BRCA2 c.8432A>G, p.(Asp2811Gly)

Reasoning

Governing framework: ENIGMA BRCA1/BRCA2 specification v1.2 (BRCA2). BS1 is met at strong strength when the maximum filter allele frequency (FAF) in a gnomAD non-founder population exceeds 0.0001 (0.01%); BS1_Supporting applies for FAF >0.00002 (0.002%) up to and including 0.0001 (0.01%). Assessment uses gnomAD v2.1 non-cancer exomes and gnomAD v3.1 non-cancer genomes separately, on maximum non-founder-population FAF (Appendix G). For BRCA2 c.8432A>G (p.Asp2811Gly), the maximum non-founder-population frequency is 5.65e-05 (1/17,688 alleles, East Asian) in gnomAD v2.1 non-cancer exomes, which the governing ENIGMA SupplementaryTables V1.2 records as a Maximum MAF of 5.653550429669833e-05 under the frequency category '>0.00002 MAF <=0.0001'. This value exceeds the 0.002% (0.00002) supporting threshold but does not exceed the 0.01% (0.0001) strong threshold, while the v3.1 non-cancer genome maximum (1.54e-05, European non-Finnish) and the gnomAD v4.1 total AF (9.91e-06) remain below the supporting band. BS1 is therefore met at supporting strength (BS1_Supporting), not strong. No founder-population caveat applies as the qualifying population is East Asian, which is not one of the BRCA2 founder populations excluded by the specification.

Rule

ENIGMA BRCA2 v1.2 BS1: maximum non-founder-population FAF > 0.0001 in gnomAD v2.1 non-cancer exomes and/or v3.1 non-cancer => strong; > 0.00002 and <= 0.0001 => BS1_Supporting.

Evidence considered

• CSPEC ENIGMA BRCA2 v1.2 BS1 rules: strong when 'FAF is above 0.01% (FAF > 0.0001)'; supporting when 'FAF is above 0.002% (FAF > 0.00002) and less than or equal to 0.01% (FAF <= 0.0001)' in gnomAD v2.1 (non-cancer, exome only subset) and/or gnomAD v3.1 (non-cancer), non-founder population(s).
• Specifications V1.2 Table 1: 'BS1 - above 0.0001 (0.01%) BS1_Supporting - >0.00002 (0.002%) to <= 0.0001 (0.01%)'; apply on maximum filter allele frequency in a gnomAD non-founder population, exome and genome assessed separately.
• Appendix G: BS1 frequency derived with the Whiffin calculator assuming prevalence 0.125, genetic heterogeneity 0.01, allelic heterogeneity 0.10, penetrance 0.69 for BRCA2; minimal credible AF 0.0000906 rounded up to 0.0001; LR-based approach additionally supports BS1_Supporting for MAF >0.00002 to <=0.0001; use highest filter allele frequency in a non-founder population (TCGA excluded), exome and genome separately.
• gnomAD v2.1 non-cancer exomes: maximum non-founder-population AF 5.653550429669833e-05 (East Asian, 1/17,688 alleles); all other populations 0 in this subset.
• ENIGMA SupplementaryTables V1.2 row for BRCA2 c.8432A>G p.(Asp2811Gly): frequency category '>0.00002 MAF <=0.0001', Maximum MAF 5.653550429669833e-05 (East Asian), gnomAD V2.1.1 non-cancer. This pre-assigned frequency band is exactly the BS1_Supporting band.
• gnomAD v3.1 non-cancer genome maximum AF 1.54297e-05 (European non-Finnish, 1/64,810 alleles) sits below the BS1_Supporting lower bound.
• HUMU-40-1557-s001 (Parsons et al. 2019, PMID:31131967) SuppT1 row for BRCA2 c.8432A>G p.(Asp2811Gly): reference-set name '(Likely) Benign', prior 0.03, posterior probability of pathogenicity 0.0131239078, IARC class 2, frequency category '>0 & <0.0001', population of maximum East Asian.

Still missing

• gnomAD read-depth at the variant site was not captured; BS1 requires read depth >=20 for applicability. This does not change the verdict materially: the qualifying frequency is only a single observation (1/17,688 EAS alleles in v2.1 non-cancer exomes), so the value is near the lower edge of the supporting band and could move within/just below the band if depth QC were applied.

Sources

Reasoning

ENIGMA BRCA1/BRCA2 V1.2 uses BP5_Variable Weight solely to capture a combined clinical likelihood ratio against pathogenicity from multifactorial likelihood clinical data, with the supporting code defined as 'BP5 - LR <=0.48 to >0.23:1' (BP5_Moderate <=0.23:1, BP5_Strong <=0.05:1, BP5_VeryStrong <=0.00285:1), and with combined LR >0.48-1.00 explicitly not informative (BP5 not applicable). The published ENIGMA/Parsons 2019 multifactorial result for this exact variant gives component LRs Co-occurrence 1.0756788212 and Family History 0.39973152851, whose product is the Combined clinical LR 0.4299827393629182 (HUMU-40-1557-s001 SuppT1); the ENIGMA reference-set table independently classifies the same variant as '(Likely) Benign' with posterior probability 0.0131239078 and IARC class 2 (SupplementaryTables_V1.2 ST7; Parsons et al. 2019). This is structurally identical to ENIGMA Appendix B Table 2 Example 5, where a Parsons-derived combined clinical LR of 0.015 (co-occurrence 1.177 x family history 0.0127) is assigned BP5_Strong. Explicit comparison in this criterion's own direction (BP5 strengthens as the LR goes DOWN): 0.430 <= 0.48? yes; and 0.430 > 0.23? yes - so the value falls in the supporting BP5 band and BP5 is met at Supporting strength. The BP5 'not applicable for co-observation' stipulation (cases carrying pathogenic variants in two or more different breast-ovarian cancer risk genes) is not engaged here: the co-occurrence component is co-occurrence with a pathogenic variant in the same gene (BRCA2), which is the VCEP's own allowed component, and the family-history component alone (LR 0.3997) already falls inside the supporting band, so the call does not depend on the co-occurrence component.

Rule

ENIGMA BRCA1/BRCA2 V1.2 BP5 (Variable Weight): LR <=0.48 to >0.23 supporting; <=0.23 moderate; <=0.05 strong; <=0.00285 very strong; combined LR >0.48-1.00 not informative.

Evidence considered

• ENIGMA BRCA1/BRCA2 V1.2 Table 1 BP5: 'Use ONLY to capture combined LR against pathogenicity, based on multifactorial likelihood clinical data. BP5_VeryStrong - LR <=0.00285:1; BP5_Strong - LR <=0.05:1; BP5_Moderate - LR <=0.23:1; BP5 - LR <=0.48 to >0.23:1. Combined LR >0.48-1.00 is not informative (BP5 not applicable).'
• ENIGMA Appendix B assigns calibrated multifactorial clinical data weights under PP4 (towards pathogenicity) or BP5 (against pathogenicity) by combined LR, and Appendix B Table 2 example 5 applies BP5_Strong to a Parsons et al. 2019 combined clinical LR of 0.015 (co-occurrence 1.177 x family history 0.0127), establishing the direct precedent for using the Parsons combined clinical LR under BP5.
• HUMU-40-1557-s001 SuppT1 exact-variant row (Gene BRCA2, HGVS Nucleotide c.8432A>G, HGVS Protein p.(Asp2811Gly), legacy 8660A>G / D2811G): Co-occurrence LR 1.0756788212, Family History LR 0.39973152851, Combined LR (Odds for Causality) 0.4299827393629182, Prior Probability 0.03, Posterior Probability 0.01312390780007919, IARC Class 'Likely Benign', comment 'Multifac new calc'.
• SupplementaryTables_V1.2 ST7 reference-set row for the same variant: '(Likely) Benign', prior 0.03, posterior 0.0131239078, IARC class 2, source 'Parsons et al. 2019 (PMID 31131967)' - a second ENIGMA-governed table giving the same posterior probability and a benign class.
• PMID:31131967 (Parsons et al. 2019) is the underlying ENIGMA multifactorial likelihood resource; the article text does not itself name c.8432A>G, but the authoritative per-variant values are carried in its supplementary tables as reproduced in the ENIGMA v1.2 tables searched here.
• PMID_31853058_BRCA2_clinical_history_LR.xlsx.txt exact match: BRCA2 c.8432A>G with LOG(LR) -0.03127995133399963, N_Probands 2, LR = exp(LOG(LR)) 0.9692042050827514 - a neutral clinical-history LR component (inside the >0.48 and <2.08 non-informative zone) that is not by itself BP5-supporting.
• Explicit threshold comparison using BP5's own operator direction (weight increases as LR decreases): 0.430 <= 0.48? yes; 0.430 > 0.23? yes. The value therefore sits in the supporting band; it is not <=0.23, so BP5_Moderate is not claimed.
• PMID:17924331 (Easton et al. 2007) was consulted for the multifactorial methodology underlying the combined clinical LR (independent co-occurrence, family-history and co-segregation LRs multiplied into a combined LR).

