴
PP5N/ANot applicable: ClinVar holds seven submissions for this variant with zero from expert panels and conflicting classifications, so the expert-panel pathogenic assertion PP5 requires does not exist.
Details ▾Hide ▴
Reasoning
PP5 under generic ACMG/AMP 2015 (PMID:25741868) is a repackaged assertion of a reputable source's classification, and this group's rule restricts it further: it may be applied only for an exact-variant ClinVar classification made by a ClinGen expert panel, where the expert-panel classification is Pathogenic or Likely pathogenic. The precondition is a property of the source, not a variable to threshold, and PP5 has no numeric threshold or comparison operator in the governing framework, so no >=/<= inequality applies. For ClinVar VCV000781999 - NM_002439.5(MSH3):c.1522A>G (p.Lys508Glu), the exact variant - the submission audit records 7 submissions from 7 clinical laboratories, of which 0 (zero) are expert-panel submissions (clinvar_submission_audit.json counts.expert_panel_submissions = 0; is_expert_panel=false for every submission), and the review status is 'criteria provided, conflicting classifications', i.e. 1 star, aggregate label 'Conflicting classifications of pathogenicity'. The submissions are ordinary laboratory assertions: 5 Likely benign, 1 Uncertain significance (GeneDx), 1 with no classification provided. There is therefore no expert-panel Pathogenic/Likely pathogenic assertion for this exact variant to repackage, and the conflicting aggregate label is explicitly excluded from use. With the criterion's gating condition absent, PP5 is not_applicable rather than not_met. The ClinVar-cited papers carrying PP5 as a candidate criterion in this case (PMID:24493721, PMID:24929052, PMID:25394175; also PMID:33451724 and PMID:33516529, neither searchable for the variant in this case) are consensus statements/reviews that do not mention MSH3 c.1522A>G or p.(K508E) and contain no expert-panel assertion of the variant's classification; PMID:25741868 is the criteria framework itself.
Rule
group_rule PP5: requires an exact-variant ClinGen expert-panel ClinVar assertion that is Pathogenic or Likely pathogenic; non-expert submissions and aggregate labels are excluded from use
Evidence considered
Group rule applied for PP5: usable only when a ClinGen expert-panel (3-star) ClinVar classification of this exact variant is Pathogenic or Likely pathogenic; ordinary laboratory submissions, aggregate ClinVar labels and non-expert assertions must never trigger it. Evaluated as a gating condition on the source, not as a numeric threshold (the generic framework defines no threshold or comparison operator for PP5).
clinvar / clinvar_submission_audit.json for VCV000781999 (NM_002439.5(MSH3):c.1522A>G, p.Lys508Glu, exact match by transcript HGVS): 7 submissions (Ambry Genetics, PreventionGenetics, Sinai Health System, GeneDx, Dasa, Quest Diagnostics, Labcorp/Invitae), expert_panel_submissions = 0, is_expert_panel = false for all, review status 'criteria provided, conflicting classifications' = 1 star, api classification 'Conflicting classifications of pathogenicity'.
Per-submission classifications: 5 x Likely benign (Ambry, PreventionGenetics, Sinai Health System, Dasa, Labcorp/Invitae), 1 x Uncertain significance (GeneDx), 1 x likely benign but with 'no classification provided' record present (Peter K. Rogan Lab, not applicable origin) - no expert-panel assertion of any kind.
ClinVar-cited PP5-candidate papers reviewed: PMID:24493721 (ASCO cancer family-history expert statement - MSH3 not among the MMR genes listed), PMID:24929052 (SGO endometrial cancer review - MSH3 not named), PMID:25394175 (ACMG/NSGC cancer-predisposition referral guideline - CMMRD MMR gene list excludes MSH3). None is an expert-panel variant classification and none mentions c.1522A>G or p.(K508E); PMID:25741868 is the ACMG/AMP criteria document itself.
