MLH3 c.628C>T, p.(Arg210Ter)

Reasoning

Governing framework check first: no gene-specific specification exists for MLH3 (cspec.found = false, gene_vcep_dir = null, vcep_materials reports 'no cspec url available' and final_classification_framework fell back to generic_acmg_fallback with framework_mode = generic_acmg), so PVS1 was applied with the generic ClinGen SVI PVS1 recommendations (Abou Tayoun et al., Hum Mutat 2018;39:1517-1524, PMID:30192042; registry key pvs1_generic_framework), which refine - not replace - the ACMG/AMP 2015 PVS1 definition (PMID:25741868). Both applicability gates are satisfied and every downgrade branch of the SVI flowchart was checked and found inapplicable, so PVS1 is met at its unmodified default strength, very strong. (1) Variant type: NM_001040108.2:c.628C>T is a stop-gained change, NP_001035197.1:p.(Arg210Ter) (one-letter p.(R210*)), consequence class 'nonsense', variant_bucket 'nonsense', canonical_splice_consensus = false - a predicted null allele, the exact variant class PVS1 addresses. (2) Transcript relevance and exon position: the variant lies in exon 2 of 13 of NM_001040108.2, the MANE Select and RefSeq Select transcript (CDS c.1-4362 in 13 exons; terminal exon c.4243 to *3307 per the normalized transcript model), and VariantValidator places both the start and end of the change in exon 2; this is the most biologically relevant MLH3 transcript, and exon 2 is a constitutive coding exon of it, not a cassette exon skipped from the major transcript (no evidence of MLH3 exon 2 skipping was found). (3) NMD: the SVI rule is that NMD is NOT predicted only when the premature termination codon lies in the 3'-most exon or within the 3'-most 50 bp of the penultimate exon. c.628 is in exon 2, several kilobases upstream of the penultimate exon-exon junction, so NMD is predicted; the NMD-escape branch (which caps PVS1 at strong, or at moderate when <10% of the protein is removed) therefore does not apply. (4) Protein length change and protein region: the reference protein NP_001035197.1 is 1453 aa and the predicted product is 210 aa, so 1244 residues (85.6%) are removed, far above the 10% threshold used in the SVI truncation branches. The removed region is critical, not dispensable: PMID:10615123 shows MLH3 is a 160-kD mismatch-repair protein that interacts with MLH1 through its C-terminal region (the N-terminal ~240 aa carry the predicted ATPase motif) and that a truncated MLH3 construct stably expressed in 293T/COS7 cells induced microsatellite instability in a frameshift beta-galactosidase reporter assay. p.(Arg210Ter) deletes the entire MLH1-interaction C-terminal domain. (5) Repeat-region and splice context: codon 210 is not in a repeat or homopolymer tract, and the change is not a splice-region variant (SpliceAI max delta 0.001; DS_AG 0.0, DS_AL 0.0, DS_DG 0.001, DS_DL 0.001), so the consequence assignment is unambiguous and the SVI caveat against stacking PVS1 with PP3 splice predictions does not arise. (6) The SVI 'LoF-enriched / dispensable exon' downgrade branch does not apply either: this nonsense is present in gnomAD v4.1 at 76/1613818 alleles (AF 4.70933e-05, grpmax FAF 4.945e-05, zero homozygotes) and in gnomAD v2.1 at 7/250738 alleles (AF 2.79176e-05, zero homozygotes) - a low-frequency heterozygous carrier state, not the high-frequency LoF enrichment used to infer exon dispensability, and well below the BS1 (0.01) and BA1 (0.05) frequency thresholds. (7) Gene-disease mechanism: LoF is a documented mechanism for MLH3 - a founder truncating allele, c.3563C>G p.(Ser1188Ter), in the homozygous state underlies adenomatous polyposis predisposition (Olkinuora et al., Genet Med 2019;21:1868-1873, PMID:30573798) - MLH3 encodes a mismatch-repair protein whose truncation produces microsatellite instability in vitro (PMID:10615123), the exact variant is curated in ClinVar (VCV000860534, variation 860534) under MLH3-associated 'Colorectal cancer, hereditary nonpolyposis, type 7' and 'Endometrial carcinoma', and OncoKB annotates p.Arg210Ter's biological effect as 'Likely Loss-of-function' (only that biological-effect field was used for the mechanism question; OncoKB's oncogenicity classification was not treated as evidence). Residual uncertainty that is recorded rather than converted into a strength reduction: the mechanism evidence for monoallelic (heterozygous) MLH3 LoF is weaker than for the biallelic setting, because the established hereditary MMR-cancer gene panels confine the phenotype to MLH1/MSH2/MSH6/PMS2 (plus EPCAM) and do not include MLH3 (PMID:34043773, PMID:33516529, PMID:24493721), and the SVI framework reserves unmodified PVS1 for gene-disease pairs at definitive or strong clinical validity that further show LoF variants in >=10% of reported pathogenic variants with at least three pathogenic LoF variants classified without PVS1 across more than one exon. Neither that quantitative precondition nor a ClinGen gene-disease validity tier for MLH3 could be verified from the sources available in this case (no CSPEC, no MLH3 VCEP, no retrievable machine-readable ClinGen gene-validity record), which is why human review is requested. Strength is nevertheless reported as very strong because no specific, named SVI downgrade rule's conditions are met: NMD is predicted, the affected exon is a biologically relevant constitutive exon, the truncated region is critical to MLH1 interaction, and there is no population LoF enrichment at this codon.

Rule

ClinGen SVI PVS1 decision tree (Abou Tayoun et al. 2018, PMID:30192042; generic ACMG/AMP fallback, no MLH3 CSPEC/VCEP): nonsense variant -> premature termination codon in exon 2 of 13, upstream of the 3'-most 50 bp of the penultimate exon, so NMD is predicted and the truncated region (85.6% of the protein, including the C-terminal MLH1-interaction domain) is critical -> PVS1 at the default unmodified strength, very strong; no named downgrade branch (NMD escape, non-critical exon, alternative splicing, population LoF enrichment) is met.