Still missing

• Segregation LR, tumour-pathology LR and case-control LR components are blank for this variant in HUMU-40-1557-s001 SuppT1, so the combined clinical LR of 0.430 rests on the co-occurrence and family-history components only; the Li et al. 2020 clinical-history LR for the same variant (0.969) is neutral and was not combined with the ENIGMA value to avoid double-counting family-history evidence.
• The variant-level source row in PMID_17924331_Easton_2007_AJHG.pdf.txt could not be located in the converted text, so the Easton-derived component of the Parsons combined LR could not be audited directly.

Sources

Reasoning

The ENIGMA BRCA1/2 v1.2 specification applies PS1 (Strong) for a predicted missense substitution that leads to the same amino-acid change as a previously classified pathogenic variant acting via protein change (SpliceAI <= 0.1); PS1_Moderate applies when the comparator is only (likely) pathogenic. No qualifying comparator exists for p.Asp2811Gly. This exact change, c.8432A>G p.(Asp2811Gly), is itself classified (Likely) Benign by the ENIGMA multifactorial likelihood analysis (combined LR 0.03, posterior probability of pathogenicity 0.0131), and ClinVar (VariationID 52585) contains only Uncertain significance (7 laboratories) and Likely benign/Benign (4 laboratories) submissions with no pathogenic submission and no expert-panel submission. The catalogued known pathogenic missense residues within the BRCA2 DNA-binding domain (Appendix Table 4) are T2607P, W2626C, I2627F, E2663V, D2723H, G2748D, I2778_Q2829del, P2992_T3033del42, R3052W and N3124I; Asp2811 is not among them, so there is no established pathogenic variant sharing this amino-acid change. One paper reporting a glycine substitution at codon 2811 (PMID:10717622) assigns a different wild-type residue (asparagine, i.e. N2811G), so it is not the same amino-acid change and cannot support PS1. PS1 is therefore not met.

Rule

ENIGMA v1.2 Table 1: PS1 (Strong) - same amino-acid change as a previously established pathogenic missense variant (acting via protein change, SpliceAI <= 0.1); PS1_Moderate for a previously established (likely) pathogenic comparator.

Evidence considered

• ENIGMA BRCA1/2 specification v1.2, Table 1 (STRONG CRITERIA, PS1_Variable Weight): PS1 (Strong) for a predicted missense substitution where a previously classified pathogenic variant is considered to act via protein change (no confirmed or predicted splicing effect, SpliceAI <= 0.1); PS1_Moderate where the comparator is (likely) pathogenic. Searched the specification text for 'PS1', 'c.8432A>G', 'p.Asp2811Gly', 'D2811G' and '2811'.
• ENIGMA SupplementaryTables_V1.2_2024-11-18.xlsx, row for BRCA2 c.8432A>G / p.(Asp2811Gly): IARC class '(Likely) Benign', combined LR 0.03, posterior probability of pathogenicity 0.0131, prior 0.03, source 'Parsons et al. 2019 (PMID 31131967)'.
• HUMU-40-1557-s001.xlsx (Parsons et al. 2019 / ENIGMA multifactorial dataset), SuppT1 row for BRCA2 c.8432A>G D2811G: Family History LR 0.3997, Combined LR 0.4300, prior probability 0.03, posterior probability 0.01312, IARC class 'Likely Benign'; SuppT2 lists c.8432A>G within the Easton et al 2007 dataset.
• ClinVar VariationID 52585: 7 laboratories report Uncertain significance, 3 report Likely benign and 1 reports Likely Benign; no pathogenic submission, no expert-panel submission.
• Appendices_V1.2_2024-11-18.docx, Appendix Table 4 (BRCA2 conserved/functional domains and their known pathogenic missense residues): DBD (DNA/DSS1 binding domain - helical, OB1, OB2, OB3) aa 2481-3186 lists T2607P, W2626C, I2627F, E2663V, D2723H, G2748D, I2778_Q2829del, P2992_T3033del42, R3052W, N3124I. Asp2811 is not listed.
• paper_extracts.json entry for PMID:10717622 records a glycine change at codon 2811 but with wild-type residue asparagine (not aspartate), i.e. a different amino-acid change from p.Asp2811Gly; variant_mentioned=false for the exact HGVS.

Sources

Reasoning

The variant is a single missense substitution, NM_000059.4:c.8432A>G (p.Asp2811Gly), whose codon 2811 sits inside the BRCA2 DNA-binding domain (aa 2481-3186), a clinically important functional domain in the ENIGMA v1.2 specification. PS3 for BRCA2 is reserved for protein-function assays (assay measuring effect via protein only, or mRNA and protein combined), with strength taken from the calibrated mammalian assay set; the governing look-up is Specifications Table 9, which is the ENIGMA-curated summary of published calibrated functional assay results and pre-assigned PS3/BS3 weights. Table 9 contains no row for this variant: the BRCA2 entries in the 2800-codon region include c.8420C>T, c.8428A>G, c.8435G>A, c.8438G>A, c.8443G>A, c.8447G>A, c.8450G>T, c.8455G>C and c.8456A>T, but not c.8432A>G (searched as c.8432A>G, p.Asp2811Gly, p.D2811G and legacy 8660A>G; zero matches for '8432', '2811', 'D2811' or 'Asp2811' anywhere in the file). The Parsons et al. 2019 ENIGMA dataset likewise carries no functional assay result for this variant: its per-variant row (HUMU-40-1557-s001 SuppT1, 'BRCA2 c.8432A>G p.(Asp2811Gly) 8660A>G D2811G') has every functional-assay column blank (Functional Paper/s, Functional Category Assigned, Findlay, Starita, Mesman, Bouwman, Fernandes2019, Petitalot2019, Hart2018), and the variant is absent from the functional-assay cross-reference tables SuppT4 and Supplementary ST4 (the neighbouring c.8428A>G p.(Ser2810Gly) IS present in both, listed as 'No functional impact', Hart, confirming the tables are per-variant and that c.8432A>G was simply not assayed). The only ENIGMA variant-level data for this variant are multifactorial likelihood clinical components (co-occurrence LR 1.0757, family-history LR 0.3997, combined LR 0.43, prior 0.03, posterior 0.0131, IARC class 2/likely benign); the ENIGMA specification states explicitly that clinical data from calibrated multifactorial likelihood analysis are assigned under PP4/BP5, not PS3/BS3, and Table 9/Appendix E reserve PS3 for assays of protein function, so this cannot be converted into PS3. Literature screening found no functional assay of this variant: PMID:10717622 reports a substitution at BRCA2 codon 2811 but performed no assay (it is a PCR/SSCP population screen that explicitly states missense significance cannot be established without a functional assay, and its description of the change as 'asparagine to glycine' in the 'BRCT domains' conflicts with p.Asp2811Gly in the DNA-binding domain), and no other fetched publication names the variant (PMID:31112341 and ClinVar-flagged interpretation-methodology papers are gene-level or in-silico only). Because no well-established functional study demonstrates a damaging effect on BRCA2 protein function, PS3 is not met, at any strength.

Rule

ENIGMA BRCA1/BRCA2 VCEP specification v1.2, Table 1 PS3 row (functional assays, protein or mRNA+protein, with Table 9 as the go-to curated list); Appendix E and Table 10 define PS3 strength calibration (complete impact in >=1 study with no conflict = Strong). No pre-assigned code exists for this variant.

