MSH6 c.2017C>G, p.(Pro673Ala)

Reasoning

Governing framework: ClinGen InSiGHT MSH6 v2.0, which provides exactly one applicable PM2 strength - Supporting - with the rule 'Absent/extremely rare allele frequency <0.00002 (<1 in 50,000 alleles) in gnomAD v4 dataset'; the specification's own PS1 moderate rule refers to 'PM2_supporting', confirming the intended strength. The variant is not absent, so it must qualify through the 'extremely rare' branch. gnomAD v4.1 gives a total AF of 4.337846e-06 (7/1,613,704 alleles, i.e. approximately 1 in 230,529 alleles; the exome-only component is 2/1,461,876 = 1.368105e-06) and a Grpmax filtering AF of 1.83e-06 - both roughly 5- to 11-fold below the 0.00002 cutoff. Concordant independent data support genuine rarity: gnomAD v2.1 total AF 1.416832e-05 (4/282,320 alleles), gnomAD v2.1 non-cancer exomes 8.458949e-06 (2/236,436 alleles), gnomAD v3.1 non-cancer genomes 3.388039e-05 (5/147,578 alleles), zero homozygotes in every dataset, and absence from gnomAD-Canada v1.0. For an autosomal dominant, incompletely penetrant condition, the absence of any homozygous carrier and an allele frequency below 1 in 50,000 in every gnomAD iteration are consistent with the rarity that PM2_supporting is designed to capture. One source-dependent threshold artefact is recorded: the gnomAD v3.1 non-cancer genome subset reports a Grpmax filtering AF of 2.849e-05 (driven by 5/64,724 European non-Finnish alleles), which nominally exceeds the 0.00002 cutoff because the non-cancer filter shrinks the denominator without removing the five observed alleles; however, the governing specification names the gnomAD v4 dataset, whose all-comers figure (1.83e-06) and total AF (4.337846e-06) are both decisively below the cutoff, so PM2 is met at Supporting. PM2 must not be escalated: the specification offers no moderate-strength PM2 rule for this gene.

Rule

Absent/extremely rare allele frequency <0.00002 (<1 in 50,000 alleles) in gnomAD v4 dataset

Evidence considered

• ClinGen InSiGHT MSH6 v2.0 (cspec, doc 1578294885) PM2 rule at Supporting strength: 'Absent/extremely rare allele frequency <0.00002 (<1 in 50,000 alleles) in gnomAD v4 dataset'. The CSPEC release notes state: 'Amendment: Minor change to PM2 allele frequency format (1/50000 = 0.00002)'.
• The same specification's PS1 moderate rule requires that 'PM2_supporting is met', confirming that the VCEP intends PM2 to be applied at Supporting strength only.
• gnomAD v4.1 (chr2-47800000-C-G, GRCh38): total AF 4.337846e-06 (7/1,613,704 alleles, ~1 in 230,529 alleles), Grpmax filtering AF 1.83e-06, 0 homozygotes; exome-only 2/1,461,876 (1.368105e-06) and genome-only 5/151,828 (3.29320e-05).
• gnomAD v2.1 (GRCh37, 2-48027139-C-G): total AF 1.416832e-05 (4/282,320 alleles), Grpmax filtering AF 2.241e-05, 0 homozygotes - also below the 0.00002 PM2 cutoff.
• gnomAD v2.1 non-cancer exomes: AF 8.458949e-06 (2/236,436 alleles), 0 homozygotes. gnomAD v3.1 non-cancer genomes: total AF 3.388039e-05 (5/147,578 alleles), Grpmax filtering AF 2.849e-05, highest subpopulation European (non-Finnish) 7.725110e-05 (5/64,724 alleles), 0 homozygotes - this subset Grpmax nominally exceeds the 0.00002 cutoff, but the VCEP-specified gnomAD v4 dataset governs.
• gnomAD-Canada v1.0 (HostSeq genomes): variant absent, providing further independent support for rarity beyond gnomAD.
• No ancestry group in gnomAD v4.1 reaches a frequency that would argue against rarity: the highest subpopulation AF is Admixed American 1.667834e-05 (1/59,958 alleles).
• Homozygote and penetrance context: 0 homozygotes are reported in every gnomAD iteration for this variant, and MSH6-associated Lynch syndrome is a dominant, incompletely penetrant adult-onset condition, so an extremely low heterozygote frequency together with no biallelic carrier is the configuration PM2_supporting addresses; the low allele count does not justify escalation beyond Supporting, which the specification does not provide for PM2 in any case.

Still missing

• Source-dependent threshold artefact: the gnomAD v3.1 non-cancer genome subset Grpmax filtering AF is 2.849e-05, nominally above the 0.00002 PM2 cutoff, whereas the VCEP-specified gnomAD v4 all-comers figure is 1.83e-06. The VCEP-specified gnomAD v4 dataset was used as required; the alternate subset figure is recorded so a reviewer can see the sensitivity of this call to source selection.
• The variant is present rather than absent in gnomAD (7 alleles in v4.1), so PM2 rests on the 'extremely rare' branch and is only a Supporting-level criterion; it should not be counted at moderate strength.

Sources

Reasoning

Scope first: c.2017C>G is a missense substitution (NP_000170.1:p.(Pro673Ala)), so BP4 is evaluated on the missense protein-impact path. The governing ClinGen InSiGHT Hereditary Colorectal Cancer/Polyposis VCEP specification for MSH6 (v2.0, cspec doc 1578294885) states BP4 (default strength Benign Supporting) as: missense variant with HCI-prior probability of pathogenicity <0.11 (the alternative clause, SpliceAI delta <=0.1, applies only to intronic and synonymous variants). The declared local VCEP lookup file HCI-PRIORS-MSH6.txt was searched under all notations of this variant; the exact row is present at line 4413 (exon 4, c.2017C>G, p.P673A, dbid MSH6_04411, reference {PMID22949387:Thompson et al., 2013}) with prior probability 0.0159. That is well below the <0.11 cutoff, so BP4 is met at supporting strength, exactly as pre-assigned by the VCEP's own supporting materials; the VCEP's default strength for this criterion is Benign Supporting, so no strength upgrade or downgrade is applied. The same table row shows a custom PP2 score of 0.409, which is a different quantity from the prior probability and was not used. The generic missense path was checked only to confirm consistency and does not change the verdict: REVEL was 0.447, in the indeterminate band (above the <=0.29 supporting BP4 cutoff and below the 0.644 PP3 cutoff, Pejaver et al 2022, PMID:36413997), which under generic rules alone would simply mean BP4 is not met; because the VCEP specifies its own missense BP4 rule for MSH6, that prior-based rule governs and is satisfied. SpliceAI max delta was 0.014, but a splice prediction is never used as BP4 support for a missense variant, so it was not substituted for the protein-level evidence.

Rule

MSH6 VCEP v2.0 BP4 (Benign Supporting): missense variant with HCI-prior probability of pathogenicity <0.11. Observed HCI prior for c.2017C>G (p.P673A) = 0.0159 -> met at supporting strength.

Evidence considered

• Scope check: c.2017C>G is a missense variant (NP_000170.1:p.(Pro673Ala)), the consequence type BP4's missense clause covers; the intronic/synonymous SpliceAI clause of the VCEP BP4 rule does not apply.
• MSH6 VCEP v2.0 BP4 rule (default strength Benign Supporting, cspec doc 1578294885): missense variant with HCI-prior probability of pathogenicity <0.11; alternatively, for intronic and synonymous variants, SpliceAI predicts no splicing impact with delta score <=0.1.
• Exact lookup: HCI-PRIORS-MSH6.txt line 4413, exon 4, c.2017C>G, p.P673A, dbid MSH6_04411, prior probability 0.0159 (MAPP 4.34, custom PP2 0.409), reference {PMID22949387:Thompson et al., 2013}. 0.0159 < 0.11, so the VCEP's missense BP4 clause is satisfied by its own pre-assigned prior value.
• The '<...>_Prior' column is taken as the HCI prior probability of pathogenicity required by the rule; the custom PP2 score on the same row (0.409) is a separate in-silico score and does not meet the <0.11 criterion, so the column identification is the decisive step in this call.
• Strength: the VCEP assigns BP4 a default of Benign Supporting with no strength tiers, so the criterion is recorded at supporting strength.
• SpliceAI for NM_000179.3:c.2017C>G (max delta 0.014; DS_AG 0.014, DS_DG 0.008) is not used here: no splice prediction may be cited as BP4 support for a missense variant.
• Generic cross-check (not governing): REVEL 0.447 sits between the generic BP4 supporting cutoff <=0.29 (and <=0.183 moderate, <=0.016 strong) and the PP3 supporting cutoff 0.644 (Pejaver et al 2022, Am J Hum Genet, PMID:36413997); under generic calibration alone BP4 would not be met, and the VCEP's own prior-based rule was therefore applied instead of the generic REVEL thresholds.
• No case publication provided variant-specific computational evidence for c.2017C>G; the seven available papers (PMID:25741868, PMID:25394175, PMID:27854360, PMID:30267214, PMID:25711197, PMID:26324357, PMID:28492532) were screened via their pre-computed extractions and none mentions the variant.
• The same prior value was not double-counted: the 0.0159 prior supports BP4 only; it is below the PP3 thresholds, so PP3 and BP4 are mutually exclusive here.