No ClinGen CSPEC or MSH3 VCEP specification exists in this case (cspec found=false; vcep_materials.json 'no cspec url available'), so no expert panel's criteria specification supplied an alternative PP5 definition.
Still missing
None material to this criterion: its precondition (an exact-variant ClinGen expert-panel Pathogenic/Likely pathogenic assertion) is verifiably absent from the ClinVar record consulted. If a ClinGen MSH3 VCEP is later established and classifies this variant, PP5 would need re-evaluation.
PMID:33451724 (screenshot only) and PMID:33516529 (not retrieved) could not be text-searched; both are endometrial-cancer clinical reviews with no expected variant-level content.
BS2N/ANot applicable: MSH3 disease is adult-onset with unestablished penetrance, so BS2's requirement of full penetrance at an early age is unmet, and its 2 gnomAD homozygotes are uninformative.
Details ▾Hide ▴
Reasoning
BS2 (strong benign; observed in a healthy adult individual for a recessive homozygous, dominant heterozygous or X-linked hemizygous disorder with full penetrance expected at an early age) is not applicable to this variant. The criterion's precondition is phenotype-, onset- and penetrance-dependent, and it fails here: MSH3-associated disease is an autosomal recessive susceptibility (attenuated adenomatous polyposis) with adult onset and unestablished penetrance, and MSH3 has minimal evidence for a role in Lynch syndrome, so there is no MSH3-associated disorder for which full penetrance at an early age can be assumed in carriers. The homozygous-state data were nevertheless examined because this group prioritizes homozygote observations: gnomAD v4.1 reports 2 homozygotes, both in the African/African American exome subset (afr_xx 1 of 41,458 alleles, afr_xy 1 of 33,526 alleles; 2 of 1,612,578 alleles overall) against a total allele count of 392, and gnomAD v2.1, gnomAD v2.1 non-cancer, gnomAD v3.1 non-cancer and gnomAD-Canada v1.0 report none. Those two homozygotes cannot be used as BS2 support because gnomAD participants are not phenotyped for colorectal neoplasia or polyposis (health status for this phenotype is unverified) and because BS2 requires an early-onset, fully penetrant disorder, which this gene-disease pair does not have; conversely, two homozygotes in a reference database do not exclude an adult-onset, reduced-penetrance recessive risk allele. Recorded as not_applicable rather than not_met: under either label BS2 contributes no benign evidence for this variant, and the difference is one of framing, not of evidence.
Rule
generic_acmg: BS2 when a homozygous (recessive) or heterozygous (dominant) carrier state is observed in a healthy adult for a disorder with full penetrance expected at an early age; precondition not satisfied for adult-onset, reduced-penetrance MSH3-associated disease
Evidence considered
gnomAD v4.1 homozygote data: total_hom 2 (exome homozygotes 2, genome homozygotes 0) within 1,612,578 alleles; homozygotes are in the African/African American subset (afr homozygote count 2, afr_xx 1, afr_xy 1).
gnomAD v2.1 homozygote data: total_hom 0 (exome 0, genome 0) within 282,466 alleles, including 0 homozygotes among 24,774 African/African American alleles.
gnomAD v2.1 non-cancer (exomes) and gnomAD v3.1 non-cancer (genomes): 0 homozygotes (101/267,946 and 170/147,900 alleles respectively).
gnomAD-Canada v1.0 (HostSeq genomes): 0 homozygotes within 18,420 alleles.
Disease-model context for the BS2 precondition: MSH3 encodes MutS-beta with MSH2; MSH3 defects are described as autosomal recessive susceptibility to adenomatous polyposis/endometrial cancer and MSH3 has minimal evidence for a role in Lynch syndrome, so neither an early-onset fully penetrant recessive phenotype nor a fully penetrant dominant phenotype can be assumed for carrier status.
gnomAD individuals are not ascertained or phenotyped for colorectal neoplasia or polyposis, so the healthy-adult premise of BS2 cannot be verified from this source for this gene-disease pair.