Evidence considered

• ENIGMA BRCA1/BRCA2 specification v1.2, Table 1, PS3 row: PS3 (functional assays) requires well-established in vitro or in vivo functional studies supportive of a damaging effect, where the assay measures effect via protein only OR mRNA and protein combined, and directs the user to Table 9 for the comprehensive table of applicable codes from published calibrated functional assay results.
• ENIGMA specification v1.2 footnote ** and Appendix C: (potentially) clinically important functional domains include BRCA2 DNA binding aa 2481-3186; p.Asp2811Gly (codon 2811) therefore lies inside a clinically important functional domain, which does not itself generate PS3 but does mean protein-level assay data are required before any functional code can be applied.
• Specifications_Table9_V1.2_2024-11-18.xlsx (sheet Table9_BRCA12VCEP_specs): no row for c.8432A>G / p.(Asp2811Gly). Directly adjacent curated rows exist for c.8420C>T p.(Ser2807Leu), c.8428A>G p.(Ser2810Gly) (BS3 Strong), c.8435G>A p.(Gly2812Glu), c.8438G>A p.(Gly2813Glu), c.8443G>A p.(Val2815Ile) (BS3 Strong), c.8447G>A p.(Gly2816Asp) (BS3 Strong), c.8450G>T p.(Cys2817Phe) (BS3 Strong), c.8455G>C p.(Asp2819His) (PS3 Strong) and c.8456A>T p.(Asp2819Val) (PS3 Strong).
• Appendices v1.2 Appendix E: PS3 and BS3 codes are reserved for assigning variant impact on protein function and must take into account assay design, variant location and predicted/observed splicing; Table 10 calibration showed complete impact in at least one study (with no conflict) provides Strong evidence for pathogenicity, and no impact in at least one study (with no conflict) provides Strong evidence against pathogenicity - neither situation applies here because there is no assay result.
• HUMU-40-1557-s001.xlsx SuppT1, row for BRCA2 c.8432A>G p.(Asp2811Gly) (legacy 8660A>G D2811G): co-occurrence LR 1.0757, family-history LR 0.3997, combined LR 0.43, prior probability 0.03, posterior probability 0.0131, IARC class 2 (Likely Benign), comment 'Multifac new calc'; all functional-assay columns (Functional Paper/s, Functional Category Assigned, Findlay_Function/RNAscore/RNAclass, Starita Depleted, Mesman_Complementation/HDR/Cisplatin/Class, Bouwman_Selection/FxnClass, Fernandes2019_fClass/fCategory, Petitalot2019_ControlGroup/Class, Hart2018_Functional Class/HDR ratio) are empty.
• ENIGMA specification v1.2, 'Multifactorial likelihood analysis' section: given the breadth of data types included, evidence weights for clinical data derived from calibrated multifactorial likelihood analysis approaches are assigned using PP4 (combined clinical evidence towards pathogenicity) or BP5 (combined evidence against pathogenicity); bioinformatic priors are excluded. This is why the Parsons c.8432A>G multifactorial result cannot be read as PS3.
• HUMU-40-1557-s001.xlsx SuppT4 'Fxnal Assay Data' and SupplementaryTables_V1.2 ST4 'Fxn assay results': no c.8432A>G row. The immediately adjacent BRCA2 variant c.8428A>G p.(Ser2810Gly) is present in both (SuppT4: posterior 0.00395, IARC class 2, 'No functional impact', Hart; ST7/ST4: listed with Hart functional data), demonstrating that a missing c.8432A>G entry reflects absence of assay data rather than a lookup failure.
• PMID:10717622 (Malone et al. 2000): reports a substitution at BRCA2 codon 2811 but performed no functional assay - 'For BRCA2 these include an asparagine to glycine change at amino acid 2811 and the glutamic acid to alanine change at amino acid 2856, both with the BRCT domains. (Results, Missense Changes)'; the paper classifies conserved-position missense changes as unclassified variants because significance cannot be established without a functional assay, and its wild-type residue assignment (asparagine) and domain assignment (BRCT) do not match p.Asp2811Gly in the DNA-binding domain.
• PMID:31131967 (Parsons et al. 2019): collated multifactorial data for 1395 BRCA1/2 predominantly intronic and missense variants, classifying 734 via posterior probability of pathogenicity (447 likely benign, 94 likely pathogenic); the article text names no individual variant assay result and the accompanying data tables carry no functional assay result for c.8432A>G.
• PMID:31112341 (CAGI 5 ENIGMA challenge): BRCA1/BRCA2-specific in-silico predictors benchmarked against HDR assay values and clinical labels; it reports gene-level performance (BRCA2 MLR accuracy 0.86, MCC 0.71) and names no individual variant, so it supplies no PS3 evidence (in-silico evidence belongs to PP3/BP4).
• PMID:25741868 (ACMG/AMP 2015): PS3 is defined as well-established in vitro or in vivo functional studies supportive of a damaging effect on the gene or gene product; the paper reports no variant-level data.

Still missing

• No entry for c.8432A>G / p.(Asp2811Gly) in Specifications_Table9_V1.2_2024-11-18.xlsx (the VCEP index-declared governing PS3/BS3 table); searched by c. notation, three-letter and one-letter protein notation, and legacy 8660A>G / D2811G, with zero matches. Neighbouring codons (2810, 2812, 2813, 2815-2819) are listed, so the absence is a true data gap rather than a notation mismatch.
• No functional assay entry for this variant in HUMU-40-1557-s001.xlsx SuppT4 (functional assay data) or SupplementaryTables_V1.2 ST4 (calibrated functional assay results); the variant's SuppT1 row has all functional-assay columns empty.
• No published functional assay of this variant was identified in the fetched case literature; only gene-level or in-silico papers were retrieved for this variant.

Sources

Reasoning

The governing ENIGMA BRCA1/BRCA2 V1.2 specification (Table 1 and Appendix F) applies PS4 only for case-control enrichment: the prevalence of the variant in affected individuals must be significantly increased versus ethnicity- and country-matched controls, with p-value <= 0.05 and relative risk/odds ratio >= 4.0 (the confidence interval around the estimate must not include 2.0). No variant-level case-control estimate exists for NM_000059.4:c.8432A>G p.(Asp2811Gly). The ENIGMA/Parsons multifactorial table (HUMU-40-1557-s001 SuppT1) carries a blank Case-Control LR cell for this variant, and its SuppT2 lists the variant only as a member of the 'Easton et al 2007 dataset' with no retrievable variant-level LR. The only case-based publication naming a change at this codon, PMID:10717622 (population-based series of 386 young breast carcinoma cases), reports a glycine substitution at amino acid 2811 but assigns a different wild-type residue (asparagine, i.e. N2811G), provides no OR, relative risk, p-value or control comparison, and classifies conserved-position missense changes as unclassified variants. Population data are consistent with rarity rather than enrichment (gnomAD v2.1 total 2/251,384 alleles, AF 7.96e-06; East Asian 1/18,390, AF 5.44e-05). ENIGMA additionally directs that where case-control likelihood-ratio estimates are available they be applied under PP4/BP5 rather than PS4, and that proband counting is 'Do not use'. Explicit inequality for the governing threshold: no OR estimate exists to compare against OR >= 4, so PS4 is not met.

Rule

ENIGMA BRCA1/BRCA2 V1.2 PS4 (Table 1 / Appendix F): case-control p-value <=0.05 and OR >=4 with lower 95% CI excluding 2.0; proband counting 'Do not use'.