Sources

Reasoning

NM_000179.3:c.2017C>G is a single-nucleotide substitution that mutalyzer resolves to NP_000170.1:p.(Pro673Ala) - one amino-acid change, with predicted protein start (residue 672) and end (residue 673) positions identical to the reference, i.e. no protein-length change and no premature termination codon. The variant lies deep inside the large exon 4 of NM_000179.3 (c.628-3172; c.2017 is 1389 bp from the 3' end of exon 3 and 1155 bp from the 5' end of exon 5), not at a consensus donor/acceptor dinucleotide, so it is not a canonical +/-1 or +/-2 splice-site variant. SpliceAI predicts no splicing aberration (max delta score 0.014; DS_AG 0.014, DS_DG 0.008, DS_AL/DS_DL 0.0 at distance 500), far below even the >=0.2 supporting cutoff, so there is no exon skipping or cryptic splice-site activation to convert the allele into a functionally null transcript, and no patient mRNA or minigene assay evidence of a splice defect exists in this case. The governing framework is the ClinGen InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel specification for MSH6 v2.0 (cspec), which is the applicable VCEP; its PVS1 decision tree restricts PVS1 to (a) nonsense/frameshift variants introducing a PTC at or before codon 1341, (b) large single- or multi-exon genomic alterations, (c) variants at IVS+/-1 or IVS+/-2 where exon skipping or cryptic splice-site use disrupts the reading frame (with NMD predictions governing Very Strong vs Moderate) or preserves the reading frame in a function-critical region (Strong), and (d) variants with constitutional mRNA assay evidence of a splicing aberration producing a PTC or a domain-disrupting in-frame deletion. None of these triggers applies: there is no nonsense or frameshift allele, no gross deletion/duplication, no canonical splice-site change, and no RNA assay. MSH6 loss of function is an established disease mechanism and the gene passed the gene-level PVS1 eligibility gate, but gene-level eligibility alone does not make a missense substitution a null variant. Because PVS1 is not met here, no strength tier is assigned and the Very Strong-versus-downgraded ambiguity does not arise; no PVS1 code string is written into the strength field.

Evidence considered

• Governing framework read first: ClinGen InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel Specifications for MSH6 v2.0 (cspec), precedence rank 1, framework_mode criteria_combination; generic ACMG PVS1 guidance was therefore not used as the decision tree.
• VCEP PVS1 applicability is 'Applicable' with default strength 'Pathogenic Very Strong'; the Very Strong rules are a nonsense/frameshift PTC at or before codon 1341 of MSH6, large single- or multi-exon genomic alterations, and IVS+/-1 or IVS+/-2 variants whose exon skipping/cryptic site disrupts the reading frame with NMD predicted.
• VCEP PVS1 decision tree (PVS1_DecisionTree_MMR.pdf summary) routes only 'Nonsense or Frameshift', 'GT-AG 1,2 splice sites', 'Deletion', and 'Duplication' branches to PVS1; nucleotide substitutions that merely change one amino acid are absent from every branch.
• Variant consequence scaffold: consequence_class = missense, canonical_splice_consensus = false, protein NP_000170.1:p.(Pro673Ala); apply_generic_pvs1_framework = false with rationale that the variant does not fall into the default generic PVS1 null-variant buckets.
• Variant normalization: mutalyzer returned NM_000179.3:c.2017C>G with predicted protein p.(Pro673Ala), position_first 672, position_last_original 673, position_last_predicted 673 - protein length unchanged, no truncation, no stop-loss.
• Exon context from the transcript selector: c.2017 sits within exon 4 of NM_000179.3 (c.628-3172), 1389 bp from the exon 4 start and 1155 bp from its end, so no canonical splice-consensus disruption is possible at this position.
• SpliceAI result (spliceailookup, hg37, distance 500): DS_AG 0.014, DS_AL 0.0, DS_DG 0.008, DS_DL 0.0, max delta score 0.014 - no predicted splice impact, so the allele cannot be treated as a spliceogenic null and rule (c) of the VCEP PVS1 set cannot be satisfied.
• Gene-level PVS1 gate: lof_mechanism_supported = true with pvs1_gene_gate = 'eligible' because the MSH6 VCEP itself includes PVS1 guidance; this establishes LoF as a disease mechanism for the gene but is not evidence that this particular missense allele is a null.
• No mRNA, minigene, or patient RNA assay data for c.2017C>G exists in the case files, so the VCEP's mRNA-assay PVS1 route (splicing aberration with demonstrated absence of full-length/reference transcript) cannot be invoked.
• OncoKB returned 'Unknown Oncogenic Effect' with no variant-specific reviewed functional evidence for MSH6 P673A, and Cancer Hotspots found no hotspot row for MSH6 P673 - neither source offers a truncating/null interpretation of this allele.
• Checked paper_extracts.json (pre-computed full-text extractions for PMID:25741868, PMID:25394175, PMID:27854360, PMID:30267214, PMID:25711197, PMID:26324357, PMID:28492532): none mentions c.2017C>G or p.(P673A), and the one LoF-focused burden study among them (PMID:30267214) analysed only potentially disruptive variants (stop-gain, frameshift, splice acceptor/donor), explicitly excluding missense changes, so no publication supplies PVS1-relevant consequence evidence for this variant.
• Because PVS1 is not met, strength is left null rather than populated with any tier word or criterion code.

Sources

Reasoning

PS1 has two branches, neither of which is satisfied. Branch 1 requires a predicted missense substitution that encodes the same amino acid change (p.Pro673Ala) with a different underlying nucleotide change already established as Pathogenic by this VCEP. MSH6 codon 673 is CCG, so alanine can only arise from a first-base change: c.2017C>G is the sole nucleotide substitution in the codon that encodes p.Pro673Ala. The other codon-673 substitutions encode different amino acids (c.2017C>A p.Pro673Thr, c.2017C>T p.Pro673Ser, c.2018C>A p.Pro673Gln, c.2018C>G p.Pro673Arg, c.2018C>T p.Pro673Leu), so no alternate-nucleotide p.Pro673Ala comparator can exist. Even setting the nucleotide question aside, no codon-673 variant carries a VCEP Pathogenic or Likely Pathogenic classification: the VCEP pilot-variant curation contains no MSH6 codon-673 entry, and the complete ClinVar set of MSH6 codon-673 missense variants (p.Pro673Thr, p.Pro673Ser, p.Pro673Ala, p.Pro673Leu) is uniformly Uncertain significance, with zero expert-panel submissions. Branch 2 (variant affecting the same non-canonical splice nucleotide as a confirmed Pathogenic splice variant with similar or worse SpliceAI prediction) is inapplicable: c.2017C>G is an exonic missense change in exon 4 of NM_000179.3, not a non-canonical splice nucleotide, and SpliceAI max delta is 0.014 with no donor/acceptor gain or loss. PS1 is not met at Strong or Moderate strength.

Rule

PS1 (Strong): A predicted missense substitution that encodes the same amino acid change with a different underlying nucleotide change previously established by this VCEP as Pathogenic (not a predicted or confirmed splice defect). OR Variants affecting the same non-canonical splice nucleotide as a confirmed Pathogenic splice variant with similar or worse splicing in silico prediction using SpliceAI. PS1 (Moderate) also requires the comparator to be established by this VCEP (Likely Pathogenic, with normal RNA result and PM2_supporting met).

Evidence considered

• Governing VCEP PS1 rule (InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel specification for MSH6 v2.0): a predicted missense substitution encoding the same amino acid change with a different underlying nucleotide change previously established by this VCEP as Pathogenic, OR variants affecting the same non-canonical splice nucleotide as a confirmed Pathogenic splice variant with similar or worse SpliceAI prediction.
• MSH6 codon 673 is CCG (residue c.2017 is the first base of codon 673; protein NP_000170.1:p.(Pro673Ala)). The only first-base change that yields alanine is C>G, i.e. c.2017C>G is the unique nucleotide change encoding p.Pro673Ala.
• The HCI prior table for MSH6 enumerates six codon-673 substitutions: c.2017C>A p.P673T, c.2017C>G p.P673A, c.2017C>T p.P673S, c.2018C>A p.P673Q, c.2018C>G p.P673R, c.2018C>T p.P673L. Only c.2017C>G encodes alanine; the remainder encode Thr, Ser, Gln, Arg and Leu.
• ClinVar query MSH6[gene] AND "Pro673"[all] returned five records: p.Pro673Thr (Uncertain significance), p.Pro673Ser (Uncertain significance), p.Pro673Ala (this variant, conflicting classifications), p.Pro673Leu (Uncertain significance), and the synonymous p.Pro673= (Benign/Likely benign). No codon-673 variant is classified Pathogenic or Likely pathogenic.
• The VCEP pilot-variant curation table contains MSH6 entries only for c.107C>T, c.3482CTG[1], c.1296T>G, c.1282A>G, c.1723G>T, c.663A>C and c.884A>G; there is no MSH6 codon-673 entry, so no VCEP-established comparator at this residue exists.
• ClinVar record for this variant (VCV000410451) has 8 submissions and 0 expert-panel submissions, so no VCEP classification of this or a sibling codon-673 variant is available from ClinVar either.
• SpliceAI max delta for NM_000179.3:c.2017C>G is 0.014 with no acceptor/donor gain or loss, and the variant lies in exon 4, so the splice-nucleotide branch of PS1 does not apply.