No paper in the pre-computed full-text extractions mentions this variant or reports homozygote/healthy-adult data relevant to BS2, so no publication contributes evidence to this criterion.
Still missing
No curated colorectal phenotype data (adenoma/polyposis status) exist for the 2 gnomAD v4.1 homozygotes, so their unaffected status for the relevant disorder could not be confirmed.
Penetrance and age of onset for biallelic MSH3-associated disease are not established in the available sources, which is why BS2 was treated as not applicable rather than assessed as met or not met.
BP3N/ABP3 (Richards et al. 2015, PMID:25741868) applies to in-frame insertions/deletions located in a repetitive region without a known function. NM_002439.5:c.1522A>G in MSH3 is a missense substitution predicted to produce NP_002430.3:p.(Lys508Glu). As a result, the variant does not alter protein length within a repeat region, so there is nothing for this criterion to evaluate, so BP3's structural prerequisite is not met by this variant class. This determination was made deterministically from the variant's consequence class, prior to LLM adjudication, because no structured VCEP criteria-combination framework governs this case (framework_complete=false); the generic ACMG/AMP 2015 fallback applies.
Details ▾Hide ▴
BP6N/ANot applicable: no expert-panel ClinVar assertion exists for this variant (seven laboratory submissions, conflicting classifications), so BP6's expert-panel benign precondition is unmet.
Details ▾Hide ▴
Reasoning
BP6 under generic ACMG/AMP 2015 (PMID:25741868) is a repackaged assertion of a reputable source's benign classification, and this group's rule restricts it further: it may be applied only for an exact-variant ClinVar classification made by a ClinGen expert panel, where the expert-panel classification is Benign or Likely benign. This is a gating condition on the source rather than a variable to threshold, and the governing framework defines no numeric threshold or comparison operator for BP6, so no >=/<= inequality applies. For ClinVar VCV000781999 - NM_002439.5(MSH3):c.1522A>G (p.Lys508Glu) - the submission audit records 7 submissions from 7 clinical laboratories with 0 expert-panel submissions (counts.expert_panel_submissions = 0; is_expert_panel=false for every submission) and review status 'criteria provided, conflicting classifications' (1 star), aggregate label 'Conflicting classifications of pathogenicity'. Five individual laboratories asserted Likely benign, but these are ordinary laboratory submissions; under this group's rule they must never trigger BP6, and neither may the conflicting aggregate ClinVar label. Because no expert panel has classified this exact variant at all, the criterion's precondition is absent and BP6 is not_applicable rather than not_met. For completeness, none of the seven submissions is from a ClinGen MSH3 VCEP, and no MSH3 VCEP/CSPEC exists in this case, so there is no criteria specification that could define BP6 differently.
Rule
group_rule BP6: requires an exact-variant ClinGen expert-panel ClinVar assertion that is Benign or Likely benign; non-expert submissions and aggregate labels are excluded from use
Evidence considered
Group rule applied for BP6: usable only when a ClinGen expert-panel (3-star) ClinVar classification of this exact variant is Benign or Likely benign; ordinary laboratory submissions, aggregate ClinVar labels and non-expert assertions must never trigger it. Evaluated as a gating condition on the source, not as a numeric threshold (no threshold or comparison operator for BP6 exists in the governing generic framework).
clinvar / clinvar_submission_audit.json for VCV000781999 (exact match by transcript HGVS: NM_002439.5(MSH3):c.1522A>G, p.Lys508Glu): expert_panel_submissions = 0; every submission has is_expert_panel = false; review status 'criteria provided, conflicting classifications' = 1 star; aggregate classification 'Conflicting classifications of pathogenicity' (explicitly excluded from use).