Evidence considered

• ENIGMA BRCA1/BRCA2 V1.2 Table 1 / Appendix F defines PS4: 'The prevalence of the variant in affected individuals is significantly increased compared with the prevalence in controls. Case-control studies; p-value <=0.05 and OR >=4 (lower confidence interval excludes 2.0).' Appendix F adds that case-control results support a strong pathogenic code only if the RR/OR is >=4.0 and the CI around the estimate does not include 2.0.
• The ENIGMA specification notes that if case-control LR estimates are available for a given dataset they should be used in preference to case-control OR data, under code PP4 (or BP5) - i.e. an available case-control LR is reported under PP4/BP5, not PS4 - and that PS4_Moderate proband counting is 'Do not use as originally described'.
• HUMU-40-1557-s001 SuppT1 row for BRCA2 c.8432A>G p.(Asp2811Gly): Segregation LR blank, Pathology LR blank, Co-occurrence LR 1.0756788212, Family History LR 0.39973152851, Case-Control LR blank, Combined LR 0.4299827393629182, prior 0.03, posterior 0.01312390780007919, IARC class 'Likely Benign'. The Case-Control LR cell is empty, so no case-control evidence is contributed by this dataset.
• HUMU-40-1557-s001 SuppT2 lists 'Easton et al 2007 dataset | BRCA2 | c.8432A>G | ... | 1', i.e. the variant appears in the Easton 2007 data set, but no variant-level LR for it is retrievable from the converted Easton text.
• gnomAD v2.1: total allele count 2/251,384 (AF 7.95596e-06), highest population East Asian 1/18,390 (AF 5.43774e-05), 0 homozygotes; a frequency this low provides no case enrichment signal and is far below any case-control-based inference.
• PMID:10717622 (Malone et al. 2000, population-based case series of 386 breast carcinoma cases in western Washington State) reports 'For BRCA2 these include an asparagine to glycine change at amino acid 2811 and the glutamic acid to alanine change at amino acid 2856, both with the BRCT domains' - a codon-2811 glycine change but with a different wild-type residue (asparagine) and no OR, RR, p-value or control comparison; no PS4-qualifying statistic is reported.
• PMID:12692171 (ASCO policy statement) and PMID:17392385 (ACS breast MRI screening guideline) were screened as PS4 candidate sources; neither names this variant nor reports variant-level case/enrichment data.

Still missing

• No variant-level case-control odds ratio, relative risk, or p-value for c.8432A>G was found in HUMU-40-1557-s001.xlsx.txt, SupplementaryTables_V1.2_2024-11-18.xlsx.txt, PMID_31853058_BRCA2_clinical_history_LR.xlsx.txt, or PMID_17924331_Easton_2007_AJHG.pdf.txt.
• PMID_17924331_Easton_2007_AJHG.pdf.txt is listed in Parsons SuppT2 as a dataset containing c.8432A>G, but no variant-level LR or odds entry for this variant could be located in the converted Easton full text.
• PMID:10717622 reports a glycine substitution at codon 2811 (341 bp/organised as BRCT domains) but with a different wild-type residue assignment (asparagine) and without any case-control enrichment statistic, so it cannot be used as a PS4 case-control source for p.(Asp2811Gly).

Sources

Reasoning

Governing framework: ENIGMA BRCA1/BRCA2 specification v1.2 (BRCA2). PM2 (supporting) is applicable only when the variant is absent from controls in gnomAD v2.1 (non-cancer, exome-only subset) and gnomAD v3.1 (non-cancer), with an average regional read depth >=25 (Appendix G). BRCA2 c.8432A>G (p.Asp2811Gly) is not absent: it is observed at 2/236,848 alleles (AF 8.44e-06) in gnomAD v2.1 non-cancer exomes and 1/147,930 alleles (AF 6.76e-06) in gnomAD v3.1 non-cancer, and is also present in the all-comers datasets (gnomAD v2.1 total 2/251,384; gnomAD v4.1 16/1,613,870). Because the variant is present in controls, the absence requirement fails and PM2 is not met; the read-depth precondition is moot.

Rule

ENIGMA BRCA2 v1.2 PM2: absent from controls in gnomAD v2.1 (non-cancer, exome only subset) and gnomAD v3.1 (non-cancer), regional average read depth >=25.

Evidence considered

• CSPEC ENIGMA BRCA2 v1.2 PM2 rule: 'Absent from controls in an outbred population, from gnomAD v2.1 (non-cancer, exome only subset) and gnomAD v3.1 (non-cancer). Region around the variant must have an average read depth >=25.' (default strength downgraded to supporting).
• gnomAD v2.1 non-cancer exomes: present, 2/236,848 alleles (AF 8.444234276835777e-06), homozygotes 0.
• gnomAD v3.1 non-cancer: present, 1/147,930 alleles (AF 6.75995e-06), homozygotes 0.
• gnomAD v2.1 all-comers: present, 2/251,384 alleles (AF 7.95596e-06). gnomAD v4.1 all-comers: present, 16/1,613,870 alleles (AF 9.91406e-06).
• ENIGMA SupplementaryTables V1.2 lists BRCA2 c.8432A>G p.(Asp2811Gly) with Frequency Category '>0.00002 MAF <=0.0001' (Maximum MAF 5.653550429669833e-05), confirming non-absence in the governing table.

Still missing

• Regional read depth at the variant site was not captured, but is moot because PM2 fails on presence in controls regardless of depth.

Sources

Reasoning

The ENIGMA BRCA2 v1.2 specification repurposes PM3 as PM3_VariableWeight and restricts it to a proband whose phenotype is consistent with BRCA1/BRCA2-related Fanconi anemia (FA) who also carries co-occurrent variants in the same gene (Specifications Table 1 and Table 6). Points are assigned per FA proband: 2 points when the co-occurrent P/LP variant is in trans or homozygous, 1 point when phase is unknown; PM3 = 2 points, PM3_Supporting = 1 point, PM3_Strong >= 4 points. Neither prerequisite is met here. No proband with an FA-consistent phenotype (increased DEB/MMC chromosome breakage plus FA clinical features, or >= 2 features across >= 2 categories) is documented anywhere in the case evidence, and no second BRCA2 P/LP variant in trans or in a homozygous state is reported for c.8432A>G. The VCEP's own curated catalogue of reported BRCA2-related FA cases (SupplementaryTables ST9, all reported biallelic BRCA2 FA probands, e.g. c.5609_5610delinsAG/c.7007G>A, c.5946delT/c.9207T>A) was searched for c.8432A>G, p.(Asp2811Gly) and 2811 and contains no entry for this variant. The only phase-adjacent observations in the VCEP files are cancer-family multifactorial-likelihood rows, not biallelic FA data: Parsons 2019 SuppT1 records a single row for BRCA2 c.8432A>G with no co-occurrence or segregation LR, and the Easton 2007 source listing gives only a family-history family count of 1. ClinVar submissions for this variant (11 submissions, no expert panel) similarly report no cis/trans pathogenic-variant co-observation. PM3 is therefore not met at any strength.

Rule

ENIGMA BRCA2 v1.2 PM3_VariableWeight (Specifications Table 1 / Table 6): apply only for a proband with BRCA1/BRCA2-related Fanconi anemia phenotype plus co-occurrent variants in the same gene; 2 points if the co-occurrent P/LP variant is in trans or homozygous, 1 point if phase unknown; PM3 = 2 points, PM3_Supporting = 1 point, PM3_Strong >= 4 points.

Evidence considered

• ENIGMA BRCA2 v1.2 Specifications Table 1, criterion PM3_VariableWeight: apply for a patient with a phenotype consistent with BRCA1- or BRCA2-related Fanconi anemia and co-occurrent variants in the same gene; generic 'in trans with a pathogenic variant' phase evidence is not accepted outside that FA context.
• Specifications Table 6: 2 points for a co-occurrent P or LP variant in trans or in a homozygote, 1 point when phase is unknown; PM3 = 2 points, PM3_Supporting = 1 point, PM3_Strong >= 4 points across unrelated probands. Prerequisite: the variant under assessment must be sufficiently rare (not meeting a benign population code).
• Searched the VCEP-curated BRCA2-related Fanconi anemia case table (SupplementaryTables ST9, all reported biallelic BRCA2 FA probands) for 'c.8432A>G', 'p.(Asp2811Gly)' and '2811': no row for this variant.
• Searched Parsons et al. 2019 (PMID:31131967) converted supplement (HUMU-40-1557-s001): the only BRCA2 c.8432A>G row in SuppT1 is a multifactorial-likelihood row with no co-occurrence or segregation LR, and the Easton 2007 dataset source listing in SuppT2 gives only a family-history family count of 1.
• Li et al. 2020 clinical-history LR for BRCA2 c.8432A>G is 0.969 (2 probands) - cancer personal/family-history data, neutral, and not an allelic/phase observation.
• Reviewed the case bundle (ClinVar submissions, gnomAD v2.1/v4.1, paper extracts) for a proband with an FA phenotype or a second BRCA2 P/LP allele in this genotype: none reported; the 11 ClinVar submissions include no expert-panel and no cis/trans co-observation.