Sources

Reasoning

Governing framework: the InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel specification for MSH6 v2.0 (CSPEC doc 1578294885), which replaces generic PS2 with a de novo points system rather than a single all-or-nothing rule. The rule assigns points per proband: a proband with a de novo variant and both maternity and paternity confirmed in a case with an MMR-deficient LS-spectrum tumor = 2 points; the same with an LS-spectrum tumor but no MSI/IHC/methylation data = 1 point; a proband with an assumed de novo variant and maternity and/or paternity unconfirmed in a case with an LS-spectrum tumor (no tumor data) = 0.5 points. Points are summed across probands and mapped to strength: >=4 points Very Strong, 2-3.5 Strong, 1-1.5 Moderate, 0.5 Supporting. PS2 therefore requires at least one reported de novo occurrence in a proband, with some level of parental data. No such observation exists for NM_000179.3:c.2017C>G (p.Pro673Ala). ClinVar variation 410451 comprises 8 single-submitter germline submissions (7 Uncertain significance, 1 Likely benign; 0 expert panels; SCV000580255 Ambry, SCV000685248 Color, SCV001788638 GeneDx, SCV004195618 Baylor, SCV004838292 All of Us, SCV005917056 Sinai Health, SCV002535691 Sema4, SCV000551135 Labcorp): none reports a de novo occurrence, parental or trio testing, maternity/paternity confirmation, or an assumed-de novo scenario, and the criterion signals extracted from these submissions are limited to PP3, PS3 and PP4 - no submitter applies PS2. None of the seven fetched full texts mentions c.2017C>G or p.(P673A) or reports a de novo occurrence. OncoKB returns no variant-specific curated evidence, COSMIC has no somatic record, Hotspots finds no significant hotspot, and the population sources (gnomAD v2.1, gnomAD v4.1) contain no case-level information. No proband-level clinical record of any kind (pedigree, parental genotypes, tumor phenotype) accompanies this variant, so the de novo point score is 0 points, below even the 0.5-point Supporting tier. Generic ACMG/AMP PS2 (de novo with both maternity and paternity confirmed in a patient with the disease and no family history; PMID:25741868) is likewise unmet.

Rule

InSiGHT MSH6 v2.0 PS2 (de novo points, summed per proband): 2 points = de novo with both maternity and paternity confirmed + MMR-deficient LS-spectrum tumor; 1 point = de novo with both confirmed + LS-spectrum tumor (no MSI/IHC/methylation data); 0.5 points = assumed de novo with maternity and/or paternity unconfirmed + LS-spectrum tumor (no tumor data). Strength: >=4 points Very Strong, 2-3.5 Strong, 1-1.5 Moderate, 0.5 Supporting. Scored points for this variant: 0.

Evidence considered

• InSiGHT MSH6 v2.0 (CSPEC doc 1578294885) lists PS2 'Applicable' with instructions: de novo with both maternity and paternity confirmed plus MMR-deficient LS-spectrum tumor = 2 points per proband; de novo with both confirmed plus LS-spectrum tumor without MSI/IHC/methylation data = 1 point per proband; assumed de novo with maternity and/or paternity unconfirmed plus LS-spectrum tumor without tumor data = 0.5 points per proband; strength by summed points (Very Strong >=4, Strong 2-3.5, Moderate 1-1.5, Supporting 0.5).
• ClinVar variation 410451: 8 germline submissions (7 Uncertain significance, 1 Likely benign), 0 expert-panel submissions, review status 'criteria provided, single submitter'; no submission records a de novo occurrence, parental or trio testing, maternity/paternity confirmation, or family testing, and the submission criterion-signal audit yields only PP3, PS3 and PP4.
• Screened the seven fetched full texts (PMID:25741868, 25394175, 27854360, 25711197, 26324357, 28492532, 30267214) for the exact variant and for any de novo or parental-testing report: none mentions c.2017C>G or p.(P673A), and none reports a de novo occurrence.
• Rosenthal et al. 2018 (PMID:30267214), the only publication cited by a submitter that reports the variant at cohort level, analysed only potentially disruptive variants (stop gain, frameshift, splice acceptor/donor) and therefore contains no de novo or family observation for this missense change.
• Other retrieved sources carry no case-level information: OncoKB 'Unknown Oncogenic Effect' (not variant-specific), COSMIC no record, Cancer Hotspots no significant hotspot, gnomAD v2.1/v4.1 population counts only.
• Generic ACMG/AMP PS2 (PMID:25741868) requires a de novo variant with both maternity and paternity confirmed in a patient with the disease and no family history; neither a confirmed nor an assumed de novo occurrence is reported for this variant.
• Net de novo score for this variant: 0 points (0 < 0.5-point Supporting minimum).

Still missing

• No proband, trio, parental-testing or tumor-phenotype data accompany this variant in the case record, so no de novo points (0.5, 1 or 2 per proband) can be assigned even if a de novo occurrence exists but is unreported.
• The specification's footnote defining the fine detail of point assignment across multiple probands and the 'assumed de novo' scenario was not present in the extracted rule text; this is immaterial here because zero de novo observations of any kind exist.

Sources

Reasoning

Governing framework: ClinGen InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel specification for MSH6 v2.0 (framework_mode = vcep), which supersedes generic ACMG/AMP for this gene. PS3 in that specification is satisfied only by (a) calibrated functional assays with functional odds for pathogenicity >18.7 (PS3_Strong), >4.3 and <=18.7 (PS3_Moderate) or >2.08 and <=4.3 (PS3_Supporting), (b) an MMR function defect documented following the MMR functional assay flowchart (PS3_Moderate), or (c) monoallelic expression with complete loss of the variant allele (<10% of wild-type in cDNA without puromycin) analysed with and without NMD block (PS3_Moderate). The variant under review is a missense substitution, NM_000179.3:c.2017C>G, NP_000170.1:p.(Pro673Ala). No functional evidence meeting any of those three routes exists for this variant in the case record. No calibrated functional assay result (mammalian/cell-free MMR repair activity, protein expression or stability, subcellular localisation, or mRNA/splicing assay) is reported for c.2017C>G in any source; SpliceAI predicts no splice impact (max delta 0.014), so the variant also cannot enter the flowchart via a predicted splicing defect, and no output of the flowchart exists. OncoKB reports no variant-specific reviewed functional evidence (gene-level Lynch syndrome context only). Two independent clinical-laboratory ClinVar submissions, All of Us Research Program (SCV004838292) and Color Diagnostics (SCV000685248), state in their PS3-related text that 'to our knowledge, functional studies have not been reported for this variant'. All seven full-text publications retrieved for this case were screened and none mentions c.2017C>G or p.(P673A), and none is a functional-assay study of MSH6 (paper_extracts.json). ClinVar's aggregate classification is deliberately not used as evidence for this criterion. The local HCI prior entry for p.P673A (prior probability 0.0159, from the MAPP/PP2 prior column of the LOVD-derived HCI prior table) is an in-silico prior that this VCEP assigns to PP3/BP4 only; it is not an odds-of-pathogenicity measurement from a calibrated functional assay and therefore cannot be used to satisfy PS3, and its low value does not create PS3 evidence in either direction. PS3 is not met at any strength because the variant-specific functional data the criterion requires were never generated.

Rule

PS3 (InSiGHT MSH6 v2.0): calibrated functional assay with functional odds for pathogenicity >18.7 (Strong); >4.3 and <=18.7, OR MMR function defect per functional assay flowchart, OR complete loss of the variant allele (<10% of wild-type, cDNA without puromycin) (Moderate); >2.08 and <=4.3 (Supporting).

Evidence considered

• ClinGen InSiGHT MSH6 v2.0 specification (cspec, doc 1578294885) PS3 rule text: 'Calibrated functional assays with functional odds forpathogenicity > 18.7' (Strong); 'Calibrated functional assays with functional odds for pathogenicity >4.3 and <= 18.7 OR MMR function defect following functional assay flowchart OR Variants with monoallelic expression: complete loss of expression (<10% of wild-type in cDNA without puromycin) of the variant allele' (Moderate); 'Calibrated functional odds for pathogenicity >2.08 and <= 4.3' (Supporting).
• Specification instruction for PS3: 'Refer to file Functional assay SVI documentation (MMR genes) for calibrated functional assays' - i.e. the functional evidence PS3 cites must come from an approved calibrated assay, not from in-silico predictors or population data.
• VCEP functional-assay documentation ('MMR assays' sheet) lists which assays are approved and calibrated (e.g. Drost et al. cell-free CIMRA MMR-activity assay, MSH6 normal-readout threshold >=100% and abnormal <18%; MSH2 saturation-genome-editing assay; MLH1 cell-based assay; MSH6 protein-expression, subcellular-localisation, cDNA full-length-transcript and MNNG-survival assays), each mapped to proposed PS3/PS3_Moderate/PS3_Supporting strengths; it contains no result for c.2017C>G.
• MMR functional assay flowchart: for an in-frame indel or missense variant without predicted splicing impact, the next decision node is 'Calibrated functional assay conducted?'; if no calibrated assay exists the flowchart requires '2 independent assays with concordant results' (mammalian MMR activity, protein expression/stability, subcellular localisation), none of which was performed here.
• OncoKB record for MSH6 P673A: 'OncoKB did not identify variant-specific reviewed functional evidence for this variant; gene-level curated context is available for reviewer follow-up.'
• ClinVar VCV000410451 submission text captured in the case record: All of Us Research Program (SCV004838292) and Color Diagnostics (SCV000685248) both report, as their PS3-related statement, 'to our knowledge, functional studies have not been reported for this variant'. No ClinVar submitter reports a functional assay result for this variant.
• Variant identity and consequence used throughout: NM_000179.3:c.2017C>G, NC_000002.11:g.48027139C>G, NP_000170.1:p.(Pro673Ala) - a missense change, so no loss-of-function or splice-assay route to PS3 applies automatically.
• SpliceAI (max delta 0.014; DS_AG 0.014, DS_AL 0.0, DS_DG 0.008, DS_DL 0.0) predicts no significant splice impact, so the variant cannot reach PS3 through the predicted-splice-defect route and must be evaluated functionally instead.
• Local HCI prior table (LOVD-derived MSH6 priors, row MSH6_04411) gives c.2017C>G / p.P673A a Custom_PP2_score of 0.409, MAPP score 4.34 and MAPP/PP2 prior of 0.0159; the governing specification and the local file index assign this HCI prior to PP3/BP4 thresholds (>0.81 PP3_Moderate, >0.68 and <=0.81 PP3_Supporting, <0.11 BP4_Supporting), not to PS3/BS3 functional odds.
• No DNA mismatch repair functional-assay publication was retrieved or extracted for this case: the seven full texts available (PMID:25394175, 25711197, 25741868, 26324357, 27854360, 28492532, 30267214) are guidelines, policy statements, frameworks or an exome burden study, and none reports an assay on this variant.