Evidence considered

• Governing framework: generic ClinGen SVI PVS1 recommendations (registry key pvs1_generic_framework, PMC6185798; Abou Tayoun et al. 2018, PMID:30192042). The rules relied on were: PVS1 applies only where LoF is a disease mechanism for the gene/disease pair; NMD is not predicted only when the premature termination codon is in the 3'-most exon or within the 3'-most 50 bp of the penultimate exon; in the NMD-escape setting strength depends on whether more than 10% of the protein is removed (strong) or less (moderate); PVS1 must not be combined with PP3 splice predictions for the same variant; exonic null variants are downgraded where the affected exon is alternatively spliced from the major transcript, is enriched with high-frequency LoF variants in the general population, or removes a region that is not critical to protein function; and full-strength PVS1 is reserved for gene-disease pairs with definitive/strong validity, LoF variants making up >=10% of reported pathogenic variants and at least three pathogenic LoF variants classified without PVS1 across more than one exon.
• Variant-level conformation (registry key pvs1_variant_assessment): consequence class 'nonsense', variant bucket 'nonsense', canonical_splice_consensus false, transcript NM_001040108.2, protein NP_001035197.1:p.(Arg210Ter), framework applied = generic ClinGen SVI PVS1 recommendations, default strength for an un-downgraded call = very strong.
• Transcript geometry from the normalized variant record: NM_001040108.2 has 13 exons with the coding sequence at c.1-4362; the terminal exon spans c.4243 to *3307; VariantValidator reports variant_exonic_positions start_exon 2 and end_exon 2 for both GRCh37 (NC_000014.8) and GRCh38 (NC_000014.9), i.e. c.628C>T sits in exon 2 of 13, several kilobases upstream of the penultimate exon-exon junction. NM_001040108.2 is annotated MANE Select and RefSeq Select (CCDS32123.1, HGNC:7128), so it is the most biologically relevant transcript.
• Protein-length consequence from the same record: reference protein length 1453 aa (mutalyzer position_last_original 1454 including the stop codon); predicted product 210 aa (position_last_predicted), first changed position 209/210 - removal of 1244 of 1453 residues, i.e. 85.6% of MLH3, including the C-terminal region that PMID:10615123 identifies as the MLH1-interaction domain.
• Splice context (registry key spliceai): SpliceAI Lookup max delta score 0.001 (DS_AG 0.0, DS_AL 0.0, DS_DG 0.001, DS_DL 0.001) - the variant is not a splice-region change, so the consequence is a pure protein-truncating null and no PVS1/PP3 double-counting issue arises.
• Population context used only to test the SVI 'exon enriched with high-frequency LoF variants' downgrade branch: gnomAD v4.1 (registry key gnomad_v4) 76/1613818 alleles, AF 4.70933e-05, grpmax FAF 4.945e-05, zero homozygotes; gnomAD v2.1 (registry key gnomad_v2) 7/250738 alleles, AF 2.79176e-05, grpmax FAF 2.298e-05, zero homozygotes. A heterozygous carrier state at ~1e-4 is not high-frequency LoF enrichment of the exon (and is far below the BS1 0.01 / BA1 0.05 calibration thresholds for population frequency).
• Gene-disease context (registry key clinvar): variation 860534 is an exact match to NM_001040108.2(MLH3):c.628C>T (p.Arg210Ter) and is curated under 'Colorectal cancer, hereditary nonpolyposis, type 7' and 'Endometrial carcinoma'; three germline submissions from ordinary clinical laboratories (2 Uncertain significance, 1 Pathogenic), zero expert-panel submissions. ClinVar was consulted for the MLH3 disease/gene context only; no criterion was inferred from any classification it reports.
• Protein-effect context (registry key oncokb): OncoKB annotates the biological effect of MLH3 R210* as 'Likely Loss-of-function' (cited to PMID:10615123), consistent with a true null consequence at the protein level. Only that biological-effect field was used; the curated oncogenicity label was not treated as evidence for PVS1.
• PMID:10615123 (MLH3 cloning and characterisation): MLH3 at 14q24.3 encodes a 160-kD mismatch-repair protein that interacts with MLH1; the N-terminal ~240 aa contain a predicted ATPase motif and the C-terminus encodes the predicted MLH1-interaction domain. Truncating-allele functional data: a truncated MLH3 construct (DN-MLH3) stably expressed in 293T/COS7 cells induced microsatellite instability in an in-frame/out-of-frame beta-galactosidase microsatellite reporter assay, and the authors hypothesise that dominant-negative MLH3 mutations might analogously induce MSI and HNPCC in rare instances. This establishes that the C-terminal region removed by p.(Arg210Ter) is functionally critical and that truncating MLH3 alleles have documented functional impact; the paper does not report c.628C>T itself, so it was not treated as variant-specific evidence.
• PMID:30573798 (Olkinuora et al., Genet Med 2019;21:1868-1873, 'Biallelic germline nonsense variant of MLH3 underlies polyposis predisposition'; fetched as an abstract-level record in this case, no full text): exome sequencing of 40 unexplained FAP/AFAP cases plus a ~1000-patient Swedish panel screen identified four homozygous carriers of the truncating founder allele MLH3 c.3563C>G p.(Ser1188Ter) with classical or attenuated adenomatous polyposis and a shared 0.8-Mb haplotype. Direct quote from the abstract conclusion: 'Our results show that a biallelic nonsense variant of MLH3 underlies a novel syndrome with susceptibility to classical or attenuated adenomatous polyposis and possibly extracolonic tumors, including breast cancer.' The same abstract states that 'Colorectal polyps from variant carriers showed no instability at mono-, di-, tri-, or tetranucleotide repeats, in agreement with previous findings of a minor role of MLH3 in MMR.' This is the primary LoF-mechanism evidence for MLH3 used here; it is mechanism-level (biallelic, recessive), not evidence about c.628C>T.
• Mechanism caveat from guideline-level sources: PMID:34043773 (EHTG/ESCP Mallorca Lynch syndrome guidelines) states 'LS is caused by inherited malfunction in one of the four MMR genes named MLH1, MSH2, MSH6 and PMS2'; PMID:33516529 (SGO endometrial cancer review) lists hereditary endometrial cancer predisposition as PTEN plus MLH1, MSH2, MSH6, PMS2 and EPCAM; PMID:24493721 (ASCO family-history statement) lists the MMR loci warranting testing as MLH1/MSH2/MSH6/PMS2/EPCAM. MLH3 appears in none of these lists, so the monoallelic MLH3 disease association is weaker than for the four canonical Lynch genes - recorded as uncertainty, not as a named SVI downgrade.
• A ClinGen gene-disease validity lookup for MLH3 was attempted but the ClinGen search interface returned no machine-readable curation (JavaScript-only responses), so no ClinGen validity tier for MLH3 could be cited; this is the principal unresolved input for the SVI gene-mechanism precondition.
• Note on a case-artifact field: the gene-context record's disease_names field contains the sentence 'Novel MSH2 splice-site mutation in a young patient with Lynch syndrome' - that is the title of an unrelated PMID:29568967 paper, i.e. a garbled value. Nothing in this adjudication rests on that field; the MLH3 gene-disease and mechanism statements rest on PMID:30573798, PMID:10615123 and the ClinVar condition labels.

Still missing

• No ClinGen CSPEC and no MLH3-specific VCEP or local framework exist, so the generic ClinGen SVI PVS1 decision tree was used and no gene-specific PVS1 thresholds or pre-assigned strengths were available.
• No machine-readable ClinGen gene-disease validity curation for MLH3 could be retrieved, so the SVI precondition for unmodified PVS1 (definitive/strong gene-disease pair, LoF variants in >=10% of pathogenic variants, and at least three pathogenic LoF variants classified without PVS1 distributed across more than one exon) could not be verified for this gene.
• The only formal MLH3 LoF disease-mechanism evidence located is biallelic (homozygous truncating carriers with adenomatous polyposis, PMID:30573798); whether heterozygous MLH3 LoF is sufficient for the ClinVar-listed MLH3 conditions (hereditary nonpolyposis colorectal cancer type 7, endometrial carcinoma) is not established by any source in this case.
• NMD was inferred from the position of the premature stop codon within the transcript model (exon 2 of 13); no RNA or functional study of this exact allele (c.628C>T) exists in the evidence, so the reduction of the transcript/protein to a true null was not measured directly.
• Exon 2 biological relevance was assessed against the MANE Select/RefSeq Select transcript NM_001040108.2 only; no MLH3 transcript-level RNA data were available to exclude a minor alternative transcript that skips exon 2.

Sources

Reasoning

PM2 requires the variant to be absent from, or present at extremely low frequency in (below the expected carrier frequency for the disorder), a large general-population or control cohort with adequate read depth (Richards et al. 2015, PMID:25741868), applied at Supporting strength per the ClinGen SVI recommendation (ClinGen SVI 2020, PMID:25741868), for which the generic calibration used in this case sets the cut-off at an allele frequency <= 0.0001. No MLH3 CSPEC/VCEP exists (cspec lookup not found; final_classification_framework.found = false, framework_complete = false), so the all-comers GNOMAD_V2_1 and GNOMAD_V4_1 totals are the default sources and no gene-specific PM2 threshold overrides the generic one. The variant is rare in every dataset: gnomAD v4.1 total AF 4.70933e-05 (76/1,613,818 alleles; grpmax FAF 4.945e-05; exomes 4.99376e-05, 73/1,461,824; genomes 1.97376e-05, 3/151,994) and gnomAD v2.1 exomes AF 2.79176e-05 (7/250,738; grpmax FAF 2.298e-05), with the highest ancestry-group frequency 6.10167e-05 (European non-Finnish, v4.1) and highest popmax FAF of 6.10e-05; the variant is absent from gnomAD-Canada v1.0 and homozygote counts are 0 in every dataset. Each of these standard frequency metrics (overall AF, exome AF, grpmax/popmax FAF) is below the 0.0001 cut-off, so PM2 is met. The strength was not escalated above Supporting: no gene-specific MLH3 calibration exists, and the only calibration supplied for this case is the SVI Supporting formulation. Confidence is medium rather than high, and the criterion is flagged for human review, because one small gnomAD v2.1 subgroup (NFE_SEU) shows 3/11,480 alleles (AF 2.61324e-04), a value above the 0.0001 cut-off; this subgroup is not the grpmax/popmax metric, its allele count is very small with a wide confidence interval, and no other ancestry group or dataset approaches it, but a reviewer who prefers a strict maximal-subgroup frequency would consider PM2 arguable rather than clearly met.

Rule

PM2 Supporting: allele frequency <= 0.0001 (ClinGen SVI 2020 downgrade to Supporting, PMID:25741868); generic default, no MLH3 VCEP-specific cutoff

Evidence considered

• Threshold applied: PM2 Supporting at allele frequency <= 0.0001 (ClinGen SVI 2020 recommendation to apply PM2 at Supporting strength, PMID:25741868, as calibrated for this case); no MLH3 VCEP or gene-specific PM2 threshold exists to supersede it.
• gnomAD v4.1 (chr14-75049028-G-A, GRCh38): total AF 4.70933e-05 (AC 76 / AN 1,613,818, homozygotes 0); exome AF 4.99376e-05 (73/1,461,824); genome AF 1.97376e-05 (3/151,994); grpmax FAF 4.945e-05; popmax/ancestry maximum European (non-Finnish) 6.10167e-05 (72/1,180,004).
• gnomAD v2.1 (14-75515731-G-A, GRCh37, exomes only): total AF 2.79176e-05 (AC 7 / AN 250,738, homozygotes 0); grpmax FAF 2.298e-05; popmax European (non-Finnish) 5.28811e-05 (6/113,462); all ancestry groups <= 5.3e-05.
• gnomAD-Canada v1.0 (HostSeq genomes): variant absent (search_status 'absent'), an independent control cohort with no observation of this allele.
• Homozygote counts are 0 in both gnomAD datasets and in every ancestry subgroup, so the allele is not segregating at a frequency consistent with a common benign or well-tolerated variant.
• Read depth/AN is sufficient for a reliable rarity call (250,738 v2.1 exome alleles; 1,613,818 v4.1 alleles; 151,994 v4.1 genome alleles), so the very low counts reflect true rarity rather than missed calls.
• Threshold artifact noted: the highest single subgroup value across both datasets is gnomAD v2.1 NFE_SEU 2.61324e-04 (3/11,480 alleles), which is above the 0.0001 cut-off; it is a small sub-cohort, is not the grpmax FAF (2.298e-05 in v2.1, 4.945e-05 in v4.1) and does not change the over-arching rarity of the allele, but it is the reason this criterion carries medium confidence and a human-review flag.
• Cross-check against the ClinVar record (ClinVarID 860534, NM_001040108.2(MLH3):c.628C>T p.Arg210Ter, 3 submissions, 2-star 'criteria provided, multiple submitters, no conflicts') is consistent with rarity: no submitter reports a population frequency for this allele, and the Labcorp/Invitae submission's criterion-level population signal cites only that it is present in population databases - the frequency assessment above was derived independently from the gnomAD data and not from any submitter's classification.