Still missing

• No proband-level Fanconi anemia phenotype (DEB/MMC chromosome-breakage result or FA clinical features) and no co-occurrent BRCA2 P/LP allele genotype for c.8432A>G were present in the case evidence bundle.
• VCEP file Specifications_V1.2_2024-11-18.docx and SupplementaryTables_V1.2_2024-11-18.xlsx (ST9 BRCA2 FA case table) were searched for c.8432A>G / p.(Asp2811Gly) / 2811 and contain no FA-related entry for this variant; absence of an entry is stated here rather than assumed.

Sources

Reasoning

Variant is a missense substitution (NM_000059.4:c.8432A>G, NP_000050.3:p.(Asp2811Gly)) and falls inside the BRCA2 DNA-binding domain, aa 2481-3186, which is a (potentially) clinically important functional domain under the ENIGMA BRCA1/BRCA2 specification v1.2. For in-domain missense/in-frame variants the ENIGMA BRCA2 PP3 path requires a predicted impact on protein (BayesDel no-AF score >=0.30 for BRCA2). The locally computed BayesDel no-AF score is 0.191743, which is below the 0.30 PP3 threshold and sits in the specification's explicitly uninformative band (>0.18 and <0.30) where PP3 and BP4 are both not met. The alternative PP3 route, predicted splice impact, is also not satisfied: SpliceAI max delta score is 0.032, well below the specification's SpliceAI >=0.2 threshold for PP3 (irrespective of variant location). On the generic missense path, REVEL 0.568 is below the ClinGen SVI supporting PP3 cutoff of 0.644 (Pejaver et al. 2022, PMID:36413997). No prediction path therefore supports a deleterious effect, and PP3 is not met. No pre-assigned PP3 entry for this variant exists in the VCEP tables: the only variant-level rows found for c.8432A>G in the converted VCEP files are multifactorial-likelihood rows in SupplementaryTables_V1.2 (posterior probability 0.0131, IARC class 2 / (Likely) Benign, Parsons et al. 2019 PMID 31131967) and HUMU-40-1557-s001 (combined LR 1.0757, posterior 0.430, 'Likely Benign'), which are PP4/BP5-type evidence and carry no PP3 assignment.

Rule

ENIGMA BRCA1/BRCA2 ACMG/AMP specification v1.2, PP3: missense/in-frame inside a (potentially) clinically important functional domain with BayesDel (no-AF) >=0.30 for BRCA2; OR SpliceAI >=0.2 for silent, missense/in-frame and intronic variants outside donor/acceptor +/-1,2.

Evidence considered

• Variant consequence is missense: variant_validator/Nutalyzer give NP_000050.3:p.(Asp2811Gly), codon 2811.
• D2811 lies inside the BRCA2 DNA-binding domain aa 2481-3186, one of the VCEP-defined (potentially) clinically important functional domains.
• ENIGMA v1.2 PP3 (missense/in-frame in-domain path): apply when BayesDel no-AF is predicted impact, i.e. >=0.30 for BRCA2 (Specifications Table 1 / Appendix J, Table 16).
• ENIGMA v1.2 PP3 (splice path, all variant types): apply when SpliceAI >=0.2 for silent, missense/in-frame and intronic variants outside the donor/acceptor +/-1,2 positions.
• Appendix J states the BayesDel category >0.18 and <0.30 for BRCA2 is the deliberately uninformative middle band in which 'PP3 or BP4 is not met'.
• Local BayesDel no-AF score for this variant = 0.191743 (bayesdel source_registry entry).
• SpliceAI max delta score = 0.032 (SpliceAI Lookup, NM_000059.4:c.8432A>G); DS_AG 0.001, DS_AL 0.032, DS_DG 0.0, DS_DL 0.004.
• Generic REVEL score for this variant = 0.568, below the ClinGen SVI supporting PP3 cutoff 0.644 (Pejaver et al. 2022, PMID:36413997); it is also above the 0.29 BP4 supporting cutoff, so it is in the non-informative zone.
• Searched the declared VCEP full texts (Specifications_V1.2, Appendices_V1.2, PMID_22505045_Houdayer_2012) for 'c.8432A>G', 'p.Asp2811Gly', 'p.D2811G', '8432' and '2811': no PP3 pre-assignment for this variant in any of them.
• The only variant-level rows for c.8432A>G in the converted VCEP tables are multifactorial-likelihood rows (SupplementaryTables_V1.2: posterior 0.0131, IARC class 2; HUMU-40-1557-s001: combined LR 1.0757, posterior 0.430), which are PP4/BP5-type evidence, not PP3.

Sources

Reasoning

ENIGMA BRCA1/BRCA2 V1.2 uses PP4_Variable Weight solely to capture the combined clinical likelihood ratio towards pathogenicity from multifactorial likelihood clinical data (co-segregation, co-occurrence with a pathogenic variant in the same gene, reported family history, tumour pathology, case-control), with weights assigned by combined LR: PP4 - LR >=2.08:1 (supporting), PP4_Moderate >=4.3:1, PP4_Strong >=18.7:1, PP4_Very Strong >=350:1, and combined LR 1.00 to <2.08 not informative (PP4 not applicable). The published ENIGMA multifactorial result for this exact variant (Parsons et al. 2019, as tabulated in HUMU-40-1557-s001 SuppT1 and referenced in SupplementaryTables_V1.2 ST7) gives Component LRs Co-occurrence 1.0756788212 and Family History 0.39973152851, yielding a Combined clinical LR of 0.4299827393629182 - i.e. below 1.00 and directed against pathogenicity rather than towards it. The independent Li et al. 2020 clinical-history LR for this exact variant is 0.9692042050827514 (LOG(LR) -0.03127995133399963, N = 2 probands), which sits inside the non-informative neutral zone (0.48 < LR < 2.08). Explicit inequality for the governing threshold: 0.430 >= 2.08? no. PP4 is therefore not met.

Rule

ENIGMA BRCA1/BRCA2 V1.2 PP4 (Variable Weight): combined clinical LR >=2.08:1 supporting; >=4.3:1 moderate; >=18.7:1 strong; >=350:1 very strong; LR 1.00 to <2.08 not informative (PP4 not applicable).

Evidence considered

• ENIGMA BRCA1/BRCA2 V1.2 Table 1 PP4_Variable Weight: 'Use ONLY to capture combined LR towards pathogenicity, based on multifactorial likelihood clinical data. PP4 - LR >=2.08:1; PP4_Moderate - LR>=4.3:1; PP4_Strong - LR>=18.7:1; PP4_Very Strong - LR>=350:1. Combined LR 1.00 to <2.08 is not informative (PP4 not applicable).'
• ENIGMA Appendix B states that evidence weights for clinical data derived from calibrated multifactorial likelihood analysis are assigned using PP4 (combined clinical evidence towards pathogenicity) or BP5 (combined evidence against pathogenicity), with weight defined by combined LR, and Appendix B Table 2 footnote a defines the combined clinical LR as the product of the individual component LRs (e.g. co-occurrence x family history).
• HUMU-40-1557-s001 SuppT1 (Parsons et al. 2019, PMID 31131967) exact-variant row BRCA2 c.8432A>G p.(Asp2811Gly): Co-occurrence LR 1.0756788212, Family History LR 0.39973152851, Combined LR 0.4299827393629182, prior probability 0.03, posterior probability 0.01312390780007919, IARC Class 'Likely Benign'. 1.0756788212 x 0.39973152851 = 0.4299827, confirming the combined clinical LR.
• SupplementaryTables_V1.2 ST7 reference-set row for BRCA2 c.8432A>G p.(Asp2811Gly): '(Likely) Benign', prior 0.03, posterior 0.0131239078, IARC class 2, source 'Parsons et al. 2019 (PMID 31131967)' - the ENIGMA reference-set classification for this variant is benign, not pathogenic, corroborating the absence of PP4-direction evidence.
• PMID_31853058_BRCA2_clinical_history_LR.xlsx.txt exact match: 'BRCA2 c.8432A>G c.8432A>G -0.03127995133399963 2 0.9692042050827514' (LOG(LR), N probands, LR = exp(LOG(LR))). LR 0.969 lies in the neutral, non-informative zone for PP4/BP5 (greater than 0.48 and less than 2.08).
• Explicit threshold comparison for PP4: combined clinical LR 0.430 >= 2.08? no - and 0.430 also falls below 1.00, so it is in the against-pathogenicity direction rather than in the 1.00-<2.08 not-informative band.
• PMID:17924331 (Easton et al. 2007) was consulted as the foundational methodology/legacy multifactorial source for PP4/BP5; it describes combining independent co-occurrence, family-history and co-segregation LRs, but carries no locatable variant-level entry for c.8432A>G in the converted text.