Still missing

• No calibrated functional assay (e.g. cell-free/mammalian MMR repair-activity or protein-expression assay as listed in the VCEP 'Functional assay SVI documentation (MMR genes)') has been performed on NM_000179.3:c.2017C>G, so no functional odds for pathogenicity can be calculated and PS3 cannot be upgraded.
• Whether this substitution appears in any published MSH6 functional-assay calibration dataset was not verifiable from the case record: no functional-assay publication (e.g. an MSH6 CIMRA/functional classification study) was fetched or extracted for this case, and none of the seven retrieved full texts reports functional data for the variant.
• No patient-derived RNA or protein data (allelic expression, full-length transcript +/- NMD block) are available, so the monoallelic-loss branch of PS3_Moderate cannot be evaluated.

Sources

Reasoning

Governing framework: the InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel specification for MSH6 v2.0 (CSPEC doc 1578294885), where PM3 is applicable only as a points-based co-occurrence rule. The rule requires a patient whose clinical features are consistent with constitutional mismatch repair deficiency (CMMRD) as per the Aronson et al. 2022 CMMRD criteria (local VCEP file Table-for-CMMRD-diagnosis.pdf), in whom a second known pathogenic/likely pathogenic sequence variant in MSH6 is present; the second variant scores 1.0 point when confirmed in trans (phase established by testing parents or offspring) and 0.5 points when phase is unknown, and points are summed across cases (>=4 points Very Strong, 2-3.5 Strong, 1-1.5 Moderate, =0.5 Supporting). No qualifying allelic observation exists for NM_000179.3:c.2017C>G (p.Pro673Ala). All eight ClinVar submissions (ClinVarID 410451, SCV000551135, SCV000580255, SCV000685248, SCV001788638, SCV002535691, SCV004195618, SCV004838292, SCV005917056) report the variant as a heterozygous germline finding in Lynch-syndrome-spectrum or hereditary cancer settings; none reports a second MSH6 variant, a phase determination, or a CMMRD-consistent phenotype, and no submission cites or asserts PM3 - the criterion signals extracted from these submissions are limited to PP3, PS3 and PP4. No affected proband carrying two MSH6 alleles in trans is described anywhere in the retrieved record. None of the seven fetched full texts names c.2017C>G or p.(P673A), so no published co-occurrence or phase data are available; the only cohort-level use of the variant is Rosenthal et al. 2018, where it is reported in 0/165 colorectal cancer/polyposis cases and 1/2512 controls from a healthy population database - a heterozygous case-control observation with no second MSH6 variant, no phase data and no CMMRD phenotype. Population data are likewise consistent with the absence of biallelic occurrence: gnomAD v4.1 records 7/1,613,704 alleles with zero homozygotes (AF 4.34e-06; grpmax FAF 1.83e-06). Points scored = 0, so PM3 is not met at any strength.

Rule

InSiGHT MSH6 v2.0 PM3: co-occurrence with a known pathogenic/likely pathogenic MSH6 sequence variant in a patient with CMMRD-consistent clinical features (Aronson et al. 2022); 1.0 point in trans, 0.5 points if phase unknown, summed across cases (Very Strong >=4, Strong 2-3.5, Moderate 1-1.5, Supporting =0.5). Scored points for this variant: 0.

Evidence considered

• InSiGHT MSH6 v2.0 (CSPEC doc 1578294885) marks PM3 'Applicable' with the instruction: co-occurrence with a known pathogenic/likely pathogenic sequence variant in the same gene in a patient with clinical features consistent with CMMRD as per Aronson et al. 2022; pathogenic/likely pathogenic in trans = 1.0 point, phase unknown = 0.5 points; sum all cases to determine strength (Very Strong >=4, Strong 2-3.5, Moderate 1-1.5, Supporting =0.5 points).
• The CMMRD diagnosis table supplied with the MSH6 VCEP (Table-for-CMMRD-diagnosis.pdf) requires >=3 points from CMMRD-defining malignancies/premalignancies and optional additional features; no patient in this record has any reported CMMRD feature, so the phenotype gate of the InSiGHT PM3 rule is not satisfied by any known carrier.
• ClinVar variation 410451: 8 germline submissions (7 Uncertain significance, 1 Likely benign), conflicting classifications, 0 expert-panel submissions; no submitter describes a second MSH6 variant, phase, or a biallelic/CMMRD case, and no submission applies PM3.
• ClinVar criterion-signal audit of the exact-matched record yields only PP3 (GeneDx), PS3+PP3 (All of Us; Color), PS3+PP3+PP4 (Ambry) - no allelic (trans/cis) criterion is raised by any submitter.
• Rosenthal et al. 2018 (PMID:30267214) - cited by the Ambry submission - detected this variant in 0/165 colorectal cancer and/or polyposis patients and in 1/2512 controls from a healthy population database; the study analysed only potentially disruptive variants (stop gain, frameshift, splice acceptor/donor, MAF <0.005), so no trans/cis or CMMRD-setting observation for this missense change exists in it.
• gnomAD v4.1: 7/1,613,704 alleles (AF 4.33785e-06; MAF 0.00043%), 0 homozygotes, grpmax FAF 1.83e-06; gnomAD v2.1: 4/282,320 alleles (AF 1.41683e-05), 0 homozygotes - no homozygous (biallelic) carrier has been observed in these reference datasets.
• Screened the seven fetched full texts (PMID:25741868, 25394175, 27854360, 25711197, 26324357, 28492532, 30267214) for the exact variant and for any co-occurrence/phase report: none mentions c.2017C>G or p.(P673A), and none reports a patient with a second MSH6 variant.

Still missing

• No patient-level data (tumor MSI/IHC, family history, parental or offspring testing to establish phase) accompany this variant, so a co-occurrence in trans with a second pathogenic MSH6 variant in a CMMRD-consistent proband could still exist but cannot be scored.
• Rosenthal et al. 2018 (PMID:30267214) reports the variant in a colorectal cancer/polyp cohort and controls but without zygosity of any second MSH6 variant or phenotype detail, so it cannot substitute for a CMMRD-setting co-occurrence observation.

Sources

Reasoning

The MSH6 VCEP PM5 rule requires a missense change at an amino acid residue where a different missense change was classified by this VCEP as Pathogenic (PM5) or Likely Pathogenic (PM5_Supporting) on the protein level and not due to aberrant splicing, and it explicitly permits PM5/PM5_Supporting only when PP3 is supporting for the missense change being assessed. Both limbs fail. (1) No VCEP-classified comparator exists at residue 673: the VCEP pilot-variant curation contains no MSH6 codon-673 entry, and the complete ClinVar set of MSH6 codon-673 missense variants (p.Pro673Thr, p.Pro673Ser, p.Pro673Ala, p.Pro673Leu) is uniformly Uncertain significance, with zero expert-panel submissions among the 8 submissions for this variant. (2) The PP3 gate is not satisfied: under the VCEP missense in-silico rule PP3 requires an HCI prior probability for pathogenicity >0.68 (PP3_Supporting >0.68 and <=0.81; PP3_Moderate >0.81), whereas this variant's HCI prior is 0.0159, which sits in the BP4_supporting range (<0.11). PP3 is therefore not supporting and PM5 cannot be applied, at Moderate or Supporting strength.

Rule

MSH6 VCEP v2.0 PM5 (Moderate): missense change at an amino acid residue where a different missense change was classified by this VCEP as Pathogenic on the protein level and not due to aberrant splicing; only use PM5 if PP3 is supporting for the missense change. PM5_Supporting: same requirement against a VCEP-classified Likely Pathogenic missense change.