Submitting laboratories and classifications (all non-expert clinical laboratories): Ambry Genetics Likely benign, PreventionGenetics Likely benign (no assertion criteria provided), Sinai Health System Likely benign, Dasa Likely benign, Labcorp/Invitae Likely benign, GeneDx Uncertain significance, Quest Diagnostics likely benign - and one record with no classification provided (Peter K. Rogan Lab, origin 'not applicable'). No expert-panel assertion exists.
No ClinGen MSH3 VCEP/CSPEC was available in this case (cspec found=false; vcep_materials.json 'no cspec url available'; final_classification_framework.json framework_complete=false), so no specification supplies an alternative BP6 definition or a numeric likelihood-ratio threshold.
Still missing
None material to this criterion: its precondition (an exact-variant ClinGen expert-panel Benign/Likely benign assertion) is verifiably absent from the ClinVar record consulted. If a ClinGen MSH3 VCEP is later established and classifies this variant, BP6 would need re-evaluation.
PMID:33451724 (screenshot only) and PMID:33516529 (not retrieved) could not be text-searched for variant-level content.
BP7N/ABP7 (Richards et al. 2015, PMID:25741868) is defined for a synonymous (silent) variant for which splicing prediction algorithms predict no impact to the splice consensus sequence and the nucleotide is not highly conserved. NM_002439.5:c.1522A>G in MSH3 is a missense substitution predicted to produce NP_002430.3:p.(Lys508Glu). As a result, the encoded protein sequence is altered, so the criterion's silent-variant premise does not hold regardless of any splicing prediction, so BP7's structural prerequisite is not met by this variant class. This determination was made deterministically from the variant's consequence class, prior to LLM adjudication, because no structured VCEP criteria-combination framework governs this case (framework_complete=false); the generic ACMG/AMP 2015 fallback applies.
Details ▾Hide ▴
Rule
deterministic_prefilter: requires_synonymous
Evidence considered
Deterministic consequence classification: missense (prefetch.json normalized protein consequence)
Sources
generic_acmg_combination_rules
Identifiers
Gene
MSH3
Transcript
NM_002439.5
Coding (HGVS c.)
NM_002439.5:c.1522A>G
Protein (HGVS p.)
NP_002430.3:p.(Lys508Glu)
GRCh38
chr5:80728919 A>G
GRCh37
chr5:80024738 A>G
Location and type
Exon 10 · missense
ClinVar
781999
Gene diagram
Every interpreted variant in MSH3, mapped onto the exon structure of NM_002439.5. Select a row to find it on the plot.
MSH3NM_002439.5
Fetching transcript structure from UCSC…
Exons
—
Transcript span
—
Strand
—
Variants mapped
—
Source
UCSC ncbiRefSeqCurated
All variants in MSH3—Select a row to find it on the plot; use the link to open its page.
AF 0.00024 · 392 / 1,612,578 alleles · 2 hom · FAF 0.004
African/African American329 / 74,984
0.00442 hom
Remaining individuals36 / 62,452
0.00058
Middle Eastern2 / 6,046
0.00033
Admixed American18 / 60,026
0.0003
South Asian1 / 91,044
1.1e-05
European (non-Finnish)6 / 1,178,706
5.1e-06
0 alternate alleles recorded in 4 other groups: European (Finnish), Amish, East Asian, Ashkenazi Jewish.
gnomAD v2.1
AF 0.00038 · 106 / 282,466 alleles · 0 hom · FAF 0.0035
African/African American96 / 24,774
0.0039
Remaining individuals2 / 7,198
0.00028
Admixed American5 / 35,414
0.00014
European (non-Finnish)3 / 129,078
2.3e-05
0 alternate alleles recorded in 4 other groups: Ashkenazi Jewish, East Asian, European (Finnish), South Asian.
gnomAD-Canada
AF not recorded · ? / ? alleles
Statements
gnomAD v4.1
This variant is present in gnomAD v4.1 (AF= 0.000243089; MAF= 0.02431%, 392/1612578 alleles, homozygotes = 2) and has highest observed frequency in the African/African American population (AF= 0.0043876; MAF= 0.43876%, 329/74984 alleles, homozygotes = 2); grpmax FAF= 0.00399731.