Still missing

• The exact-variant row in PMID_17924331_Easton_2007_AJHG.pdf.txt could not be located in the converted text (the paper is retained here for methodology context only); the combined clinical LR used for this assessment is taken from the ENIGMA/Parsons resource tables.
• No variant-level co-segregation, tumour-pathology or case-control LR component is populated for this variant (Segregation LR, Pathology LR and Case-Control LR cells are all blank in HUMU-40-1557-s001 SuppT1); the combined LR rests on the co-occurrence and family-history components only.

Sources

Reasoning

PP5 ('reputable source recently reports variant as pathogenic but the evidence is not available to the laboratory to perform an independent evaluation') is listed as 'Do not use' in the governing ENIGMA BRCA1/BRCA2 V1.2 specification, and the operative rule for this case applies PP5 only when there is an exact-variant ClinVar expert-panel classification of Pathogenic or Likely pathogenic. ClinVar VCV000052585 (ClinVarID 52585) for NM_000059.4:c.8432A>G p.(Asp2811Gly) has 11 submissions from 11 clinical laboratories, all with review status 'criteria provided, single submitter' (1 star; overall API label 'Conflicting classifications of pathogenicity'), and contains no expert-panel assertion (expert_panels field empty; expert-panel submissions = 0). Submitter classifications are conflicting (7 Uncertain significance, 3 Likely benign, 1 Likely Benign), and every one is an ordinary laboratory submission, which the PP5 rule explicitly excludes. No exact-variant expert-panel pathogenic/likely pathogenic assertion exists, so PP5 cannot be applied.

Rule

ENIGMA BRCA1/BRCA2 V1.2 PP5: 'Do not use' for BRCA2; under the case policy PP5 is applied at supporting strength only for an exact-variant ClinVar expert-panel (3-star) Pathogenic/Likely pathogenic assertion.

Evidence considered

• ENIGMA BRCA1/BRCA2 V1.2 Table 1 lists PP5 as 'Do not use' (reputable source reports variant as pathogenic but the evidence is not available to the laboratory to perform an independent evaluation).
• ClinVar VCV000052585: overall classification 'Uncertain significance' (API: 'Conflicting classifications of pathogenicity'), review status 'criteria provided, single submitter', review_stars 1, 11 submissions from 11 submitters, and an empty expert_panels object - i.e. no expert-panel assertion.
• ClinVar submission summary: 7 clinical laboratories Uncertain significance, 3 clinical laboratories Likely benign, 1 clinical laboratory Likely Benign - all ordinary laboratory submissions with single-submitter review status.
• Case ClinVar audit fields: submissions_total 11, expert_panel_submissions 0, expert_panel_plp_candidates 0, non_expert_majority_plp_candidates 0.

Still missing

• ClinVar VCV000052585 contains no expert-panel (ClinGen/VCEP) submission for this variant; expert_panels is empty and expert_panel_submissions = 0, so the PP5 trigger condition cannot be evaluated against an expert assertion.

Sources

Reasoning

Governing framework: ENIGMA BRCA1/BRCA2 specification v1.2 (BRCA2). BA1 (benign stand-alone) requires a maximum filter allele frequency (FAF) above 0.001 (0.1%) in a gnomAD non-founder population, assessed on gnomAD v2.1 non-cancer exomes and gnomAD v3.1 non-cancer genomes considered separately, and applied conservatively on filter allele frequency rather than minor allele frequency (Appendix G). For BRCA2 c.8432A>G (p.Asp2811Gly), the highest non-founder-population frequency observed is 5.65e-05 (1/17,688 alleles, East Asian) in gnomAD v2.1 non-cancer exomes; the governing ENIGMA SupplementaryTables V1.2 also assigns this variant a Maximum MAF of 5.653550429669833e-05 with the frequency category '>0.00002 MAF <=0.0001'. The gnomAD v3.1 non-cancer genome maximum is 1.54e-05 (1/64,810, European non-Finnish), and even the all-comers gnomAD v4.1 total AF is only 9.91e-06 (16/1,613,870 alleles; grpmax FAF 7.63e-06). No population comes anywhere near the 0.001 stand-alone threshold, so BA1 is not met.

Rule

ENIGMA BRCA2 v1.2 BA1: FAF > 0.001 in gnomAD v2.1 (non-cancer, exomes) and/or gnomAD v3.1 (non-cancer), maximum non-founder-population FAF, exome/genome assessed separately, read depth >=20.

Evidence considered

• CSPEC ENIGMA BRCA2 v1.2 BA1 rule: 'Filter allele frequency (FAF) is above 0.1% (FAF > 0.001) in gnomAD v2.1 (non-cancer, exome only subset) and/or gnomAD v3.1 (non-cancer), non-founder population(s).'
• Specifications V1.2 Table 1: BA1 threshold 'Above 0.001 (0.1%)'; preferred datasets 'gnomAD v2.1 non-cancer, exomes' and 'gnomAD v3.1 non-cancer, genomes'; apply on maximum filter allele frequency in a gnomAD non-founder population, exome and genome assessed separately; do not apply if read depth <20; do not apply to well-established pathogenic founder variants.
• gnomAD v2.1 (all-comers, exomes): total AF 7.95596e-06 (2/251,384 alleles, 0 homozygotes); highest population EAS AF 5.43774e-05 (1/18,390).
• gnomAD v2.1 non-cancer exomes: AF 8.44423e-06 (2/236,848 alleles, 0 homozygotes); highest population EAS AF 5.653550429669833e-05 (1/17,688 alleles).
• gnomAD v3.1 non-cancer: AF 6.75995e-06 (1/147,930 alleles, 0 homozygotes); highest population European (non-Finnish) AF 1.54297e-05 (1/64,810 alleles); grpmax FAF reported as 0 in the capture.
• gnomAD v4.1 (all-comers): AF 9.91406e-06 (16/1,613,870 alleles, 0 homozygotes); highest population EAS AF 2.22717e-05 (1/44,900 alleles); grpmax FAF 7.63e-06.
• ENIGMA SupplementaryTables V1.2 row for BRCA2 c.8432A>G p.(Asp2811Gly): frequency category '>0.00002 MAF <=0.0001', Maximum MAF 5.653550429669833e-05, population of maximum East Asian, derived from gnomAD V2.1.1 non-cancer (download date 2019-03-29). This pre-assigned value governs and is well below the 0.001 BA1 cutoff.

Still missing

• gnomAD read-depth at the variant site was not captured in the case evidence; BA1 requires read depth >=20 for applicability. Depth is not limiting here because all observed frequencies are ~18-fold or more below the 0.001 threshold, so no plausible depth correction could bring the variant to BA1.

Sources

Reasoning

Governing framework: ENIGMA BRCA1/BRCA2 specification v1.2 (BRCA2). Under this specification BS2 is not a pooled-allele-frequency code; it is assigned by proband-level points (Table 8, Appendix H) based on co-occurrent BRCA2 variants in individuals without a Fanconi Anemia phenotype (BS2 = >=4 points; BS2_Moderate = 2; BS2_Supporting = 1). Appendix H explicitly excludes data used to assign frequency-based codes from BS2, noting that variants observed as >=2 homozygotes in gnomAD are already captured by BA1/BS1. No proband with biallelic or co-occurrent BRCA2 variants and an unlikely-FA phenotype is reported in this case, so no BS2 points can be assigned. The population evidence is likewise negative for the generic homozygote-observation premise: gnomAD reports 0 homozygotes for c.8432A>G in v2.1 non-cancer exomes (2/236,848 alleles), v3.1 non-cancer (1/147,930 alleles) and v4.1 (16/1,613,870 alleles). With no qualifying proband and no homozygote observation, BS2 is not met.

Rule

ENIGMA BRCA2 v1.2 BS2: proband-based point tally (Table 8 / Appendix H); >=4 points = BS2, 2 = moderate, 1 = supporting; frequency-based data excluded.