Evidence considered

• Governing VCEP PM5 rule (MSH6 v2.0): 'Missense change at an amino acid residue where a different missense change was classified by this VCEP as Pathogenic on the protein level and not due to aberrant splicing. Only use PM5 if PP3 is supporting for the missense change. Use PM5_Supporting if other variant is Likely Pathogenic due to a missense alteration.'
• This variant's local HCI prior probability of pathogenicity is 0.0159 (MAPP score 4.34, custom PP2 score 0.409), which is below the VCEP BP4_supporting cutoff of <0.11 and far below the PP3_supporting threshold of >0.68, so the VCEP's mandatory PP3 gate for PM5 is not met.
• pm5_candidates.json reports pm5_mode 'classic_same_residue_missense' with 0 same-residue candidates and 0 expert-panel P/LP candidates, and records a soft retrieval failure (HTTP Error 429: Too Many Requests).
• Manual ClinVar query MSH6[gene] AND "Pro673"[all] returned five records: c.2017C>A p.Pro673Thr (Uncertain significance), c.2017C>T p.Pro673Ser (Uncertain significance), c.2017C>G p.Pro673Ala (this variant, conflicting classifications), c.2018C>T p.Pro673Leu (Uncertain significance) and c.2019A>G p.Pro673= (Benign/Likely benign). No codon-673 missense variant is Pathogenic or Likely pathogenic, so no PM5 comparator exists.
• ClinVar VCV000410451 lists 8 submissions (7 usable) with 0 expert-panel submissions; no VCEP-classified codon-673 comparator is available from ClinVar.
• The VCEP pilot-variant curation table contains no MSH6 codon-673 entry, so this VCEP has not established any Pathogenic or Likely Pathogenic allele at residue 673.

Still missing

• The pipeline's ClinVar same-residue comparator harvest (pm5_candidates.json) errored with HTTP 429 and returned no candidates; this was closed manually with a live ClinVar query, which found only Uncertain-significance codon-673 missense variants (p.Pro673Thr, p.Pro673Ser, p.Pro673Leu) plus the synonymous p.Pro673=.
• The VCEP rule requires the comparator to have been classified by this VCEP itself; no VCEP codon-673 classification exists in any case source, so a non-VCEP ClinVar classification could not satisfy the rule regardless.

Sources

Reasoning

Governing framework: the InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel specification for MSH6 v2.0 (CSPEC doc 1578294885), which grades PP1 solely from pedigree co-segregation expressed as a combined Bayes Likelihood Ratio: Supporting when >2.08 and <=4.3, Moderate when >4.3 and <=18.7, Strong when >18.7 in >=2 families, with results from multiple pedigrees combined multiplicatively and the recommended tool being COOL v3 (overall Bayes Factor used to determine strength). PP1 therefore requires at least one informative pedigree with genotype data from affected and unaffected family members. No pedigree and no family-member genotype data exist for NM_000179.3:c.2017C>G (p.Pro673Ala) in any consulted source: this case contains no proband-level record, ClinVar variation 410451 carries no submission-level statement about family testing or co-segregation (criterion signals across the 8 submissions are limited to PP3, PS3 and PP4), and none of the seven fetched full texts names c.2017C>G or p.(P673A) or reports family data. The only cohort-level use of the variant, Rosenthal et al. 2018 (PMID:30267214), analysed potentially disruptive variants only and reports no pedigrees or segregation for this change. With no genotyped informative meiosis, no likelihood ratio can be computed, so PP1 is unmet at every strength; generic PP1 (co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease; PMID:25741868) is likewise unmet. The assessment is confidence-moderate because it rests on absence of pedigree data rather than on a computed ratio.

Rule

InSiGHT MSH6 v2.0 PP1 (combined pedigree Bayes Likelihood Ratio, pedigrees combined multiplicatively; COOL v3 recommended): Supporting >2.08 and <=4.3; Moderate >4.3 and <=18.7; Strong >18.7 in >=2 families. Likelihood ratio for this variant: not calculable (no pedigree data); PP1 unmet at all strengths.

Evidence considered

• InSiGHT MSH6 v2.0 (CSPEC doc 1578294885) lists PP1 'Applicable' with the instructions: for multiple pedigrees results are combined by multiplying them together; the recommended segregation analysis tool is COOL v3, and the 'overall Bayes Factor' is used to determine evidence strength. Rule thresholds: Supporting >2.08 and <=4.3, Moderate >4.3 and <=18.7, Strong >18.7 in >=2 families.
• ClinVar variation 410451: 8 germline submissions (7 Uncertain significance, 1 Likely benign), 0 expert panels; no submission states co-segregation, family testing, or pedigree information, and no submitter applies PP1 (criterion signals found: PP3, PS3, PP4 only).
• Screened the seven fetched full texts for the exact variant and for any family or co-segregation report: none mentions c.2017C>G or p.(P673A) and none provides pedigree or segregation data.
• Rosenthal et al. 2018 (PMID:30267214), the paper cited by the Ambry submission, reports the variant in a colorectal cancer/polyp cohort versus controls but analysed only potentially disruptive variants and presents no pedigree, family genotypes or co-segregation result for this missense change.
• No pedigree or family-member genotype data are present anywhere in this case record, so a combined Bayes likelihood ratio cannot be computed and the PP1 thresholds cannot be applied.
• Generic ACMG/AMP PP1 (PMID:25741868): co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease; no affected carrier relatives are described for this variant, so it is unmet in the generic reading as well.

Still missing

• No pedigree, family-member genotypes, or co-segregation results accompany this variant, so no Bayes likelihood ratio can be derived and no PP1 strength (Supporting >2.08, Moderate >4.3, Strong >18.7 in >=2 families) can be evaluated; a published or in-house pedigree with genotyped relatives carrying this variant would be needed.
• Assessing PP1 would require age-of-onset and tumour-phenotype information for affected relatives, none of which is present in this record.

Sources

Reasoning

Scope first: NM_000179.3:c.2017C>G encodes NP_000170.1:p.(Pro673Ala), a missense substitution in MSH6 exon 4, so PP3 is evaluated on the protein-impact path. For MSH6 the governing ClinGen InSiGHT Hereditary Colorectal Cancer/Polyposis VCEP specification (v2.0, cspec doc 1578294885) replaces generic REVEL calibration for missense variants with the HCI prior probability of pathogenicity: >0.81 = PP3_Moderate, >0.68 and <=0.81 = PP3_Supporting. The declared local VCEP lookup HCI-PRIORS-MSH6.txt was searched for every notation of this variant (c.2017C>G, p.Pro673Ala, p.P673A); the exact row exists at line 4413 (dbid MSH6_04411, reference {PMID22949387:Thompson et al., 2013}) and gives prior probability 0.0159 (with MAPP 4.34 and custom PP2 score 0.409 on the same row). 0.0159 is far below the >0.68 threshold for PP3_Supporting, so PP3 is not met at any strength. The VCEP's alternative PP3 clause (predicted splice defect for non-canonical splicing nucleotides with SpliceAI delta score >=0.2, Walker et al 2023) is out of scope here: c.2017 is an exonic coding position of a missense variant, not a non-canonical splice nucleotide, and in any case the SpliceAI max delta was 0.014, so no splice prediction was used as protein-level PP3 support. The generic missense path was checked only to confirm it does not add support: REVEL was 0.447, inside the indeterminate 0.29-0.644 zone (Pejaver et al 2022, PMID:36413997), so the generic path would also give no PP3; per the VCEP-precedence rule the prior-based MSH6 rule is the governing answer.

Rule

MSH6 VCEP v2.0 PP3 (missense): PP3_Moderate if HCI prior probability of pathogenicity >0.81; PP3_Supporting if >0.68 and <=0.81. Observed HCI prior for c.2017C>G (p.P673A) = 0.0159 -> not met.

Evidence considered

• Consequence type fixes the path: c.2017C>G is a missense substitution (NP_000170.1:p.(Pro673Ala), exon 4), so PP3 was evaluated on the missense protein-impact path only; the SpliceAI splice path was excluded by scope, not because evidence was missing.
• MSH6 VCEP v2.0 PP3 rules (ClinGen InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel Specification, cspec doc 1578294885): missense variant with HCI prior probability of pathogenicity >0.81 = PP3_Moderate; >0.68 and <=0.81 = PP3_Supporting (otherwise SpliceAI delta >=0.2 for non-canonical splice nucleotides).
• Declared local VCEP file HCI-PRIORS-MSH6.txt (also mirrored in the case evidence as hci_prior) contains the exact variant: line 4413 = exon 4, c.2017C>G, p.P673A, prior probability 0.0159, MAPP 4.34, custom PP2 0.409, dbid MSH6_04411. Searching all three notations (c.2017C>G, p.Pro673Ala, p.P673A) returned only this single row group at codon 673; p.Pro673Ala appears nowhere in the file.
• Column identification on that row: the HCI prior probability of pathogenicity is the '<...>_Prior' value 0.0159; the 0.409 value is a custom PP2 score and the 4.34 value is the MAPP score, neither of which is the prior. Adjacent rows in the same table behave consistently with the VCEP thresholds (c.2017C>T/p.P673S prior 0.1023 just below 0.11; c.2017C>A/p.P673T prior 0.1190 just above 0.11), supporting that the prior column is the intended lookup value.
• HCI prior 0.0159 fails the >0.81 (moderate) and >0.68 (supporting) PP3 thresholds, so PP3 is not met; no PP3 strength is assigned.
• SpliceAI lookup for NM_000179.3:c.2017C>G returned max delta 0.014 (DS_AG 0.014, DS_DG 0.008, DS_AL/DS_DL 0.0), consistent with no splice impact, but this is NOT used as PP3 support for a missense variant and the VCEP's SpliceAI-based PP3 clause applies only to non-canonical splice nucleotides.
• Generic calibration cross-check (not governing): REVEL 0.447 is in the indeterminate range between the BP4 supporting cutoff 0.29 and the PP3 supporting cutoff 0.644 (Pejaver et al 2022, Am J Hum Genet, PMID:36413997), so the generic missense path would give no PP3 either. BayesDel was not used; no validated ACMG-strength calibration exists for it.
• Seven case publications were screened through their pre-computed extractions (PMID:25741868, PMID:25394175, PMID:27854360, PMID:30267214, PMID:25711197, PMID:26324357, PMID:28492532); none reports variant-specific computational or in-silico evidence for c.2017C>G.