gnomAD v2.1
This variant is present in gnomAD v2.1 (AF= 0.000375266; MAF= 0.03753%, 106/282466 alleles, homozygotes = 0) and has highest observed frequency in the African/African American population (AF= 0.00387503; MAF= 0.38750%, 96/24774 alleles, homozygotes = 0); grpmax FAF= 0.00348064.
gnomAD-Canada
This variant is present in gnomAD-Canada v1.0 (AF= 0.00016286644951140066, 3/18420 alleles, homozygotes = 0).
BP4SupportingMet at supporting strength: REVEL 0.191 is within the <=0.29 BP4 supporting threshold but above the <=0.183 moderate cut-off.
Details ▾Hide ▴
Reasoning
Scope settled first: NM_002439.5:c.1522A>G is a missense substitution (NP_002430.3:p.(Lys508Glu)), so BP4 is adjudicated on the missense in-silico path (REVEL) only; the SpliceAI score is not admissible as BP4 support for a missense variant. With no ClinGen CSPEC specification or gene-specific VCEP framework for MSH3, the generic ACMG/AMP 2015 fallback and the published ClinGen SVI REVEL calibration apply. The local REVEL v1.3 score is 0.191, which satisfies the BP4 supporting threshold of <=0.29 (Pejaver et al. 2022, PMID:36413997) but does not reach the moderate (<=0.183) or strong (<=0.016) thresholds, so BP4 is met at supporting strength only. The score is not in the indeterminate gray zone (0.29-0.644). This criterion is not double-counted with any splice-prediction or protein-length evidence, and the same REVEL value is the basis of the PP3 not-met call rather than a second, independent assertion. The supporting-strength call is made directly from the calibrated score alone; no other source's classification was used to reach it.
Path selection: missense consequence (NP_002430.3:p.(Lys508Glu)) selects the REVEL missense path for BP4; SpliceAI max delta 0.002 was not used as BP4 support.
Local REVEL v1.3 lookup returned revel_score 0.191 for chrom 5, pos 80728919 (GRCh38), ref A, alt G; found=true.
REVEL thresholds applied verbatim from the ClinGen SVI calibration (Pejaver et al. 2022, Am J Hum Genet, PMID:36413997): BP4 supporting <=0.29, moderate <=0.183, strong <=0.016.
BayesDel score -0.230708 was not used; no established generic ACMG-strength calibration exists for BayesDel, so it is not_available for BP4.
Governing framework: no ClinGen CSPEC entry for MSH3 and no gene-specific VCEP materials or pre-assigned PP3/BP4 lookup code in this case, so generic ACMG/AMP 2015 applies.
Corroborative context only, not the basis of the call: a ClinVar submission for this variant (GeneDx, SCV002558398, Uncertain significance) states that in silico analysis suggests this missense variant does not alter protein structure/function; that submission carries no explicit PP3/BP4 code and was not used to derive the strength.
PP3Not met: REVEL 0.191 for this missense variant is below the >=0.644 PP3 supporting threshold.
Details ▾Hide ▴
Reasoning
Scope settled first: NM_002439.5:c.1522A>G in MSH3 is a missense substitution, predicted to encode NP_002430.3:p.(Lys508Glu), so it falls on the missense in-silico path (REVEL) and the splice path is not used for this criterion. No ClinGen CSPEC specification and no gene-specific VCEP framework exist for MSH3 (cspec found=false; framework_complete=false; framework_mode=generic_acmg), so the generic ACMG/AMP 2015 fallback with the published ClinGen SVI REVEL calibration applies. The local REVEL v1.3 lookup returned a score of 0.191 for the variant's GRCh38 coordinates (chr5:80728919 A>G, chromosome 5, ref A, alt G), which is below the PP3 supporting threshold of >=0.644 and therefore also below the moderate (>=0.773) and strong (>=0.932) thresholds (Pejaver et al. 2022, PMID:36413997). The BayesDel score in the evidence (-0.230708) was not used because no generic ACMG-strength calibration for BayesDel is established; the SpliceAI max delta (0.002) was not used because splice prediction is not the applicable path for a missense variant. PP3 is therefore not met.