Evidence considered

• CSPEC ENIGMA BRCA2 v1.2 BS2 rules: 'Applied in absence of features of recessive disease, namely Fanconi Anemia phenotype. See Specifications Table 8 for approach to assign points per proband... BS2 = >=4 points; BS2_Moderate = 2 points; BS2_Supporting = 1 point.'
• Specifications V1.2 Table 1 BS2_VariableWeight: 'Observed in a healthy adult individual for a recessive (homozygous)... disorder, with full penetrance expected at an early age'; 'Applied in absence of features of recessive disease, namely Fanconi Anemia phenotype'; co-occurrent P or LP variant should be assigned classification using VCEP specifications.
• Appendix H footnote: 'Apply only for phenotyped individuals from clinical or research cohorts. NOT to be applied for data used to assign frequency-based codes; we have observed that variants observed as >=2 homozygotes in GnomAD data (assumed unaffected adults of unspecified age) are already captured by the BA1 or BS1 frequency codes.'
• gnomAD v2.1 non-cancer exomes: 2/236,848 alleles, homozygotes = 0. gnomAD v3.1 non-cancer: 1/147,930 alleles, homozygotes = 0. gnomAD v4.1: 16/1,613,870 alleles, homozygotes = 0.
• No case-level proband homozygous/co-occurrent genotype observation is present in the case evidence reviewed (review_manifest/case_summary/source_registry/prefetch/evidence bundle).

Still missing

• No proband-level biallelic/co-occurrence genotype data were available in this case, so the ENIGMA Table 8 point tally for BS2 could not be fully evaluated; the verdict rests on the absence of any such observation plus 0 gnomAD homozygotes.
• gnomAD homozygote counts are reported per dataset but the raw genotype/read-depth detail was not captured; this does not affect the negative verdict.

Sources

Reasoning

BS3 requires well-established in vitro or in vivo functional studies showing no damaging effect on protein function (assay measuring effect via protein only, or mRNA and protein combined), with weight from the calibrated mammalian assay set; per Appendix E and Figure 1C, for a missense variant inside a (potentially) clinically important functional domain - here the BRCA2 DNA-binding domain, aa 2481-3186, containing codon 2811 - protein functional data (or a protein/mRNA-combined assay) are required before any BS3 weight can be applied, and mRNA-only splicing results cannot substitute. The governing ENIGMA curated table (Specifications Table 9) contains no entry for c.8432A>G / p.(Asp2811Gly): the surrounding BRCA2 codons are curated (c.8443G>A p.Val2815Ile, c.8447G>A p.Gly2816Asp, c.8450G>T p.Cys2817Phe and c.8428A>G p.Ser2810Gly are each BS3 Strong on the basis of a calibrated study), but c.8432A>G is absent, and '8432', '2811', 'D2811' and 'Asp2811' return zero matches across the whole file. The Parsons et al. 2019 ENIGMA dataset likewise contains no functional result for this variant: its SuppT1 row has every functional-assay column empty, and the variant is absent from the functional-assay cross-reference tables SuppT4 and Supplementary ST4 despite the adjacent c.8428A>G being listed there as 'No functional impact'. The only ENIGMA variant-level evidence for this variant is multifactorial clinical likelihood data (combined LR 0.43; posterior probability 0.0131; IARC class 2, likely benign), which the specification routes to BP5/PP4 rather than BS3, so the benign-direction multifactorial result is not usable here. No published assay of this variant was identified in the case literature (PMID:10717622 performed no functional work; other fetched papers are gene-level or in-silico only). Absence of assay data is not evidence of preserved function, so BS3 is not met, at any strength.

Rule

ENIGMA BRCA1/BRCA2 VCEP specification v1.2, Table 1 BS3 row (functional assays, protein or mRNA+protein, with Table 9 as the go-to curated list) plus footnote ** domain definition (BRCA2 DNA binding aa 2481-3186); Appendix E/Figure 1C require protein functional data for missense variants inside clinically important domains. No pre-assigned code exists for this variant.

Evidence considered

• ENIGMA BRCA1/BRCA2 specification v1.2, Table 1, BS3 row: BS3 (functional assays) requires well-established in vitro or in vivo functional studies showing no damaging effect on protein function, with the assay measuring effect via protein only OR mRNA and protein combined; the separate BS3 (mRNA assay only) route is handled as BP7_Strong (RNA) and is restricted to intronic, silent and missense/in-frame variants located outside a (potentially) clinically important functional domain.
• ENIGMA specification v1.2 footnote ** and Appendix C define BRCA2 DNA binding aa 2481-3186 as a clinically important functional domain; p.Asp2811Gly lies within it, so only protein-level (or protein+mRNA) functional data could support BS3 here.
• Appendices v1.2 Appendix E and Table 10: no impact in at least one calibrated assay with no conflict provides Strong evidence against pathogenicity (i.e. BS3 Strong); for exonic missense variants within (potentially) clinically important functional domains, protein functional results are required, and in their absence splicing-only results carry no code weight. No such no-impact assay exists for this variant.
• Specifications_Table9_V1.2_2024-11-18.xlsx (sheet Table9_BRCA12VCEP_specs): no row for c.8432A>G / p.(Asp2811Gly). Nearby BS3 Strong rows that would have been the calibration precedent are c.8428A>G p.(Ser2810Gly) (Richardson 2021, Neutral), c.8443G>A p.(Val2815Ile) (Richardson 2021, Neutral), c.8447G>A p.(Gly2816Asp) (Richardson 2021, Neutral) and c.8450G>T p.(Cys2817Phe) (Richardson 2021, Neutral); c.8455G>C p.(Asp2819His) and c.8456A>T p.(Asp2819Val) are PS3 Strong. None of these is the queried variant.
• HUMU-40-1557-s001.xlsx SuppT1, row for BRCA2 c.8432A>G p.(Asp2811Gly) (legacy 8660A>G D2811G): combined LR 0.43, prior 0.03, posterior probability 0.0131, IARC class 2 (Likely Benign), comment 'Multifac new calc'; Functional Paper/s and all individual assay columns (Findlay, Starita, Mesman, Bouwman, Fernandes2019, Petitalot2019, Hart2018) are empty, so there is no BS3-directed functional result in this dataset.
• ENIGMA specification v1.2, 'Multifactorial likelihood analysis' section: clinical data from calibrated multifactorial likelihood analysis are assigned under PP4 or BP5 (with bioinformatic priors excluded); therefore the benign-direction multifactorial posterior probability of 0.0131 for this variant cannot be applied as BS3.
• HUMU-40-1557-s001.xlsx SuppT4 and SupplementaryTables_V1.2 ST4: no c.8432A>G row, whereas the adjacent c.8428A>G p.(Ser2810Gly) is present in SuppT4 as 'No functional impact' (Hart) and in ST7/ST4 with its multifactorial class - confirming a genuine lack of assay data for c.8432A>G.
• PMID:10717622 (Malone et al. 2000): a population-based PCR/SSCP screen with no functional assay; it notes that missense changes at conserved positions, including the change at codon 2811, cannot be classified without a functional assay, and its residue/domain description (asparagine-to-glycine in the BRCT domains) does not match p.Asp2811Gly in the BRCA2 DNA-binding domain.
• PMID:31131967 (Parsons et al. 2019): the source of the ENIGMA multifactorial and functional-assay cross-reference tables; the variant carries multifactorial clinical likelihoods only, with no functional assay result recorded in the accompanying data.
• PMID:31112341 and PMID:25741868: in-silico predictor benchmarking and general ACMG/AMP criteria definitions respectively; neither provides functional assay data for this variant.

Still missing

• No entry for c.8432A>G / p.(Asp2811Gly) in Specifications_Table9_V1.2_2024-11-18.xlsx (the VCEP index-declared governing PS3/BS3 table); searched by c. notation, three-letter and one-letter protein notation, and legacy 8660A>G / D2811G, with zero matches, while the neighbouring codons 2810, 2812, 2813, 2815-2819 are curated.
• No functional assay entry for this variant in HUMU-40-1557-s001.xlsx SuppT4 or SupplementaryTables_V1.2 ST4; the variant's SuppT1 row has all functional-assay columns empty, so no BS3-directed assay result exists in the ENIGMA multifactorial dataset either.
• No published functional assay of this variant was identified in the fetched case literature, so the benign direction of BS3 rests on absence of data rather than a negative assay.