Sources

Reasoning

Governing rule (MSH6 VCEP v2.0, PP4 applicable, default strength Supporting; count-based tiers, so evidence strengthens as the tumor count rises): Strong = 'reach or exceed 3 independent CRC/Endometrial MSI-H tumors in 2 or more families using a standard panel of 5-10 markers or tumor genome and/or loss of MMR protein expression consistent with the variant location'; Moderate = 2 such tumors; Supporting = 1 such CRC/endometrial MSI-H tumor, or tumor genome, and/or MMR protein loss consistent with the variant location (for an MSH6 variant the consistent pattern is loss of MSH6, alone or with MSH2). Explicit comparison with this case's data: qualifying MSI-H / MSH6-consistent-IHC-loss tumors reported for c.2017C>G = 0, so '0 >= 1?' = no, and the criterion is not_met. There is no tumor phenotype evidence of any kind for this variant: the eight ClinVar submissions for VCV000410451 (none from an expert panel) carry variant-level classifications only, with no MSI or IHC result; OncoKB returns 'Unknown Oncogenic Effect' with no variant-specific curated evidence; and the only paper triaged toward PP4 (PMID:30267214) is an exome burden study restricted to potentially disruptive variants that neither names c.2017C>G nor reports tumor MSI/IHC status. PP4 would require at least one CRC or endometrial MSI-H tumor with MSH6-consistent protein loss before it could be reconsidered at Supporting strength.

Rule

PP4 (MSH6 VCEP v2.0). Supporting: 1 CRC/Endometrial MSI-H tumor using a standard panel of 5-10 markers or tumor genome and/or loss of MMR protein expression consistent with the variant location. Moderate: 2 independent CRC/Endometrial MSI-H tumors as above. Strong: at least 3 independent CRC/Endometrial MSI-H tumors in at least 2 families as above.

Evidence considered

• MSH6 VCEP v2.0 PP4 is applicable with count-based tiers: Supporting = 1, Moderate = 2 and Strong = 3 or more independent CRC/endometrial MSI-H tumors (standard 5-10 marker panel or tumor genome) and/or loss of MMR protein expression consistent with the variant location; MSH6 loss is the IHC pattern consistent with an MSH6 variant.
• Explicit comparison for this variant: the number of qualifying MSI-H / MSH6-consistent-IHC-loss tumors reported is 0, so 0 >= 1? No - PP4 is not met at any tier.
• ClinVar VCV000410451 (NM_000179.3(MSH6):c.2017C>G, p.Pro673Ala) has 8 submissions, none from an expert panel, listing conditions such as Lynch syndrome, hereditary nonpolyposis colorectal neoplasms and endometrial carcinoma, but no tumor MSI or MMR IHC data for any carrier.
• PMID:30267214 was the only paper triaged toward PP4; its exome case-control analysis was limited to potentially disruptive variants and it neither names c.2017C>G nor reports tumor MSI/IHC phenotype for this variant.
• OncoKB returned 'Unknown Oncogenic Effect' for MSH6 P673A with no variant-specific curated evidence, and the hotspot screen found this residue is not a significant cancer hotspot; neither provides tumor MSI/IHC phenotype data.

Still missing

• No tumor MSI (standard 5-10 marker panel or tumor genome) and no MMR protein IHC result is available for any carrier of c.2017C>G; at least one CRC/endometrial MSI-H tumor with MSH6-consistent IHC would be needed for PP4_Supporting.

Sources

Reasoning

Governing framework: ClinGen InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel specification for MSH6 v2.0 (cspec, doc 1578294885), which overrides the generic ACMG stand-alone threshold of 5%. The only applicable BA1 rule is stand-alone at a gnomAD v4 Grpmax filtering allele frequency >= 0.0022 (0.22%), with the qualifier that the variant is excluded as a founder pathogenic variant. The variant's gnomAD v4.1 Grpmax filtering AF is 1.83e-06 (0.000183%), roughly 1,200-fold below the 0.0022 threshold, and the total gnomAD v4.1 frequency is 4.33785e-06 (7/1,613,704 alleles, 0 homozygotes). No ancestry group comes near the threshold: the highest subpopulation frequency in v4.1 is 1.66783e-05 in Admixed American (1/59,958 alleles), and the highest ancestry-restricted Grpmax seen anywhere in the available data is 2.849e-05 (European non-Finnish, 5/64,724 alleles in the v3.1 non-cancer genome subset). gnomAD v2.1 (Grpmax FAF 2.241e-05; total AF 1.41683e-05, 4/282,320 alleles, 0 homozygotes), gnomAD v2.1 non-cancer exomes (8.45895e-06, 2/236,436 alleles) and gnomAD v3.1 non-cancer genomes (total AF 3.38804e-05) are all three to four orders of magnitude below 0.0022, and the variant is absent from gnomAD-Canada v1.0. BA1 is therefore not met under the VCEP-specified source or under any alternative gnomAD subset, and because no threshold is approached the founder-pathogenic-variant exclusion clause is moot. MSH6-associated Lynch syndrome is a dominant but incompletely penetrant condition, which is already reflected in the VCEP's unusually permissive 0.22% stand-alone cutoff; an allele frequency near 0.0002% is nowhere close to being incompatible with pathogenicity for this gene.

Rule

GnomAD v4 Grpmax filtering allele frequency >= 0.0022 (0.22%) and variant is excluded as founder pathogenic variant.

Evidence considered

• ClinGen InSiGHT MSH6 v2.0 (cspec, doc 1578294885) defines BA1 stand-alone solely as: 'GnomAD v4 Grpmax filtering allele frequency >= 0.0022 (0.22%) and variant is excluded as founder pathogenic variant.' No other strength level is applicable for BA1 in this specification.
• gnomAD v4.1 (GRCh38, chr2-47800000-C-G): total AF 4.337846e-06 (7/1,613,704 alleles), Grpmax filtering AF 1.83e-06, 0 homozygotes; exome component 2/1,461,876 (1.368105e-06), genome component 5/151,828 (3.29320e-05).
• gnomAD v4.1 highest subpopulation frequency is Admixed American, 1.667834e-05 (1/59,958 alleles), 0 homozygotes; the highest ancestry-restricted Grpmax visible in the v4 genome component is 2.849e-05, i.e. European (non-Finnish).
• gnomAD v2.1 (GRCh37, 2-48027139-C-G): total AF 1.416832e-05 (4/282,320 alleles), Grpmax filtering AF 2.241e-05, 0 homozygotes; highest subpopulation Admixed American 2.822786e-05 (1/35,426 alleles), 0 homozygotes.
• gnomAD v2.1 non-cancer exomes: AF 8.458949e-06 (2/236,436 alleles), 0 homozygotes. gnomAD v3.1 non-cancer genomes: total AF 3.388039e-05 (5/147,578 alleles), Grpmax filtering AF 2.849e-05, highest subpopulation European (non-Finnish) 7.725110e-05 (5/64,724 alleles), 0 homozygotes.
• gnomAD-Canada v1.0 (HostSeq genomes) was queried for the GRCh38 position and the variant is reported absent.
• Threshold-artifact note: the gnomAD v4 genome component is the v3.1 genome call set, and restricting to non-cancer samples shrinks the denominator while leaving the 5 observed alleles, which inflates the subset frequency (v3.1 non-cancer Grpmax 2.849e-05 versus v4 all-comers Grpmax 1.83e-06). Neither figure is within two orders of magnitude of the 0.0022 BA1 threshold, so the source choice does not change this verdict.
• Disease-prevalence/penetrance consideration: the VCEP's BA1 cutoff of 0.22% is already far below the generic 5% stand-alone guideline because MSH6 Lynch syndrome is dominantly inherited with reduced penetrance; the observed allele frequency therefore cannot be argued to exceed what the disease prevalence/penetrance could accommodate in the benign direction.

Sources

Reasoning

Governing framework: ClinGen InSiGHT MSH6 v2.0. The applicable BS1 rule is benign strong at a gnomAD v4 Grpmax filtering allele frequency >= 0.00022 and < 0.0022 (0.022-0.22%), with the same founder-pathogenic-variant exclusion qualifier as BA1. The variant's gnomAD v4.1 Grpmax filtering AF is 1.83e-06, about 120-fold below the 0.00022 lower bound of the band, and the total v4.1 frequency (4.337846e-06; 7/1,613,704 alleles; v4.1 exome-only AF 1.368105e-06) is likewise far below. The most permissive available non-cancer or exome-restricted figure, gnomAD v3.1 non-cancer genomes (total AF 3.388039e-05, Grpmax FAF 2.849e-05), is still roughly 8-fold below 0.00022, and gnomAD v2.1 (Grpmax FAF 2.241e-05), gnomAD v2.1 non-cancer exomes (8.458949e-06) and gnomAD-Canada (variant absent) provide nothing higher. Because MSH6 disease is dominant with incomplete penetrance, the VCEP has already set BS1 very low (0.022%) relative to the generic 1% SVI default; even against that lowered bar the variant's frequency is insufficient, so BS1 is not met and the founder exclusion clause is moot. No ancestry-specific frequency (highest: Admixed American 1.667834e-05 in v4.1; European non-Finnish 7.725110e-05 in the v3.1 non-cancer genome subset) reaches the threshold either.