Still missing

• No non-cancer-cohort gnomAD subset (GNOMAD_V2_1_NON_CANCER exomes-only or GNOMAD_V3_1_NON_CANCER genomes-only) was retrieved for this case, so the cancer-cohort-inclusive all-comers totals were used; this is the prescribed default because no MLH3 VCEP specifies a non-cancer source, and any non-cancer subset would be equal to or rarer than the totals.
• No gene-specific MLH3 PM2 (or BA1/BS1) frequency derivation exists, so the generic SVI Supporting calibration was applied without gene-specific adjustment.
• Small-subgroup ambiguity: gnomAD v2.1 NFE_SEU 3/11,480 (AF 2.61e-04) exceeds the 0.0001 PM2 cut-off while the grpmax FAF (2.298e-05) and overall AF (2.79e-05) do not; the criterion was evaluated on the grpmax/popmax and overall metrics.

Sources

Reasoning

PS3 (Richards et al. 2015, PMID:25741868: a well-established functional study shows a deleterious effect on the gene or gene product) is not met for MLH3 NM_001040108.2:c.628C>T (NP_001035197.1:p.(Arg210Ter)). Scope first: this is a stop-gained allele located in exon 2 of 13 (c.-63_c.3280) of a 1454-residue protein, so none of the computational paths (REVEL, SpliceAI, BayesDel) can supply PS3-type evidence and the criterion rests entirely on direct functional assay data. Review of the functional literature retrieved for this case found no assay performed on this allele: the single MLH3 functional study available (PMID:10615123) reports that stable overexpression of an engineered N-terminal MLH3 deletion construct (DN-MLH3, analogous to a Pms1p dominant-negative mutant) induced microsatellite instability in 293T and COS7 cells carrying an in-frame/out-of-frame ATG(CA)27/29 beta-galactosidase reporter, that stable clones expressing full-length MLH3, MLH3-delta-exon-7 or an MLH3 C-terminal deletion mutant could not be generated (attributed to toxicity of the MLH3 N terminus), and that this dominant-negative behaviour might induce MSI/HNPCC in rare instances; the variant c.628C>T, its protein change p.(Arg210Ter) and the residue R210 are never mentioned, no patient-derived material was assayed, and no variant-level pathogenic or benign control series was used to validate the readout. An engineered deletion construct tested by overexpression in an MMR-proficient cell line is not the same allele as a patient germline nonsense change, and this assay was not applied to c.628C>T, so it cannot establish a deleterious effect of the variant and does not satisfy PS3's requirement for a well-established, variant-level functional demonstration. The OncoKB entry for R210* ('Likely Loss-of-function', 'Likely Oncogenic', sourced to the same PMID:10615123) is a curation-level inference drawn from a truncating change rather than an assay result, and a criterion may not be inferred from another source's classification; it was therefore not scored as PS3 evidence. Biological-relevance check: MLH3 is absent from the established Lynch-syndrome mismatch repair gene set (PMID:34043773 states Lynch syndrome is caused by inherited malfunction in MLH1, MSH2, MSH6 and PMS2 only; PMID:33516529 lists MLH1, MSH2, MSH6, PMS2 and EPCAM), so there is no disease-validated, gene-specific MLH3 functional assay framework that could license a reduced-strength application, and no ClinGen CSPEC/VCEP exists for MLH3 (final classification framework resolved to the generic ACMG/AMP 2015 fallback), leaving the generic PS3 definition as the governing rule. Mechanistic expectation from protein architecture - the p.Arg210Ter product lacks the C-terminal MLH1-interaction domain, and the exon-7-encoded helix is critical for MLH1 binding (PMID:10615123) - is predictive reasoning, not functional evidence, and cannot substitute for an assay. PS3 is consequently not met; this means the criterion's evidentiary requirement is unmet, not that retained function of the allele has been demonstrated.

Evidence considered

• PS3 requires a well-established functional study showing a deleterious effect on the gene or gene product (Richards et al. 2015, PMID:25741868); no such study exists for this variant.
• PMID:10615123 is the only MLH3 functional study retrieved: stable overexpression of an engineered N-terminal MLH3 deletion construct (DN-MLH3) induced microsatellite instability in 293T/COS7 cells using in-frame ATG(CA)27 (pCAR-IF) / out-of-frame ATG(CA)29 (pCAR-OF) beta-galactosidase reporters, and no stable clones expressing full-length MLH3, MLH3-delta-exon-7 or an MLH3 C-terminal deletion mutant could be obtained.
• The same paper never mentions c.628C>T, p.(Arg210Ter), R210* or any patient carrying a MLH3 truncating allele, and reports no variant-specific assay, no control series and no patient material, so its construct-level MSI result cannot be transferred to this allele.
• The OncoKB record for R210* reports 'Likely Loss-of-function' / 'Likely Oncogenic' with PMID:10615123 as its only citation; this is a database curation inference from a truncating change, not a functional assay result, and per the case rules a criterion is never inferred from another source's final classification.
• Biological-relevance check: MLH3 is not one of the four path_MMR Lynch-syndrome genes (PMID:34043773) and is not among the MMR genes listed for testing in MMR-deficient tumours (PMID:33516529: MLH1, MSH2, MSH6, PMS2, EPCAM), so no established disease-specific MLH3 functional assay framework is available for PS3 calibration.
• Gene architecture argument (not scored): the p.Arg210Ter product ends at residue 209 of 1454 and therefore lacks the C-terminal MLH1-interaction domain, but predicted loss of a domain is not functional evidence for PS3.
• A PubMed abstract reviewed for gene-level context (PMID:30573798, Olkinuora et al., Genet Med 2019, 'Biallelic germline nonsense variant of MLH3 underlies polyposis predisposition') supports MLH3 loss of function being disease-relevant, but in a biallelic/recessive polyposis setting; that record is an abstract only, is not part of the case source registry, and reports no assay of c.628C>T.
• No ClinGen CSPEC or gene-specific MLH3 specification was available (cspec.found = false; gene_vcep_dir null; framework_mode generic_acmg), so PS3 was assessed against the generic ACMG/AMP 2015 criterion definition.

Still missing

• No variant-level functional assay exists for c.628C>T (p.Arg210Ter): no MMR-activity, MSI, MLH1-interaction or protein-stability assay has been performed on this allele in any retrieved source.
• The only MLH3 functional study (PMID:10615123) used an engineered N-terminal deletion construct in overexpression cell lines, not this germline nonsense allele, and no validated pathogenic/benign control series was established for that readout.
• MLH3 has no ClinGen CSPEC/VCEP, so there is no gene-specific functional assay framework; PS3 was scored against the generic ACMG/AMP 2015 definition.
• PMID:33451724 and PMID:24929052 were triaged for this case but never retrieved as full text, so they could not be screened for MLH3 functional content (their endometrial-cancer review themes make variant-level assay data unlikely).

Sources

Reasoning

PS4 requires that the variant's prevalence in affected individuals be statistically increased over controls. No case-control study, cohort enrichment analysis, or variant-specific case series for MLH3 c.628C>T exists in this case's evidence. ClinVar variation 860534 holds three germline submissions, all from ordinary clinical laboratories with 'criteria provided, single submitter' review status (SCV005920953 Sinai Health System, Uncertain significance; SCV001231874 Labcorp/Invitae, Uncertain significance; SCV002657186 Ambry Genetics, Pathogenic) and no case counts, cohort denominator, or case-control design; expert_panel_submissions = 0. The candidates retrieved as case-observation leads (PMID:24493721, PMID:33516529, PMID:34043773) are gene-level consensus/review documents that never mention this variant, and the MLH3 primary paper (PMID:10615123) reports no patient or family carrying a MLH3 truncating allele. Population data describe controls only and cannot establish enrichment in cases: gnomAD v2.1 7/250738 alleles (AF 2.79176e-05, zero homozygotes), gnomAD v4.1 76/1613818 alleles (AF 4.70933e-05, zero homozygotes), gnomAD-Canada absent; 7-76 control alleles is the opposite of a case-side enrichment signal. On the numeric question: no PS4 odds-ratio threshold is available for this gene. There is no ClinGen CSPEC or MLH3-specific VCEP specification, and the calibration block supplied for this case contains no PS4 cutoff, so no operator (>=, <=, >, <) and no substitution were defined and no inequality was evaluated; any PS4 number recalled from memory without a named source publication would be indistinguishable from a fabricated threshold. PS4 is therefore not met on absence of any enrichment evidence, not on a threshold comparison. The COSMIC entry (COSV53134876, n = 1 somatic occurrence) is a tumour-database record, not enrichment data.