Path selection: missense consequence (NP_002430.3:p.(Lys508Glu)) selects the REVEL missense path for PP3/BP4; the SpliceAI path was not used for either criterion.
Local REVEL v1.3 lookup returned revel_score 0.191 for chrom 5, pos 80728919 (GRCh38), ref A, alt G, matching the case's normalized GRCh38 coordinates; found=true.
REVEL thresholds applied verbatim from the ClinGen SVI calibration (Pejaver et al. 2022, Am J Hum Genet, PMID:36413997): PP3 supporting >=0.644, moderate >=0.773, strong >=0.932.
BayesDel score -0.230708 present but not used: no generic ACMG-strength PP3/BP4 calibration exists for BayesDel.
Governing framework: no ClinGen CSPEC entry for MSH3 and no gene-specific VCEP materials or PP3/BP4 lookup table in this case (vcep_materials.json found=false, framework_complete=false), so generic ACMG/AMP 2015 applies.
SpliceAI lookup for this variant reported max delta score 0.002 (DS_AG 0.002, all other deltas 0.0); this was deliberately not carried into PP3/BP4 because the variant is missense.
BP7N/ABP7 (Richards et al. 2015, PMID:25741868) is defined for a synonymous (silent) variant for which splicing prediction algorithms predict no impact to the splice consensus sequence and the nucleotide is not highly conserved. NM_002439.5:c.1522A>G in MSH3 is a missense substitution predicted to produce NP_002430.3:p.(Lys508Glu). As a result, the encoded protein sequence is altered, so the criterion's silent-variant premise does not hold regardless of any splicing prediction, so BP7's structural prerequisite is not met by this variant class. This determination was made deterministically from the variant's consequence class, prior to LLM adjudication, because no structured VCEP criteria-combination framework governs this case (framework_complete=false); the generic ACMG/AMP 2015 fallback applies.
Details ▾Hide ▴
Rule
deterministic_prefilter: requires_synonymous
Evidence considered
Deterministic consequence classification: missense (prefetch.json normalized protein consequence)
Sources
generic_acmg_combination_rules
Statement
In silico
SpliceAI predicts no significant splice impact for this variant (max delta score = 0.00). REVEL score = 0.191. BayesDel score = -0.230708.
8 paper records listed. 1 cited in the classification. None of the listed paper records explicitly indicates that the variant is named.
Cited in the classification
Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
Richards S et al. · 2015 May
Richards et al. 2015 ACMG/AMP standards and guidelines for sequence variant interpretation. It is a consensus framework, not a study of MSH3 or of c.1522A>G, and reports no variant-specific data. It defines the criteria semantics used for this group (PS1 same-amino-acid-change with a different nucleotide change; PM1 mutational hotspot or critical functional domain without benign variation; PM5 different pathogenic missense at the same residue; PP2 missense in a gene with low benign missense rate where missense is a common disease mechanism; BP1 missense in a gene where primarily truncating variants cause disease) and the combining rules, and it explicitly states that gene/disease-specific groups should refine criteria weights.
→ BP1 · supporting→ BP4 · supportingDoes not name this variant
We should also note that those working in specific disease groups should continue to develop more focused guidance regarding the classification of variants in specific genes given that the applicability and weight assigned to certain criteria may vary by gene and disease.
Location Methods (workgroup/literature review), final paragraph of the Methods section
Context Not applicable - expert-opinion/consensus guideline development: workgroup surveys of >100 GeneTests.org sequencing laboratories (Feb 2013), analysis of 11 submitted laboratory variant-assessment protocols, a second survey (~2000 AMP list-serve members, Aug 2013), and a Nov 2013 AMP workshop audience-response weighting exercise; no bench or cohort data on MSH3.