Rule

GnomAD v4 Grpmax filtering allele frequency >= 0.00022 and < 0.0022 (0.022-0.22%) and variant is excluded as founder pathogenic variant.

Evidence considered

• ClinGen InSiGHT MSH6 v2.0 (cspec, doc 1578294885) defines BS1 strong solely as: 'GnomAD v4 Grpmax filtering allele frequency >= 0.00022 and < 0.0022 (0.022-0.22%) and variant is excluded as founder pathogenic variant.' No other strength level is applicable for BS1 in this specification.
• gnomAD v4.1 (chr2-47800000-C-G): Grpmax filtering AF 1.83e-06 and total AF 4.337846e-06 (7/1,613,704 alleles, 0 homozygotes); v4.1 exome-only AF 1.368105e-06 (2/1,461,876 alleles).
• gnomAD v2.1: total AF 1.416832e-05 (4/282,320 alleles), Grpmax filtering AF 2.241e-05, 0 homozygotes.
• gnomAD v2.1 non-cancer exomes: AF 8.458949e-06 (2/236,436 alleles). gnomAD v3.1 non-cancer genomes: total AF 3.388039e-05 (5/147,578 alleles) and Grpmax filtering AF 2.849e-05, the highest population-frequency figure available for this variant in any subset, yet still ~8-fold below the 0.00022 BS1 floor.
• gnomAD-Canada v1.0: variant absent.
• No ancestry group reaches the BS1 band: highest subpopulation AF in gnomAD v4.1 is Admixed American 1.667834e-05 (1/59,958 alleles) and the highest ancestry-restricted value anywhere is European (non-Finnish) 7.725110e-05 (5/64,724 alleles) in the v3.1 non-cancer genome subset.
• Disease-prevalence/penetrance consideration: for Lynch syndrome the penetrance of MSH6-associated colorectal/endometrial cancer is incomplete, which is why the VCEP's BS1 threshold (0.022%) is much lower than the generic 1% default; even so the variant's allele frequency is roughly 120-fold below the cutoff and cannot be used as benign population evidence.

Sources

Reasoning

Governing framework: ClinGen InSiGHT MSH6 v2.0, where BS2 is redefined from the generic 'observed in a healthy adult individual' rule to a gene-specific co-occurrence rule: co-occurrence in trans with a known pathogenic sequence variant in the same gene in a patient with colorectal cancer after age 45 (or another Lynch-spectrum cancer above its median age of onset in Lynch syndrome), in a patient with no previous or current evidence of clinical manifestations of CMMRD as per Aronson et al. 2022, with confirmation of phase by testing of parents or offspring. No such observation exists in the available evidence. The case carries no proband genotype and no phase/parental testing data for this variant; none of the seven ClinVar submissions contributes an expert-panel or criterion-level statement about co-occurrence (the record is 'criteria provided, single submitter' only); no literature source identified the variant at all; and gnomAD reports zero homozygotes for this variant in v2.1 and v4.1 as well as in the non-cancer subsets, so there is no homozygous or biallelic carrier observation of any kind, whether healthy or affected. BS2 is therefore not met. This is an evidence-absence rather than evidence-against result: if the proband is shown to carry a second pathogenic MSH6 allele in trans with confirmed phase, or a healthy homozygous adult were identified, BS2 would need to be re-evaluated, and that determination sits with the allelic/phase and segregation evidence groups rather than with population data.

Rule

Co-occurrence in trans with a known pathogenic sequence variant in the same gene in a patient with colorectal cancer after age 45 (or other LS cancer above the median age of onset for that cancer in LS), and who has no previous or current evidence of clinical manifestations of CMMRD as per Aronson et al 2022. Confirmation of phase requires testing of parents or offspring.

Evidence considered

• ClinGen InSiGHT MSH6 v2.0 (cspec, doc 1578294885) BS2 strong rule: 'Co-occurrence in trans with a known pathogenic sequence variant in the same gene in a patient with colorectal cancer after age 45 (or other LS cancer above the median age of onset for that cancer in LS), and who has no previous or current evidence of clinical manifestations of CMMRD as per Aronson et al 2022 (Refer to "Table for CMMRD diagnosis.pdf"). Confirmation of phase requires testing of parents or offspring.'
• No in-trans co-occurrence observation with phase confirmation is present anywhere in the case evidence: there is no proband genotype block, no parental/offspring testing result, and no second MSH6 pathogenic variant reported in the case files.
• gnomAD v2.1 and gnomAD v4.1 both report 0 homozygotes for this variant (v2.1: 4/282,320 alleles; v4.1: 7/1,613,704 alleles), and the gnomAD v2.1 non-cancer exome and v3.1 non-cancer genome subsets also report 0 homozygotes, so no homozygous (biallelic) carrier of any phenotype has been observed in population data.
• ClinVar record 410451 is composed of single-submitter criteria-based entries (5 clinical laboratories Uncertain significance, 1 Uncertain Significance, 1 Likely benign) with no expert-panel review and no criterion-level co-occurrence or phase statement; it therefore contributes no BS2 evidence.
• Papers screened for this variant-level curation (including the mismatch-repair/CMMRD practice guideline and the colorectal cancer candidate-gene case/control exome study) do not mention MSH6 c.2017C>G or p.(P673A) and provide no co-occurrence, phase, or CMMRD-status data for it.
• CMMRD context for the exclusion clause: biallelic mismatch repair gene mutations cause constitutional mismatch repair deficiency (childhood cancer predisposition), so an in-trans observation in an adult with later-onset colorectal cancer and no CMMRD features would be the benign-supporting configuration this VCEP rule targets; no such patient is documented here.

Still missing

• No proband-level genotype, co-occurring MSH6 variant, or parent/offspring phase-testing data are present in the case evidence, so the VCEP BS2 co-occurrence precondition could not be positively tested; the criterion is recorded as not met on the basis of absent evidence.
• No healthy homozygous adult carrier has been reported for this variant in any consulted source (gnomAD v2.1, v4.1 and both non-cancer subsets report 0 homozygotes), so the classic generic BS2 observation also does not exist.

Sources

Reasoning

Governing framework: ClinGen InSiGHT MSH6 v2.0 specification (vcep mode). BS3 there is met only by (a) calibrated functional assays with functional odds for pathogenicity <=0.05 (BS3_Strong) or >0.05 and <=0.48 (BS3_Supporting), or (b) variant-specific proficient function in protein- and mRNA-based laboratory assays as per the MMR functional assay flowchart (BS3_Supporting). The synonymous/intronic branch of BS3_Strong (no mRNA aberration in laboratory assays performed with NMD inhibition) is structurally inapplicable: c.2017C>G is a missense substitution, p.(Pro673Ala), not a synonymous or intronic variant. No calibrated functional assay has been performed for this variant, and no protein- or mRNA-based assay reports proficient (wild-type-like) function for it, so neither BS3_Strong nor BS3_Supporting is satisfied. The VCEP's approved-assay documentation describes the calibrated MMR assays and their normal/abnormal readout thresholds but contains no measurement for c.2017C>G, and the VCEP pilot-variant table does not list this variant. The one numerical value for this variant that superficially resembles a benign functional odds - the HCI prior probability 0.0159 (row MSH6_04411 of the local LOVD-derived HCI prior table) - is not functional-assay data: that table's columns are Custom_PP2_score (0.409), MAPP score (4.34) and MAPP/PP2 prior (0.0159), i.e. an in-silico MAPP/PP2 prior, and both the specification's own criteria text and the local file index assign it exclusively to the PP3/BP4 in-silico thresholds (<0.11 = BP4_Supporting). Converting it into a BS3 functional odds would misapply the framework, so it is not used here; the PP3/BP4 consequences of that prior belong to the in-silico group. ClinVar includes one Likely benign submission and ClinVar's aggregate classification was not used as BS3 evidence. Absent the variant-specific functional data BS3 requires, BS3 is not met at any strength.

Rule

BS3 (InSiGHT MSH6 v2.0): calibrated functional assay with functional odds for pathogenicity <=0.05, OR synonymous/intronic variant with no mRNA aberration under NMD inhibition (Strong); functional odds >0.05 and <=0.48, OR variant-specific proficient function in protein- and mRNA-based assays per the MMR functional assay flowchart (Supporting).