Evidence considered

• ClinVar variation 860534 (NM_001040108.2(MLH3):c.628C>T, p.Arg210Ter) carries three germline submissions, each 'criteria provided, single submitter', from ordinary clinical laboratories: Sinai Health System (Uncertain significance), Labcorp Genetics/Invitae (Uncertain significance) and Ambry Genetics (Pathogenic); no expert-panel submission is present (expert_panel_submissions = 0).
• Population-frequency records used only as the control side of the comparison: gnomAD v2.1 exome 7/250738 alleles (AF 2.79176e-05, grpmax FAF 2.298e-05, 0 homozygotes), gnomAD v4.1 76/1613818 alleles (AF 4.70933e-05, grpmax FAF 4.945e-05, 0 homozygotes), gnomAD-Canada v1.0 absent. These describe unaffected population controls, not affected cases, so they cannot demonstrate enrichment.
• The ClinVar-cited papers screened for case-observation content are endometrial-cancer and Lynch-syndrome reviews and consensus guidelines: PMID:24493721 (ASCO family-history statement), PMID:33516529 (SGO endometrial cancer review), PMID:34043773 (EHTG/ESCP Mallorca Lynch syndrome guidelines), PMID:33451724 and PMID:24929052 (endometrial cancer reviews, no full text retrieved) and PMID:24905773 (no registry key or full text in this case). The fetched ones never mention MLH3 or c.628C>T/p.(R210*).
• PMID:10615123, the MLH3 cloning/characterisation paper, reports no patient or family carrying a MLH3 nonsense allele and no MLH3 coding polymorphism, so it contributes no case or cohort counts for PS4.
• No ClinGen CSPEC was retrieved for MLH3 (cspec.found = false) and no gene-specific VCEP or local framework was available, so generic ACMG/AMP 2015 criterion definitions govern; no published PS4 odds-ratio threshold was supplied in the calibration data for this case, so no numeric operator or inequality could be stated for PS4.
• COSMIC records this variant once as a somatic change (COSV53134876, somatic count 1); a single somatic database entry is not case-control enrichment evidence, and COSMIC has no source_registry key that could be cited as a reference.
• A single laboratory's Pathogenic assertion (SCV002657186, Ambry Genetics, no validated PMIDs) is an ordinary submitter classification and was not treated as case-enrichment or expert-panel evidence.

Still missing

• No case-control study or case-enrichment analysis involving MLH3 c.628C>T was identified in any retrieved source.
• The three ClinVar submissions for this variant are single-submitter laboratory assertions with no case counts, cohort denominators, or case-control design.
• No MLH3-specific VCEP/CSPEC specification exists, and the generic ACMG/AMP framework supplies no numeric PS4 odds-ratio threshold, so no threshold comparison was possible.

Sources

Reasoning

PM3 (Richards et al. 2015, PMID:25741868) applies for a recessive disorder when the variant is detected in trans with a pathogenic variant. NM_001040108.2:c.628C>T is a germline nonsense allele in MLH3 predicted to give NP_001035197.1:p.(Arg210Ter). MLH3 does have a documented biallelic mechanism - biallelic germline MLH3 nonsense variants have been reported to underlie polyposis predisposition (Olkinuora et al., Genet Med 2019, PMID:30573798) - so PM3 was assessed on its merits rather than dismissed as not applicable to a gene whose principal reported indication (hereditary nonpolyposis colorectal cancer type 7) is dominant. No in-trans configuration is documented for this variant anywhere in the case. The three ClinVar submissions for VCV000860534 (SCV005920953, Sinai Health System; SCV001231874, Labcorp Genetics/Invitae; SCV002657186, Ambry Genetics) are all germline clinical-testing submissions from single submitters with no expert-panel review, and none of them provides phase, zygosity, second-variant or pedigree information; condition labels are hereditary nonpolyposis colorectal cancer type 7 and endometrial carcinoma. No proband, trio, family or segregation data exist in any case artifact, the comparator-variant collection (pm5_candidates) contains zero MLH3 candidates, and none of the papers retrieved for this case mentions c.628C>T, p.(Arg210Ter), R210* or 628C>T. Population data also show no biallelic observation (gnomAD v2.1 homozygotes 0; gnomAD v4.1 homozygotes 0). With no second pathogenic MLH3 variant documented on the opposite allele, the in-trans requirement of PM3 is unmet.

Evidence considered

• ClinVar VCV000860534 (NM_001040108.2(MLH3):c.628C>T p.Arg210Ter): three germline clinical-testing submissions, all single-submitter and none from an expert panel - SCV005920953 (Sinai Health System, Uncertain significance, 2023), SCV001231874 (Labcorp Genetics/Invitae, Uncertain significance, 2025), SCV002657186 (Ambry Genetics, Pathogenic, 2024). Classification summary: 2x Uncertain significance, 1x Pathogenic. Conditions: hereditary nonpolyposis colorectal cancer type 7, endometrial carcinoma, not specified. No submission records phase, zygosity, cis/trans configuration, or a co-occurring second MLH3 variant.
• Case-wide search for allelic-phase evidence: the strings proband, in trans, in cis, phase (in a genotype sense), segregation, zygosity, compound heterozygous and biallelic do not occur in any case artifact searched (review_manifest, case_summary, prefetch, evidence, literature_pass, fulltext_map, paper_extracts, clinvar_submission_audit); pm5_candidates.json reports 0 candidate comparator variants in MLH3, so no same-gene second variant was available to phase against.
• PM3 criterion text (Richards et al. 2015, PMID:25741868): for recessive disorders, detected in trans with a pathogenic variant. The guideline is gene-agnostic and reports no variant-specific, phase or family data for MLH3 c.628C>T.
• MLH3 recessive/biallelic mechanism context: the case's MLH3 germline-disease context artifact records biallelic germline MLH3 nonsense variants underlying polyposis predisposition (Olkinuora et al., Genet Med 2019, PMID:30573798, title and truncated abstract only), which is why PM3 was evaluated rather than treated as not applicable; that record does not name c.628C>T.
• No full-text paper in this case mentions the variant: c.628C>T, p.(Arg210Ter), R210* and 628C>T were searched and not found in PMID:10615123 (MLH3 cloning/functional study), PMID:25741868 (ACMG/AMP 2015), PMID:24493721 (ASCO cancer family history), PMID:33516529 (SGO endometrial cancer review) and PMID:34043773 (EHTG/ESCP Lynch syndrome guidelines).
• Population data show no biallelic/homozygous observation of this allele: gnomAD v2.1 total homozygotes 0 (7/250,738 alleles, AF 2.79176e-05) and gnomAD v4.1 total homozygotes 0 (76/1,613,818 alleles, AF 4.70933e-05), with 0 homozygotes in every subpopulation reported.

Still missing

• No phase-resolved genotype exists for a second MLH3 allele: the variant is present in all case sources as a single heterozygous germline finding with no cis/trans, zygosity, parental or pedigree information.
• Whether this specific allele has ever occurred in trans with a pathogenic MLH3 variant (including in the biallelic MLH3 polyposis families reported in the literature, e.g. PMID:30573798) cannot be settled from this case corpus: only a truncated abstract was available for that paper, and no other retrieved full text mentions c.628C>T.

Sources

Reasoning

PP5 applies only when the exact variant has been classified Pathogenic or Likely pathogenic by a ClinVar expert panel (a three-star expert-panel assertion). ClinVar variation 860534, NM_001040108.2(MLH3):c.628C>T (p.Arg210Ter), has three submissions and zero expert-panel submissions; the expert-panel field is empty and every submission is an ordinary clinical laboratory assertion with 'criteria provided, single submitter' status (SCV005920953 Sinai Health System, Uncertain significance; SCV001231874 Labcorp/Invitae, Uncertain significance; SCV002657186 Ambry Genetics, Pathogenic). A single laboratory's Pathogenic assertion, especially one with no validated PMID trail, is not an expert-panel classification and must never trigger PP5, and the aggregate ClinVar label is Uncertain significance, not Pathogenic/Likely pathogenic. The papers retrieved as PP5 candidate leads (PMID:24493721, PMID:33516529, PMID:34043773) do not mention this variant at all, so no expert-panel variant classification is available from them either. No numeric threshold and no comparison operator apply to this criterion; it is an applicability gate on the existence of an exact-variant expert-panel Pathogenic/Likely pathogenic classification, which does not exist here. PP5 is therefore not met.

Evidence considered

• ClinVar variation 860534 (exact match to NM_001040108.2:c.628C>T, p.Arg210Ter): submissions_all = 3, expert_panel_submissions = 0, expert panel field empty, aggregate significance Uncertain significance; individual SCVs are SCV005920953 (Sinai Health System, Uncertain significance, single submitter), SCV001231874 (Labcorp/Invitae, Uncertain significance, single submitter) and SCV002657186 (Ambry Genetics, Pathogenic, single submitter, no validated PMIDs).
• No three-star expert-panel assertion exists for this variant, so the PP5 applicability gate fails before any strength question arises; ordinary laboratory submissions and aggregate ClinVar labels are explicitly not usable for PP5.
• PMID:24493721, PMID:33516529 and PMID:34043773 were retrieved as PP5 candidate leads but are gene-level consensus/review documents that never mention MLH3 or c.628C>T, so they provide no expert-panel variant classification.
• PMID:25741868 defines the PP5 criterion (reputable source asserting pathogenicity without shared data) and its expert-panel/three-star interpretation used here; the guideline contains no variant-specific data.
• No ClinGen CSPEC or MLH3-specific VCEP specification exists (cspec.found = false; generic ACMG/AMP 2015 fallback), so no gene-specific PP5 pre-classification applies.

Still missing

• No ClinVar expert-panel (three-star) classification exists for this variant; the only Pathogenic assertion is a single-laboratory submitter (Ambry Genetics) with no validated PMID support.
• No expert-panel variant classification for MLH3 c.628C>T was found in the literature sources retrieved as PP5 leads.