Other recorded publications
American Society of Clinical Oncology Expert Statement: collection and use of a cancer family history for oncology providers.
Lu KH et al. · 2014
An ASCO expert consensus statement on collecting and using cancer family history for oncology providers; it is not a gene- or variant-level study and does not mention MSH3 or c.1522A>G / p.(K508E) anywhere. Its only MSH3-relevant content is a generic recommendation that mismatch repair-deficient colorectal and endometrial cancers should prompt genetic counselling and testing, and the MMR loci it names are MLH1, MSH2, MSH6, PMS2 and EPCAM - MSH3 is not included. It provides no de novo or segregation evidence for this variant; it was reviewed because it was triaged as a family-history/case-observation source for this case.
Colorectal cancer demonstrating mismatch repair deficiency (via tumor studies including microsatellite instability analysis and/or immunohistochemistry, excluding known somatic causes including hypermethylation of MLH1 promoter and somatic BRAF mutation) - MLH1/MSH2/MSH6/PMS2/EPCAM
Table 2, 'Common adult cancers' (Cancers for Which Genetic Counseling and Testing Should Be Considered, Even in Absence of Family History)
Endometrial cancer: a review and current management strategies: part II.
SGO Clinical Practice Endometrial Cancer Working Group et al. · 2014
A narrative review of endometrial cancer management by the SGO Clinical Practice Endometrial Cancer Working Group. MSH3 is not named anywhere in the text and this variant is absent. Its only genetics content concerns MMR/Lynch syndrome in endometrial cancer - one cited study found 7 of 102 women (7%) with synchronous endometrial and ovarian cancer met clinical or molecular criteria suggestive of Lynch syndrome, and the authors stress that family history is critical in deciding whether to test - which is general MMR/Lynch context, not segregation or de novo evidence for MSH3 c.1522A>G.
In a recent study, 7 out of 102 women (7%) with synchronous endometrial and ovarian cancer had either clinical or molecular criteria suggestive for Lynch syndrome, with most of both tumors having endometrioid histology [41]. The incidence of Lynch syndrome (HNPCC) was low unless there was a family history of HNPCC-associated cancers. Thus, family history is critical in the decision to test for HNPCC
Synchronous endometrial and ovarian carcinoma - 'Do genetics play a role in synchronous endometrial and ovarian cancers?', paragraph 1
A practice guideline from the American College of Medical Genetics and Genomics and the National Society of Genetic Counselors: referral indications for cancer predisposition assessment.
Hampel H et al. · 2015
An ACMG/NSGC clinical practice guideline on referral indications for cancer predisposition assessment. It reports no functional, case, segregation, or population-frequency data and does not mention MSH3 or c.1522A>G / p.(K508E). Its only variant-relevant content is a gene-level statement in the constitutional mismatch repair deficiency section listing MLH1, MSH2 (including methylation due to an EPCAM deletion), MSH6 and PMS2 as the implicated MMR genes - MSH3 is not among them - so the paper provides no evidence for or against this variant in this group.
Constitutional mismatch repair deficiency is a recessive condition caused by biallelic mutations in the mismatch repair genes (MLH1, MSH2 (including methylation due to an EPCAM deletion), MSH6, and PMS2) and is characterized by a high risk of developing cancers during childhood, including Lynch syndrome (LS)-associated cancers, hematologic malignancies, and embryonic tumors.
A Standardized DNA Variant Scoring System for Pathogenicity Assessments in Mende
A methodology report describing a rules-based, weighted 7-point variant pathogenicity scoring system aligned to ACMG/AMP categories, applied retrospectively to 16,500 assessments of 11,894 variants in 338 genes in a CLIA-certified laboratory. It does not mention MSH3 or c.1522A>G / p.(K508E) and provides no variant-specific data; its only relevance to this group is that family segregation (LOD) is one of the five weighted evidence types in its system, which was reviewed when considering PP1.