Evidence considered

• ClinGen InSiGHT MSH6 v2.0 specification BS3 rule text: 'Calibrated functional assays with functional odds for Pathogenicity <= 0.05 OR Synonymous substitutions and intronic variants with no associated mRNA aberration (either splicing or allelic imbalance) as determined by laboratory assays conducted with nonsense-mediated decay inhibition' (Strong); 'Calibrated functional assays with functional odds for Pathogenicity >0.05 & <=0.48 OR Variant-specific proficient function in protein and mRNA-based lab assays as per MMR functional assay flowchart' (Supporting).
• Specification instruction for BS3: 'Refer to file Functional assay SVI documentation (MMR genes) for calibrated functional assays'; the accompanying footnote states the flowchart is for functional assays already performed or from historic publications and that the VCEP recommends the calibrated assays for prospective testing.
• Variant type check: c.2017C>G is a missense change (NP_000170.1:p.(Pro673Ala)), so the synonymous/intronic branch of BS3_Strong does not apply and BS3 rests entirely on functional assay data.
• VCEP functional-assay documentation ('MMR assays' sheet) lists the approved calibrated assays and their readouts (e.g. cell-free CIMRA MMR activity with >=100% normal / <18% abnormal for MSH6; mammalian MMR activity complementation; protein expression with >=75% relative expression considered wild-type-like; subcellular localisation; cDNA full-length transcript +/- puromycin) but reports no measurement for c.2017C>G.
• MMR functional assay flowchart: a missense variant without predicted splicing impact that has no calibrated functional assay must be assessed by two independent concordant assays (mammalian MMR activity, protein expression/stability, subcellular localisation); no such assay was performed, so proficient function cannot be established.
• Local HCI prior table (LOVD-derived MSH6 priors, row MSH6_04411): columns are Variant/Exon, Variant/DNA, Variant/RNA, Variant/Protein, Custom_PP2_score, MAPP_score, MAPP/PP2_Prior - for c.2017C>G the values are 0.409 / 4.34 / 0.0159, i.e. an in-silico MAPP/PP2 prior; the local file index assigns this table to PP3/BP4 thresholds only (<0.11 = BP4_Supporting), so the 0.0159 value is not a calibrated functional odds and was not used for BS3.
• SpliceAI for this variant (max delta 0.014) indicates no splice impact, so there is no mRNA/splicing aberration to count either for or against BS3, and no laboratory mRNA assay (with NMD inhibition) was performed.
• OncoKB: no variant-specific reviewed functional evidence for MSH6 P673A; nothing in OncoKB reports proficient or deficient MMR function for this variant.
• No full-text publication retrieved for this case (PMID:25394175, 25711197, 25741868, 26324357, 27854360, 28492532, 30267214) mentions c.2017C>G or p.(P673A) or reports a functional assay on it.
• The ClinVar record contains one Likely benign submission (Labcorp, SCV000551135, citing PMID:28492532 only) but provides no functional assay documentation; ClinVar's classification was not used as evidence for BS3.

Still missing

• No calibrated functional assay result exists for NM_000179.3:c.2017C>G, so neither BS3_Strong (odds <=0.05) nor BS3_Supporting (odds >0.05 and <=0.48) can be calculated.
• No protein-based (MMR activity, protein expression/stability, localisation) or mRNA-based (full-length transcript +/- NMD inhibition) assay has been performed on this variant, so the flowchart-based proficient-function branch of BS3_Supporting is unevaluable.
• The in-silico HCI/MAPP prior for this variant (0.0159) points toward the benign end of the PP3/BP4 scale; if the VCEP's PP3/BP4 assessment is confirmed, that is a BP4-type (not BS3) conclusion and should not be duplicated as functional evidence.

Sources

Reasoning

Governing framework: the InSiGHT Hereditary Colorectal Cancer/Polyposis Expert Panel specification for MSH6 v2.0 (CSPEC doc 1578294885), where BS4 is graded from lack of co-segregation expressed as a combined Bayes Likelihood Ratio: Strong when <0.05, Supporting when >0.05 and <=0.48, with multiple pedigrees combined multiplicatively and COOL v3 recommended. BS4 therefore requires an informative pedigree in which the variant demonstrably fails to track with disease. No such pedigree exists for NM_000179.3:c.2017C>G (p.Pro673Ala): this case contains no proband, pedigree, or family-genotype record at all, ClinVar variation 410451 contains no non-segregation statement from any of its 8 single-submitter germline submissions (criterion signals limited to PP3, PS3 and PP4), and none of the seven fetched full texts names the variant or reports family data. Under the VCEP's own method, pedigrees with uninformative meioses yield a likelihood ratio of approximately 1, so the absence of segregation data is not evidence of non-segregation and cannot be converted into BS4 evidence. The assessment is confidence-moderate because it rests on absence of pedigree data rather than on a computed ratio. The single ClinVar 'Likely benign' submission (SCV000551135, Labcorp Genetics) is another source's final classification and was not used to infer, support, or satisfy BS4.

Rule

InSiGHT MSH6 v2.0 BS4 (combined pedigree Bayes Likelihood Ratio, pedigrees combined multiplicatively; COOL v3 recommended): Strong <0.05; Supporting >0.05 and <=0.48. Likelihood ratio for this variant: not calculable (no pedigree data); BS4 unmet at all strengths.

Evidence considered

• InSiGHT MSH6 v2.0 (CSPEC doc 1578294885) lists BS4 'Applicable' with the instructions: for multiple pedigrees results are combined by multiplying them together; COOL v3 is the recommended tool and the 'overall Bayes Factor' determines strength. Rule thresholds: Strong for a combined Bayes Likelihood Ratio <0.05; Supporting for >0.05 and <=0.48. The specification's caveat notes that phenocopies for common phenotypes such as cancer can mimic lack of segregation.
• ClinVar variation 410451: 8 germline submissions (7 Uncertain significance, 1 Likely benign), 0 expert panels; no submission reports non-segregation, family testing or an informative pedigree, and no submitter applies BS4.
• Screened the seven fetched full texts for the exact variant and for any non-segregation report: none mentions c.2017C>G or p.(P673A) and none presents family genotypes or segregation results.
• No pedigree or meiosis-level data exist anywhere in this case record, so no likelihood ratio can be shown to fall below 0.05 (Strong) or within 0.05-0.48 (Supporting); absence of segregation data is not evidence of non-segregation.
• The one ClinVar submission classifying the variant Likely benign (SCV000551135, Labcorp Genetics) was not used as BS4 evidence; BS4 requires pedigree-level non-segregation, not another source's final classification.
• Generic ACMG/AMP BS4 (PMID:25741868): lack of segregation in affected members of a family; no family with affected non-carriers is described for this variant, so the generic criterion is unmet as well.

Still missing

• No pedigree or family-member genotype data accompany this variant, so lack of co-segregation cannot be evaluated and no Bayes likelihood ratio (<0.05 for Strong, >0.05-0.48 for Supporting) can be computed; an informative family showing the variant in unaffected relatives (with age and tumour-phenotype data) would be required to apply BS4.
• Because MSH6 Lynch syndrome is a common adult-onset cancer phenotype, published pedigrees would also need to account for phenocopies and possible second pathogenic variants before any non-segregation call could be made (the specification's own BS4 caveat).

Sources

Reasoning

Governing rule (MSH6 VCEP v2.0, BP5 applicable, default strength Supporting; count-based tiers): Strong = '4 or more tumors: CRC/Endometrial tumors with MSS and/or no loss of MMR protein expression and/or LS spectrum tumors with loss of MMR protein(s) that is inconsistent with the gene demonstrating genetic variation', or 2 or more BRAF V600E (CRC only) / MLH1 methylation (LS spectrum tumor only) results with MSI-H/MLH1 loss; Supporting = '2 or 3 tumors' with the same MSS / no-MMR-loss pattern, or 1 BRAF V600E / MLH1 methylation result with MSI-H/MLH1 loss. The operator here is a count threshold that the observed number of qualifying tumors must reach or exceed (a higher count yields stronger benign evidence). Explicit comparison with this case's data: qualifying MSS / no-MMR-loss (or MMR-loss-inconsistent-with-MSH6) tumors reported for c.2017C>G = 0, so '0 >= 2?' = no, and BP5 is not_met at any strength. No tumor data of any kind are reported for this variant - the ClinVar record VCV000410451 carries only variant-level classifications, and no MSS, MMR IHC, BRAF V600E or MLH1 methylation result is available in any consulted source. Note on operator direction for this criterion: BP5 in this framework is tumour-count based and carries no likelihood-ratio threshold; the 0.48 likelihood-ratio value that appears elsewhere in the MSH6 v2.0 framework belongs to BS3_Supporting (functional odds) and BS4 (co-segregation Bayes likelihood ratio), neither of which is part of this group, and it was not applied here.

Rule

BP5 (MSH6 VCEP v2.0). Supporting: 2 or 3 tumors - CRC/Endometrial tumors with MSS and/or no loss of MMR protein expression and/or LS spectrum tumors with loss of MMR protein(s) inconsistent with the gene demonstrating genetic variation; OR 1 BRAF V600E (Colon only)/MLH1 methylation (in LS spectrum tumor only) with MSI-H/MLH1 loss. Strong: 4 or more such tumors; OR 2 or more BRAF V600E (CRC only)/MLH1 methylation results with MSI-H/MLH1 loss.

Evidence considered

• MSH6 VCEP v2.0 BP5 is applicable with count-based tiers: Strong = 4 or more MSS / no-MMR-loss CRC or endometrial tumors (or 2 or more BRAF V600E / MLH1 methylation results with MSI-H/MLH1 loss); Supporting = 2 or 3 such tumors (or 1 BRAF V600E / MLH1 methylation result).
• Explicit comparison for this variant: the number of qualifying tumors reported is 0, so 0 >= 2? No - BP5 is not met.
• ClinVar VCV000410451 reports no MSS, MMR IHC, BRAF V600E or MLH1 methylation result for this variant; the record contains variant-level classifications only.
• No fetched paper reports tumour testing for c.2017C>G (PMID:30267214 analysed only potentially disruptive variants; the remaining papers are policy or review documents).
• The 0.48 likelihood-ratio value in the MSH6 v2.0 framework belongs to BS3_Supporting and BS4, not to BP5; BP5 here is tumour-count based, so no numeric ratio threshold was applied.

Still missing

• No MSS, MMR protein IHC, BRAF V600E or MLH1 methylation data are available for this variant; at least 2 qualifying CRC/endometrial MSS or no-MMR-loss tumors would be needed for BP5_Supporting.

Sources