Sources

Reasoning

BA1 requires a stand-alone allele frequency of at least 5% (Richards et al. 2015, PMID:25741868: an allele frequency in a control population over 5% is stand-alone support for a benign interpretation). No MLH3-specific ClinGen CSPEC or VCEP exists for this case (cspec lookup not found; final_classification_framework.found = false, framework_complete = false, so the generic ACMG/AMP 2015 fallback applies) and no gene-specific stand-alone threshold has been derived for MLH3, so the all-comers gnomAD totals GNOMAD_V2_1 (exomes) and GNOMAD_V4_1 are the governing sources. Observed frequencies: gnomAD v4.1 total 4.71e-05 (76/1,613,818 alleles; 4.99e-05 exomes 73/1,461,824; 1.97e-05 genomes 3/151,994); gnomAD v2.1 exomes 2.79e-05 (7/250,738). The highest ancestry-group frequency is European (non-Finnish): 6.10e-05 in v4.1 (72/1,180,004) and 5.29e-05 in v2.1 (6/113,462, homozygotes 0); the highest value in any finer gnomAD subgroup is 2.61e-04 (v2.1 NFE_SEU, 3/11,480). Every value - including that small-subgroup outlier - remains more than two orders of magnitude below 0.05, and the variant is absent from gnomAD-Canada v1.0. Read depth is sufficient for the frequency call (250,738 v2.1 exome alleles; 1,613,818 v4.1 alleles), so the low frequency is not a coverage artifact. BA1 is not met at any strength.

Rule

BA1 stand-alone: allele frequency >= 0.05 (Richards et al. 2015, PMID:25741868); no MLH3 VCEP-specific override

Evidence considered

• Threshold applied: BA1 stand-alone at allele frequency >= 0.05 (Richards et al. 2015, PMID:25741868); no MLH3-specific stand-alone threshold exists to supersede it.
• Framing: MLH3 has no ClinGen CSPEC/VCEP and no local gene-specific framework (final_classification_framework.found = false, framework_complete = false), so the all-comers GNOMAD_V2_1 and GNOMAD_V4_1 entries are the default population source; non-cancer subsets would only lower the frequency.
• gnomAD v4.1 (GRCh38 chr14-75049028-G-A): total AF 4.70933e-05 (AC 76 / AN 1,613,818, homozygotes 0); exome AF 4.99376e-05 (73/1,461,824); genome AF 1.97376e-05 (3/151,994); grpmax FAF 4.945e-05; highest ancestry group European (non-Finnish) AF 6.10167e-05 (72/1,180,004, homozygotes 0).
• gnomAD v2.1 (GRCh37 14-75515731-G-A, exomes only): total AF 2.79176e-05 (AC 7 / AN 250,738, homozygotes 0); grpmax FAF 2.298e-05; highest ancestry group European (non-Finnish) AF 5.28811e-05 (6/113,462, homozygotes 0).
• Highest value in any finer subgroup of either dataset: gnomAD v2.1 NFE_SEU 2.61324e-04 (3/11,480 alleles), a small sub-cohort with a wide confidence interval; even this outlier is ~190-fold below 0.05.
• gnomAD-Canada v1.0 (HostSeq genomes): variant absent (search_status 'absent').
• Allele numbers are large enough that a truly common benign allele would not be missed: 1,613,818 v4.1 alleles and 250,738 v2.1 exome alleles give ample power to detect an allele frequency >= 5%, and 0 homozygotes were reported in both datasets.

Sources

Reasoning

BS1 requires an allele frequency greater than that expected for the disorder (Richards et al. 2015, PMID:25741868); with no MLH3 CSPEC/VCEP to provide a gene-specific derivation, the generic strong-strength default of >= 0.01 was applied (the generic SVI default derived for this purpose, Whiffin et al. 2017, PMID:28518168). No MLH3-specific BS1 derivation exists (cspec not found; final_classification_framework.found = false, framework_complete = false), so the all-comers GNOMAD_V2_1 and GNOMAD_V4_1 totals are the governing sources. Observed frequencies: gnomAD v4.1 total 4.71e-05 (76/1,613,818; exomes 4.99e-05, genomes 1.97e-05); gnomAD v2.1 exomes 2.79e-05 (7/250,738); highest ancestry group European (non-Finnish) 6.10e-05 (72/1,180,004, v4.1). The variant is absent from gnomAD-Canada v1.0. Even the single highest subgroup value in either dataset (v2.1 NFE_SEU, 3/11,480 = 2.61e-04) sits approximately 38-fold below 0.01. The allele counts are far too low to signal a benign-allele frequency for a rare Mendelian disorder at any plausible derivation of BS1 from disease prevalence, penetrance and allelic heterogeneity, and read depth is sufficient that low counts are real rather than missed calls. BS1 is not met at strong strength (nor at supporting).

Rule

BS1 strong: allele frequency >= 0.01 (generic SVI default, Whiffin et al. 2017, PMID:28518168); no MLH3 gene-specific derivation available

Evidence considered

• Threshold applied: BS1 strong at allele frequency >= 0.01 (generic SVI default as supplied for this case, derived per Whiffin et al. 2017, PMID:28518168); no MLH3 VCEP or gene-specific derivation exists, so the generic default was used rather than a gene-specific value.
• gnomAD v4.1 (chr14-75049028-G-A): total AF 4.70933e-05 (76/1,613,818, homozygotes 0); exome AF 4.99376e-05 (73/1,461,824); genome AF 1.97376e-05 (3/151,994); grpmax FAF 4.945e-05; European (non-Finnish) 6.10167e-05 (72/1,180,004).
• gnomAD v2.1 exomes (14-75515731-G-A): total AF 2.79176e-05 (7/250,738, homozygotes 0); grpmax FAF 2.298e-05; European (non-Finnish) 5.28811e-05 (6/113,462).
• Highest finer-subgroup value in either dataset: gnomAD v2.1 NFE_SEU 2.61324e-04 (3/11,480); ~38-fold below 0.01, so the BS1 verdict is robust to the choice of population metric (overall AF, grpmax FAF, or maximal subgroup AF).
• Allele counts by ancestry in v4.1 (all small): NFE 72/1,180,004, EAS 2/44,892, AFR 1/74,846, AMR 1/59,972, SAS 0/91,058, ASJ 0/29,600, FIN 0/63,970, MID 0/6,084 - i.e. no ancestry group shows the elevated frequency that BS1 requires.
• gnomAD-Canada v1.0: variant absent.

Sources

Reasoning

BS2 requires the variant to be observed in a well-documented healthy adult for a disorder that is fully penetrant at an early age (Richards et al. 2015, PMID:25741868). Neither arm of this criterion is satisfied. (1) Zygosity: no homozygous carrier has been observed in any dataset examined. gnomAD v4.1 reports 0 homozygotes overall (76 heterozygotes / 1,613,818 alleles) with 0 homozygotes in every ancestry group, including European (non-Finnish) (72/1,180,004) and East Asian (2/44,892); gnomAD v2.1 exomes report 0 homozygotes (7/250,738) with 0 in every ancestry group and every finer subgroup, including the NFE subcohorts. A recessive interpretation of BS2 therefore has no supporting observation, which is the relevant read-out given the documented biallelic loss-of-function MLH3 mechanism (polyposis predisposition, PMID:30573798 as recorded in the gene-level PVS1 context). (2) Penetrance and age of onset: the MLH3-associated phenotype is adult-onset hereditary nonpolyposis colorectal cancer (HNPCC7) rather than a fully penetrant early-onset disorder; the disease-context literature reviewed for this case attributes clinically actionable high-penetrance Lynch syndrome risk to MLH1/MSH2/MSH6/PMS2 only and does not include MLH3 among the path_MMR genes, and the same sources place endometrial cancer risk at adult ages. Heterozygous observations of this variant in a population database therefore cannot be used as BS2 evidence, because BS2 presupposes full penetrance at an early age in a well-documented healthy adult. BS2 is not met.