Sherloc: a comprehensive refinement of the ACMG-AMP variant classification crite
Sherloc, a semiquantitative refinement of the ACMG-AMP variant classification criteria (108 criteria, version 4.2), derived from >40,000 clinically observed variants across >500 genes. MSH3 is not mentioned and c.1522A>G (p.Lys508Glu) is not named; the paper provides no variant-specific functional, case, segregation, or frequency data. Its relevance to this group is limited to how segregation-type evidence (and its refinements) is defined and weighted generically, which was consulted for PP1 and BS4.
Targeted sequencing of 36 known or putative colorectal cancer susceptibility genes.
DeRycke MS et al. · 2017
Targeted sequencing of 36 known or putative colorectal cancer susceptibility genes in 1231 CRC cases plus 93 spouse controls treats MSH3 only at gene level (transcript NM_002439.4). It does not mention MSH3 c.1522A>G or p.Lys508Glu anywhere. Relevant to this group only as gene-level evidence that MSH3 missense variation is common in populations: 12 predicted-damaging Tier 2 MSH3 variants were identified, and MSH3 carrier rates were raised in unselected CRC cases largely because of three recurrent missense variants (p.Ser490Tyr, p.His827Arg, p.Leu911Trp) that were absent from Caucasian public controls but present in African American controls at 0.38-0.91%. This supports treating MSH3 as a gene with abundant benign/uncertain missense variation (relevant to PP2) but it provides no variant-specific evidence for p.Lys508Glu and is not cited as a variant-level source.
In the unselected cases, MSH3 also had a high percent carrier count, primarily due to three variants present in five (p.Ser490Tyr and p.His827Arg) or four (p.Leu911Trp) individuals each (Table S2).
Results, 'Tier 2 variants' section (paragraph discussing unselected cases and MSH3); also Table 4 and Table S2
Endometrial cancer: A society of gynecologic oncology evidence-based review and recommendations, part II.
The evidence statements recorded by the interpretation pipeline. They have not been reinterpreted by the page.
gnomAD v4.1
This variant is present in gnomAD v4.1 (AF= 0.000243089; MAF= 0.02431%, 392/1612578 alleles, homozygotes = 2) and has highest observed frequency in the African/African American population (AF= 0.0043876; MAF= 0.43876%, 329/74984 alleles, homozygotes = 2); grpmax FAF= 0.00399731.
gnomAD v2.1
This variant is present in gnomAD v2.1 (AF= 0.000375266; MAF= 0.03753%, 106/282466 alleles, homozygotes = 0) and has highest observed frequency in the African/African American population (AF= 0.00387503; MAF= 0.38750%, 96/24774 alleles, homozygotes = 0); grpmax FAF= 0.00348064.
gnomAD-Canada
This variant is present in gnomAD-Canada v1.0 (AF= 0.00016286644951140066, 3/18420 alleles, homozygotes = 0).
ClinVar
This variant is present in ClinVar (Variation ID: 781999); submission details unavailable.
In silico
SpliceAI predicts no significant splice impact for this variant (max delta score = 0.00). REVEL score = 0.191. BayesDel score = -0.230708.
Functional
OncoKB did not identify variant-specific reviewed functional evidence for this variant; gene-level curated context is available for reviewer follow-up. MSH3, a DNA mismatch repair protein, is frequently mutated in colorectal cancers.
Somatic
This variant does not lie in a statistically significant hotspot. This variant has not previously been reported in somatic cancers (COSMIC).
Cancer Hotspots
This variant does not lie in a statistically significant hotspot.
This variant was interpreted by pipeline 8.0.0.
The current version is 11.0.07.
Re-running queues a fresh interpretation; the result replaces this page when complete.
Re-runs are not free. Blank draws on today's public interpretation allowance; with a code, it uses one of that code's uses.