Rule

BS2 strong: variant observed in a documented healthy adult with full penetrance expected at an early age (Richards et al. 2015, PMID:25741868); not satisfied - 0 homozygotes observed and MLH3 disease is adult-onset

Evidence considered

• Criterion definition applied: BS2 requires a variant observed in a well-documented healthy adult individual for a recessive (homozygous), dominant (heterozygous) or X-linked condition with full penetrance expected at an early age (Richards et al. 2015, PMID:25741868).
• gnomAD v4.1: 0 homozygotes overall (AC 76 / AN 1,613,818); homozygote_count = 0 in every ancestry group, including European (non-Finnish) 0/1,180,004 alleles (72 heterozygotes), East Asian 0/44,892 (2), African/African American 0/74,846 (1), Admixed American 0/59,972 (1), South Asian 0/91,058 (0), Ashkenazi Jewish 0/29,600 (0), Finnish 0/63,970 (0), Middle Eastern 0/6,084 (0), Amish 0/910 (0), Remaining 0/62,482 (0).
• gnomAD v2.1 exomes: 0 homozygotes overall (AC 7 / AN 250,738); homozygote_count = 0 in every ancestry group and in every finer subgroup, e.g. European (non-Finnish) 0/113,462 (6 heterozygotes), NFE_SEU 0/11,480 (3), NFE_XY 0/63,496 (5), Admixed American 0/34,544 (1).
• gnomAD-Canada v1.0: variant absent, so no additional homozygous or healthy-adult observation is available from that cohort.
• Gene/disease context (gene-level PVS1 context, pvs1_gene_context): MLH3 loss of function is a recognised germline disease mechanism, with a documented biallelic germline nonsense mechanism underlying polyposis predisposition (PMID:30573798) alongside the classic adult-onset HNPCC7 association; this is an adult-onset, incompletely penetrant phenotype rather than an early-onset fully penetrant one, so heterozygous population carriers are not BS2-qualifying observations.
• Disease-context literature reviewed for penetrance and age of onset: the EHTG/ESCP Mallorca guideline (PMID:34043773) states 'LS is caused by inherited malfunction in one of the four MMR genes named MLH1, MSH2, MSH6 and PMS2' and never mentions MLH3, and the SGO endometrial cancer review (PMID:33516529) lists the hereditary MMR genes as MLH1, MSH2, MSH6, PMS2 and EPCAM, with no MLH3 content; neither provides MLH3-specific penetrance data, and neither reports this variant, so BS2 cannot be satisfied through them either.
• No homozygous carrier exists in any dataset to test the recessive arm of BS2, and the variant is far too rare (max subgroup frequency 2.61e-04) for an adult healthy-population observation to be considered a full-penetrance, early-onset disqualifier of pathogenicity.

Still missing

• No per-carrier phenotype or age data are available for the 76 heterozygous gnomAD v4.1 carriers (7 in v2.1), so 'healthy adult' status cannot be verified directly from the population data itself; the verdict rests on the absence of any homozygous carrier plus the adult-onset, incompletely penetrant nature of MLH3-associated disease.

Sources

Reasoning

BS3 (Richards et al. 2015, PMID:25741868: a well-established functional study shows no damaging effect on protein function) is not met for MLH3 NM_001040108.2:c.628C>T (NP_001035197.1:p.(Arg210Ter)). No functional study in the case evidence assays this allele at all, and none reports retained or normal function for it. The only MLH3 functional study retrieved (PMID:10615123) used an engineered N-terminal MLH3 deletion construct (DN-MLH3) overexpressed in 293T/COS7 cells, measured microsatellite instability via an ATG(CA)27/29 beta-galactosidase reporter, and reported a positive (MSI-inducing, possibly dominant-negative) result; it performed no wild-type-versus-p.Arg210Ter comparison, used no patient material, contains no normal-function result for c.628C>T, and does not mention the variant. Absence of functional data is not benign functional data: BS3 requires a study demonstrating a non-damaging effect, and no such result exists here. The near-zero SpliceAI max delta (0.001) captured for this variant is not usable as benign support either - splice-delta prediction addresses splice disruption, not the consequence of a premature termination codon, and PP3/BP4 belong to a different criterion group. The OncoKB entry for R210* ('Likely Loss-of-function') points in the opposite direction and is in any case a curation-level inference rather than assay data. No ClinGen CSPEC or gene-specific MLH3 specification exists (final classification framework resolved to the generic ACMG/AMP 2015 fallback), so the generic BS3 definition governs, and under it the criterion is not met for want of any validated assay showing a non-damaging effect on MLH3 function.

Evidence considered

• BS3 requires a well-established functional study demonstrating no damaging effect on protein function (Richards et al. 2015, PMID:25741868); no retrieved source reports such a result for MLH3 c.628C>T.
• PMID:10615123 is the only MLH3 functional study in the case evidence; its readout (beta-galactosidase MSI reporter in 293T/COS7 cells) was positive for an engineered N-terminal deletion construct, reports no wild-type-controlled assessment of the p.Arg210Ter product and never mentions the variant, so it supplies no benign functional evidence.
• The variant is a stop-gained allele in exon 2 of 13; SpliceAI max delta 0.001 indicates no predicted splice effect and is explicitly not treated as benign functional evidence for BS3, because splice prediction does not assess the impact of a premature termination codon.
• OncoKB labels R210* as 'Likely Loss-of-function' (cited to PMID:10615123), which argues against rather than for BS3 and is a curation inference rather than assay data; no criterion was inferred from that classification.
• No ClinGen CSPEC or gene-specific MLH3 specification exists for this case (cspec.found = false; generic ACMG/AMP 2015 fallback), so the generic BS3 definition was applied without gene-specific modification.
• Reviewed full-text extractions for all five fetched papers (PMID:10615123, PMID:25741868, PMID:34043773, PMID:33516529, PMID:24493721): none reports a functional assay of c.628C>T or of any MLH3 missense/truncating allele in patient material, so BS3 gains no support from any source.

Still missing

• No functional assay of c.628C>T exists that could demonstrate a non-damaging (normal or retained) effect, so BS3 is unmet rather than satisfied.
• No validated MLH3 functional assay with benign and pathogenic control variants was identified in any retrieved source, so no negative functional result could be calibrated to BS3.
• MLH3 has no ClinGen CSPEC/VCEP, so no gene-specific BS3 rule or benign functional weighting is available; the generic ACMG/AMP 2015 definition governs.
• PMID:33451724 and PMID:24929052 were never retrieved as full text and could not be screened for functional content.

Sources

Reasoning

BP2 (Richards et al. 2015, PMID:25741868) requires that the variant be 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. Neither configuration is documented for NM_001040108.2:c.628C>T (MLH3 p.Arg210Ter). The three ClinVar submissions for VCV000860534 are all germline clinical-testing submissions that report only a classification (2x Uncertain significance, 1x Pathogenic) and provide no phasing, zygosity, cis/trans or second-variant information, and none is an expert-panel submission. No proband, trio, pedigree or segregation data are present anywhere in the case, pm5_candidates.json contains no MLH3 comparator variants to phase against, and no retrieved full text mentions the variant. Population data give no phase or biallelic information either: homozygote counts are 0 in gnomAD v2.1 (7/250,738 alleles) and 0 in gnomAD v4.1 (76/1,613,818 alleles), and the allele is absent from gnomAD-Canada v1.0. The only available allele-level facts - that the allele is carried heterozygously in unaffected population reference samples at very low frequency - bear on population-based benign evidence rather than on the cis/trans observation BP2 requires. With no phase-resolved co-occurrence with a pathogenic MLH3 variant, BP2 is not met.

Evidence considered

• BP2 criterion 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; no variant-specific or family data are provided by the guideline.
• ClinVar VCV000860534 (NM_001040108.2(MLH3):c.628C>T p.Arg210Ter): three germline clinical-testing submissions, all single-submitter, none expert panel - SCV005920953 (Sinai Health System, Uncertain significance), SCV001231874 (Labcorp Genetics/Invitae, Uncertain significance), SCV002657186 (Ambry Genetics, Pathogenic). None reports a second allele, phase, zygosity, or cis/trans configuration, so no in-cis or in-trans observation can be constructed from the ClinVar record.
• Case-wide search for relatedness evidence: proband, in trans, in cis, segregation, zygosity, compound heterozygous and biallelic do not occur in any case artifact searched (review_manifest, case_summary, prefetch, evidence, literature_pass, fulltext_map, paper_extracts, clinvar_submission_audit); pm5_candidates.json reports 0 MLH3 comparator variants.
• gnomAD v4.1 (GRCh38 chr14-75049028-G-A): total AF 4.70933e-05 (76/1,613,818 alleles), homozygotes 0; highest subpopulation European (non-Finnish) AF 6.10167e-05 (72/1,180,004 alleles), homozygotes 0. No homozygous carrier is observed in any subpopulation, so there is no 'homozygous in an unaffected adult' configuration to invoke for a benign allelic reading.
• gnomAD v2.1 (GRCh37 14-75515731-G-A): total AF 2.79176e-05 (7/250,738 alleles), homozygotes 0; grpmax FAF 2.298e-05. gnomAD-Canada v1.0: absent (search status 'absent').
• No full-text paper in this case mentions c.628C>T, p.(Arg210Ter), R210* or 628C>T: PMID:10615123, PMID:25741868, PMID:24493721, PMID:33516529 and PMID:34043773 were searched for the variant strings, so none of them documents a cis/trans configuration for this allele.
• PMID:10615123 (MLH3 cloning and characterisation) reports microsatellite instability induced by a truncated MLH3 construct in a beta-galactosidase reporter assay and speculates that dominant-negative MLH3 alleles might induce MSI in rare cases; it reports no patient or family carrying a MLH3 truncating allele and does not mention c.628C>T, so it supplies no cis/trans or second-allele evidence for BP2.
• OncoKB curation for MLH3 R210* (Likely Loss-of-function; Likely Oncogenic) carries no phase, zygosity or second-allele information and was not used as allelic evidence.

Still missing

• No cis/trans or zygosity information is available for this variant: neither the ClinVar submissions nor any retrieved paper reports a phase-resolved second MLH3 variant, and no family material was available to the case.
• Homozygosity in unaffected reference individuals (a common alternative route to BP2-type benign support) cannot be assessed because the variant has zero homozygotes in every gnomAD dataset queried (v2.1, v3.1 non-cancer, v2.1 non-cancer, v4.1) and is absent from gnomAD-Canada v1.0.

Sources

Reasoning

BP5 requires that the variant be observed in a case for whom an alternate molecular basis for the disease has been identified (for example a second pathogenic variant in a different disease gene, a confirmed non-genetic cause, or a somatic explanation documented in the proband). No such case exists in the retrieved evidence for MLH3 c.628C>T. The three ClinVar submissions carry no case detail, no co-testing of other disease genes and no proband molecular data, and none of the retrieved literature sources reports a patient or family carrying this variant. The population-database records (gnomAD v2.1 7/250738 alleles; v4.1 76/1613818 alleles, zero homozygotes; absent from gnomAD-Canada) describe unaffected carrier states, not a case with an alternate molecular cause, and the COSMIC entry (COSV53134876, n = 1) is a somatic tumour occurrence rather than a documented alternate explanation for disease in a proband. In the gene-specific context, MLH3 is not among the established Lynch-syndrome genes (MLH1, MSH2, MSH6, PMS2) per PMID:34043773, PMID:33516529 and PMID:24493721, but that weakens the gene-disease premise rather than supplying a documented alternative cause in a specific case, so it is not BP5 evidence. With no case-level alternate-basis observation, BP5 is not met.

Evidence considered

• ClinVar variation 860534 reports no case-level details for any of its three single-submitter laboratory submissions (no alternate-variant testing, no second diagnosis, no proband phenotype), so no alternate molecular basis is documented.
• No retrieved publication reports a patient or family carrying MLH3 c.628C>T; the fetched candidate papers (PMID:24493721, PMID:33516529, PMID:34043773) are gene-level guidelines/reviews that never mention the variant.
• gnomAD v2.1 (7/250738 alleles, AF 2.79176e-05, 0 homozygotes), gnomAD v4.1 (76/1613818 alleles, AF 4.70933e-05, 0 homozygotes) and gnomAD-Canada (absent) describe asymptomatic population carriers; carrier status in a control cohort is not the case-with-alternate-cause scenario BP5 requires.
• COSMIC records one somatic occurrence of this variant (COSV53134876); a somatic tumour record is not a documented alternate germline molecular basis for disease in a proband, and COSMIC has no source_registry key for citation.
• PMID:34043773, PMID:33516529 and PMID:24493721 restrict the hereditary MMR-cancer phenotype to MLH1, MSH2, MSH6, PMS2 and EPCAM, which bears on the strength of the MLH3 gene-disease premise rather than constituting BP5 evidence in any individual case.
• No ClinGen CSPEC or MLH3-specific VCEP specification exists (cspec.found = false; generic ACMG/AMP 2015 fallback).

Still missing

• No case-level data exist for any carrier of c.628C>T (no co-testing of other disease genes, no tumour MMR/MSI results, no non-genetic or somatic explanation documented in a proband).

Sources

Reasoning

BP6 applies only when the exact variant has been classified Benign or Likely benign by a ClinVar expert panel. ClinVar variation 860534, NM_001040108.2(MLH3):c.628C>T (p.Arg210Ter), has three submissions and zero expert-panel submissions; no Benign or Likely benign assertion exists at any tier, expert or ordinary, and the aggregate ClinVar label is Uncertain significance. The three submissions are single-laboratory assertions classified Uncertain significance (Sinai Health System, Labcorp/Invitae) and Pathogenic (Ambry Genetics). Ordinary laboratory submissions and aggregate ClinVar labels must never trigger BP6, and here they would not in any case because none asserts benignity. No numeric threshold or comparison operator applies; the criterion is an applicability gate on the existence of an exact-variant expert-panel Benign/Likely benign classification, which does not exist. BP6 is therefore not met.

Evidence considered

• ClinVar variation 860534 (exact match to NM_001040108.2:c.628C>T, p.Arg210Ter): expert_panel_submissions = 0, aggregate classification Uncertain significance, and the three submitting laboratories report Uncertain significance (Sinai Health System), Uncertain significance (Labcorp/Invitae) and Pathogenic (Ambry Genetics).
• No Benign or Likely benign assertion for this variant exists at any review tier, so there is neither an expert-panel classification nor even a lower-tier benign signal for BP6 to act on.
• Ordinary laboratory submissions and non-expert aggregate ClinVar labels are explicitly not usable for BP6, consistent with the criterion's requirement of an expert-panel assertion.
• No ClinGen CSPEC or MLH3-specific VCEP specification exists (cspec.found = false; generic ACMG/AMP 2015 fallback), and no gene-specific BP6 rule applies.

Still missing

• No ClinVar expert-panel classification exists for this variant, and no Benign or Likely benign submission exists at any review tier.

Sources

Reasoning

PS2 (de novo occurrence, confirmed parental relationships) requires an affected proband carrying the variant whose parents have been tested and shown not to carry it. This case contains no proband at all: it is a variant-level evaluation assembled from population, database and literature sources, with no patient record, no parental genotypes, no trio and no stated inheritance of c.628C>T (NM_001040108.2, NP_001035197.1:p.(Arg210Ter), MLH3). The ClinVar record VCV000860534 (ID 860534) carries three germline submissions (Ambry Genetics; Department of Pathology and Laboratory Medicine, Sinai Health System; Labcorp Genetics/Invitae), all classified Uncertain significance with 'criteria provided, single submitter' review status and no expert-panel review; none of the submission summaries reports a de novo occurrence or parental testing, and the validated PMID trails cited by those submitters (PMID:25741868, PMID:28492532) are generic criteria/methodology papers rather than family reports. Full-text review of every paper fetched for this case (PMID:10615123, PMID:24493721, PMID:25741868, PMID:33516529, PMID:34043773) found no mention of c.628C>T, p.Arg210Ter or R210*, and no family carrying a MLH3 truncating allele; the abstracts retrieved for additional MLH3 papers report biallelic (recessive) polyposis predisposition and intronic sequencing cohorts rather than a de novo occurrence of this variant. No ClinGen CSPEC or gene-specific MLH3 specification exists (final classification framework resolved to the generic ACMG/AMP 2015 fallback), so there is no gene-specific modification of PS2 that could be applied either. The criterion is therefore not assessable from the available evidence; this is a data gap, not evidence that the variant is inherited or that PS2 is refuted.

Evidence considered

• No proband is described for this variant: the case holds only variant-level database and literature evidence, with no patient record, no trio, no parental genotypes and no reported relatives.
• ClinVar VCV000860534 (variation ID 860534, updated 2026-02-15) has three germline submissions - Ambry Genetics (SCV002657186), Sinai Health System (SCV005920953) and Labcorp Genetics/Invitae (SCV001231874) - all Uncertain significance, 'criteria provided, single submitter'; no submission reports a de novo occurrence, parental testing or family segregation data, and there is no expert-panel (VCEP) review.
• The two submissions flagged as criterion leads cite only PMID:25741868 (ACMG/AMP 2015 standards) and PMID:28492532 (Sherloc criteria refinement), i.e. generic interpretation-methodology papers that contain no case or family data for this variant.
• Searched all five fetched full texts (PMID:10615123, PMID:24493721, PMID:25741868, PMID:33516529, PMID:34043773) for c.628C>T, NP_001035197.1:p.(Arg210Ter) and R210*: none of them mentions the variant, and a search of the same full texts for 'de novo', 'segregation', 'proband' and 'pedigree' returned no hits.
• Generic ACMG/AMP 2015 PS2 requires a confirmed de novo occurrence in a patient with the relevant disease plus confirmed parental relationships (paternity/maternity); with no proband genotype and no parental testing, the criterion cannot be evaluated in either direction.
• No ClinGen CSPEC or gene-specific MLH3 specification is available for this case, so the generic ACMG/AMP 2015 PS2 definition is the governing rule; no gene-specific de novo weighting applies.

Still missing

• No proband genotype: the case contains no patient carrying c.628C>T.
• No parental testing: neither parent of any carrier has been tested, so a de novo event cannot be confirmed.
• No published family with this variant: none of the five full texts reviewed mentions c.628C>T, p.Arg210Ter or R210*.

Sources

Reasoning

PM6 (assumed de novo occurrence, parental testing not performed or not confirmed) applies to an affected proband carrying the variant whose parents are reportedly unaffected or untested. This case provides no proband and no parental clinical or genotype information whatsoever - neither an affected index patient, nor parents reported as unaffected, nor any statement about whether parental testing was performed for c.628C>T. The ClinVar record VCV000860534 lists three germline submissions (Ambry Genetics; Sinai Health System; Labcorp Genetics/Invitae) but records only classification, review status and validated PMID trails (PMID:25741868, PMID:28492532) - no pedigree, no parental phenotype and no de novo statement - and the screenshots/records collected for this case contain no clinical or family section. None of the five fetched full texts (PMID:10615123, PMID:24493721, PMID:25741868, PMID:33516529, PMID:34043773) mentions the variant or any family carrying a MLH3 truncating allele, and searching those full texts for 'de novo', 'proband', 'pedigree' and 'segregation' produced no hits. With no proband and no parental information, PM6 cannot be assigned even at its reduced, assumed-de-novo weight; the criterion is not assessable, which is a data gap rather than a negative result.

Evidence considered

• PM6 requires an affected proband with the variant and parents who are assumed (not confirmed) not to carry it; this case supplies neither a proband nor any parental phenotype or testing status for c.628C>T.
• ClinVar VCV000860534 (ID 860534) records three germline submissions (Ambry Genetics, Sinai Health System, Labcorp Genetics/Invitae) with classifications and validated PMID trails only; no submission text, family history or de novo assertion is available in the collected record.
• The PMIDs attached to the two criterion-lead submissions (PMID:25741868, PMID:28492532) are ACMG/AMP criteria and Sherloc methodology documents, not family or case reports for this variant.
• Searched the five fetched full texts (PMID:10615123, PMID:24493721, PMID:25741868, PMID:33516529, PMID:34043773) for c.628C>T, p.Arg210Ter and R210* with no hits, and searched them for 'de novo', 'proband', 'pedigree' and 'segregation' with no hits; no assumed de novo occurrence of this variant is published.
• The only gene-level functional paper (PMID:10615123) reports dominant-negative behaviour of a truncated MLH3 construct in cell culture and discusses HNPCC speculation, but reports no patient, no family and no de novo occurrence of any MLH3 variant.
• No ClinGen CSPEC or gene-specific MLH3 specification exists for this case, so the generic ACMG/AMP 2015 PM6 definition governs and no gene-specific modification is available.

Still missing

• No proband: no index patient is described, so the affected-individual requirement of PM6 cannot be tested.
• No parental phenotype or testing status: it is unknown whether parents exist, are unaffected, or were tested.
• No published family report linking c.628C>T to a de novo or assumed-de-novo occurrence.

Sources

Reasoning

PP1 (co-segregation with disease in multiple affected family members) requires a pedigree in which the variant tracks with disease across at least two affected relatives, with the number of informative meioses determining whether supporting, moderate or strong weight is justified. This case provides no pedigree, no affected relatives and no genotype data for anyone: there is no proband, no family structure, and no statement of inheritance for c.628C>T. The three ClinVar submissions on VCV000860534 describe germline origin only and give no family data or segregation counts, and the papers the case retrieved for family-history and cancer-predisposition purposes (PMID:24493721, PMID:33516529, PMID:34043773) are consensus/guideline documents that name only MLH1, MSH2, MSH6, PMS2 and EPCAM as Lynch-syndrome genes; none mentions MLH3, let alone this variant or a family carrying it. The two remaining full texts reviewed (PMID:25741868, PMID:10615123) contain no family segregation data either. Because the requirement is co-segregation in a family that does not exist in the available evidence, no meioses can be counted and PP1 cannot be scored at any strength; this is a data gap rather than evidence of non-segregation.

Evidence considered

• PP1 requires co-segregation of the variant with disease in multiple affected family members, with the strength set by the number of informative meioses; this case contains no pedigree and no genotype data for any relative.
• ClinVar VCV000860534 lists three germline submissions (Ambry Genetics, Sinai Health System, Labcorp Genetics/Invitae) but no family history, pedigree or segregation counts are recorded; no submission asserts co-segregation.
• The papers the case routed to the family-history/case-observation theme (PMID:24493721 ASCO cancer family history statement; PMID:33516529 SGO endometrial cancer review; PMID:34043773 EHTG/ESCP Lynch syndrome guidelines) were each read in full and none mentions MLH3 or c.628C>T - e.g. the Mallorca guideline states that Lynch syndrome is caused by inherited malfunction in MLH1, MSH2, MSH6 and PMS2 only, so these documents supply no variant-level segregation evidence.
• PMID:25741868 supplies the generic ACMG/AMP 2015 definition of PP1 (co-segregation in multiple affected family members) but contains no variant-specific or family data for MLH3.
• Gene-level context collected for MLH3 records an empty inheritance field and no official CSPEC/VCEP or local framework, so no gene-specific PP1 rule or segregation weighting is available for this case.

Still missing

• No pedigree or family structure for any carrier of c.628C>T.
• No genotyped affected relatives, therefore no informative meioses for segregation analysis.
• No published family report of c.628C>T tracking with disease.

Sources

Reasoning

PP4 requires that the patient's phenotype or family history be highly specific for a disease with a single genetic etiology. No proband-level information is recorded anywhere in this case: the case summary, prefetch and evidence records contain only variant-level data (HGVS normalization, ClinVar submissions, gnomAD frequencies, SpliceAI, BayesDel, OncoKB, COSMIC, Cancer Hotspots). There is no disease presentation, no tumour MSI/MMR immunohistochemistry result, no age at diagnosis, no family history, and no indication for testing, so the criterion's required input is absent and PP4 cannot be evaluated. Independently of that missing input, the retrieved guideline sources show the associated phenotype is not specific to MLH3: PMID:34043773 states that Lynch syndrome is caused by inherited malfunction in one of the four MMR genes MLH1, MSH2, MSH6 and PMS2 and stratifies all risk and surveillance by those four genes alone; PMID:33516529 lists hereditary endometrial cancer predisposition as PTEN (Cowden syndrome) and the Lynch genes MLH1, MSH2, MSH6, PMS2 and EPCAM; PMID:24493721 lists the MMR loci warranting testing as MLH1/MSH2/MSH6/PMS2/EPCAM. MLH3 appears in none of these lists, so even a colorectal or endometrial cancer presentation would not be highly specific for an MLH3 variant. PMID:10615123 describes MLH3 as an MMR gene interacting with MLH1 and speculates that dominant-negative MLH3 alleles might induce microsatellite instability and HNPCC in rare instances, which is a gene-level hypothesis and supplies no phenotype-specificity data for this variant. PP4 is therefore not_assessed rather than not_met: the deficiency is a missing patient-level input, not a demonstrated negative.

Evidence considered

• PMID:34043773 (EHTG/ESCP Mallorca guidelines) defines Lynch syndrome as caused by inherited malfunction in one of four MMR genes, MLH1, MSH2, MSH6 and PMS2 (path_MMR), with all risk, surveillance and surgical guidance stratified by those genes; MLH3 is not among them and the paper never mentions MLH3 or c.628C>T.
• PMID:33516529 (SGO endometrial cancer review) attributes hereditary endometrial cancer predisposition to PTEN (Cowden syndrome) and to the Lynch genes MLH1, MSH2, MSH6, PMS2 and EPCAM; MLH3 is not named.
• PMID:24493721 (ASCO family-history expert statement) lists the mismatch-repair loci for which germline testing should be considered as MLH1/MSH2/MSH6/PMS2 (plus EPCAM for colorectal); MLH3 is not listed and the variant is not mentioned.
• PMID:10615123 describes MLH3 as a mismatch repair gene whose protein interacts with MLH1 and hypothesises that dominant-negative MLH3 mutations might induce MSI and HNPCC in rare instances; it provides no patient phenotype data and no phenotype-specificity evidence for c.628C>T.
• Case records contain variant-level evidence only (HGVS normalization, ClinVar, gnomAD, SpliceAI, BayesDel, OncoKB, COSMIC, Cancer Hotspots) and no proband phenotype, tumour MMR/MSI status or family history.
• No ClinGen CSPEC or MLH3-specific VCEP specification exists (cspec.found = false; generic ACMG/AMP 2015 fallback), so no gene-specific PP4 phenotype checklist is available.

Still missing

• No proband phenotype, tumour MSI/MMR-IHC result, age at diagnosis, family history, or testing indication is recorded for this case, so phenotype specificity for c.628C>T cannot be evaluated.
• No published case-level phenotype description for this exact variant was found in the retrieved sources.

Sources

Reasoning

BS4 (lack of segregation in affected family members) requires genotype data from an informative family in which affected individuals do not carry the variant, using the same meioses-based framework as PP1 in the benign direction. This case has no pedigree, no genotyped relatives and, since there is no proband either, no family in which non-segregation could be observed for c.628C>T. Nothing in the collected evidence reports an affected relative who lacks the variant: the ClinVar record VCV000860534 contains three germline classifications without pedigree detail, and the retrieved papers (PMID:24493721, PMID:33516529, PMID:34043773, PMID:25741868, PMID:10615123) neither mention the variant nor report families typed for it - the MLH3-focused functional paper reports cell-culture work on a truncated MLH3 construct with no patient cohort at all. Neither segregation nor non-segregation can therefore be demonstrated, and BS4 cannot be scored; the absence of an assertion of non-segregation must not be read as support for BS4, and equally the criterion is not refuted by the missing data.

Evidence considered

• BS4 requires demonstrated lack of segregation of the variant with disease in affected family members; with no pedigree, no proband and no genotyped relatives, non-segregation cannot be observed for c.628C>T.
• ClinVar VCV000860534 contains three germline submissions (Ambry Genetics, Sinai Health System, Labcorp Genetics/Invitae) and no pedigree, genotype matrix or statement that any affected relative lacks the variant.
• Full-text review of PMID:24493721, PMID:33516529 and PMID:34043773 (the family-history and cancer-predisposition papers retrieved for this case) found no MLH3 content and no family typed for this variant; PMID:34043773 states that Lynch syndrome is caused by MLH1, MSH2, MSH6 and PMS2, and PMID:33516529 lists the Lynch genes as MLH1, MSH2, MSH6, PMS2 and EPCAM.
• PMID:10615123, the only MLH3-specific full text retrieved, is an in vitro/in vivo functional study (truncated MLH3 construct inducing microsatellite instability in 293T/COS7 reporters) with no patient cohort and therefore no segregation or non-segregation data.
• PMID:25741868 provides the generic ACMG/AMP 2015 definition of BS4 but no variant-level data; no ClinGen CSPEC or gene-specific MLH3 specification exists for this case, so the generic definition governs.

Still missing

• No family pedigree or genotyped relatives in which a non-segregating affected individual could be identified.
• No published report of an affected individual lacking c.628C>T.
• No proband exists in this case, so any interpretation-level family data would have to be supplied externally.

Sources