Sources

Reasoning

The ENIGMA BRCA1/2 v1.2 specification applies BP1_Strong (Benign Strong) for silent, missense or in-frame insertion/deletion/delins variants outside a (potentially) clinically important functional domain AND with no splicing predicted (SpliceAI <= 0.1). The splicing prong is satisfied here: SpliceAI max delta for c.8432A>G is 0.032 (<= 0.1). The domain prong, however, fails. Residue Asp2811 lies within the BRCA2 DNA-binding domain, aa 2481-3186 (exon 19; Specifications Table 4 annotates exon 19 as 'Coding DBD', protein range p.2778_2829), which the specification footnote and Appendix Tables 3/4 define as one of the (potentially) clinically important functional domains. Because the variant is inside a clinically important functional domain, BP1/BP1_Strong cannot be applied, and no supporting-strength BP1 provision exists for in-domain missense variants. BP1 is therefore not met.

Rule

ENIGMA v1.2 Table 1: BP1_Strong applies for silent/missense/in-frame variants outside a (potentially) clinically important functional domain AND SpliceAI <= 0.1.

Evidence considered

• ENIGMA BRCA1/2 specification v1.2, Table 1 (BP1_Strong): apply for silent substitution, missense or in-frame insertion, deletion or delins variants OUTSIDE a (potentially) clinically important functional domain AND no splicing predicted (SpliceAI <= 0.1); missense prediction not applicable.
• Specification footnote defining (potentially) clinically important functional domains: BRCA2 PALB2 binding domain aa 10-40; BRCA2 DNA binding aa 2481-3186. Asp2811 (residue 2811) falls inside aa 2481-3186.
• Appendices_V1.2_2024-11-18.docx, Appendix Table 4: BRCA2 DBD (DNA/DSS1 binding domain - helical, OB1, OB2, OB3) aa 2481-3186 is a clinically important functional domain.
• Specifications_Table4_V1.2_2024-11-18.xlsx, BRCA2 exon 19 rows: 'Coding DBD', protein range p.2778_2829; c.8432 is located in exon 19 (c.8332-c.8487).
• SpliceAI for NM_000059.4:c.8432A>G: max delta score 0.032 (DS_AG 0.001, DS_AL 0.032, DS_DG 0.0, DS_DL 0.004), i.e. no splicing predicted and the SpliceAI <= 0.1 prong of BP1_Strong is satisfied.
• CSPEC compact evidence: BP1 defaultStrength Benign Supporting, with the BP1_Strong rule text naming BRCA2 DNA binding aa 2481-3186 as a clinically important functional domain.

Sources

Reasoning

Variant is a missense substitution inside the BRCA2 DNA-binding domain (aa 2481-3186), a (potentially) clinically important functional domain. The ENIGMA BRCA2 v1.2 BP4 rule for in-domain missense/in-frame variants requires BOTH no predicted impact on protein (BayesDel no-AF <=0.18 for BRCA2) AND no predicted impact on splicing (SpliceAI <=0.1). The splice component is satisfied (SpliceAI max delta 0.032 <= 0.1), but the protein component is not: the local BayesDel no-AF score is 0.191743, above the <=0.18 BP4 cutoff, placing the variant in the specification's uninformative band (>0.18 and <0.30) where PP3 and BP4 are both explicitly not met. The clean SpliceAI result cannot be substituted for the failed missense-prediction component, and an in-domain missense variant's PP3/BP4 evidence must come from the protein-impact path. On the generic missense path, REVEL 0.568 exceeds the ClinGen SVI supporting BP4 cutoff of 0.29 (Pejaver et al. 2022, PMID:36413997), so no BP4 arises there either. BP4 is therefore not met; it is neither supported nor opposed by a borderline-satisfied single component.

Rule

ENIGMA BRCA1/BRCA2 ACMG/AMP specification v1.2, BP4: missense/in-frame inside a (potentially) clinically important functional domain with no predicted protein impact and no predicted splice impact (BayesDel no-AF <=0.18 for BRCA2 AND SpliceAI <=0.1).

Evidence considered

• ENIGMA v1.2 BP4 (missense/in-frame in-domain path): apply when there is no predicted protein impact AND no predicted splice impact, i.e. BayesDel no-AF <=0.18 for BRCA2 AND SpliceAI <=0.1 (Specifications Table 1 / Appendix J, Table 16).
• Local BayesDel no-AF score = 0.191743, which is >0.18 and therefore fails the BP4 protein component and falls in Appendix J's declared uninformative band (>0.18 and <0.30 for BRCA2) where PP3/BP4 are not met.
• SpliceAI max delta score = 0.032, which satisfies the <=0.1 BP4 splice component.
• Per the specification, an in-domain missense variant's bioinformatic evidence comes from the protein-impact (BayesDel) path; a clean SpliceAI score is not a substitute for a failed missense prediction.
• Generic REVEL score 0.568 is above the ClinGen SVI supporting BP4 cutoff of 0.29 (Pejaver et al. 2022, PMID:36413997), so the generic missense path also yields no BP4.
• Searched the declared VCEP full texts for the variant in all notations; no BP4 pre-assignment exists for c.8432A>G in Specifications_V1.2, Appendices_V1.2 or PMID_22505045_Houdayer_2012.
• Variant-level rows for c.8432A>G in SupplementaryTables_V1.2 and HUMU-40-1557-s001 are multifactorial-likelihood (posterior 0.0131 / 0.430) PP4/BP5-type evidence, not BP4, and were not used here.

Sources

Reasoning

BP6 ('reputable source recently reports variant as benign but the evidence is not available to the laboratory to perform an independent evaluation') is listed as 'Do not use' in the governing ENIGMA BRCA1/BRCA2 V1.2 specification, and the operative rule for this case applies BP6 only when there is an exact-variant ClinVar expert-panel classification of Benign or Likely benign. ClinVar VCV000052585 (ClinVarID 52585) for NM_000059.4:c.8432A>G p.(Asp2811Gly) contains 11 submissions from 11 clinical laboratories, all with review status 'criteria provided, single submitter' (1 star), with an overall API label of 'Conflicting classifications of pathogenicity' and an empty expert_panels object (expert-panel submissions = 0). The four benign-direction submissions (3 categorized 'Likely benign' and 1 'Likely Benign') are ordinary laboratory submissions, which the BP6 rule explicitly excludes, and the aggregate ClinVar label is not an expert-panel assertion. No exact-variant expert-panel benign/likely benign assertion exists, so BP6 cannot be applied.

Rule

ENIGMA BRCA1/BRCA2 V1.2 BP6: 'Do not use' for BRCA2; under the case policy BP6 is applied at supporting strength only for an exact-variant ClinVar expert-panel (3-star) Benign/Likely benign assertion.

Evidence considered

• ENIGMA BRCA1/BRCA2 V1.2 Table 1 lists BP6 as 'Do not use' (reputable source recently reports variant as benign but the evidence is not available to the laboratory to perform an independent evaluation).
• ClinVar VCV000052585: overall classification 'Uncertain significance' (API: 'Conflicting classifications of pathogenicity'), review status 'criteria provided, single submitter', review_stars 1, 11 submissions from 11 submitters, and an empty expert_panels object - i.e. no expert-panel assertion.
• ClinVar submission summary: 3 clinical laboratories Likely benign and 1 clinical laboratory Likely Benign (plus 7 clinical laboratories Uncertain significance) - all ordinary laboratory submissions with single-submitter review status, none from an expert panel.
• Case ClinVar audit fields: submissions_total 11, expert_panel_submissions 0, expert_panel_plp_candidates 0.

Still missing

• ClinVar VCV000052585 contains no expert-panel (ClinGen/VCEP) submission for this variant; expert_panels is empty and expert_panel_submissions = 0, so the BP6 trigger condition cannot be evaluated against an expert assertion.

Sources

Reasoning

The ENIGMA BRCA2 VCEP specification v1.2 defines PP1 strictly as 'Co-segregation with disease in multiple affected family members ... as measured by a quantitative co-segregation analysis method' (Appendix I; recommended tools COOL or CAL-Leiden), with weights assigned from the Bayes score: PP1 LR >=2.08:1, PP1_Moderate LR >=4.3:1, PP1_Strong LR >=18.7:1, PP1_Very Strong LR >=350:1. No variant-specific co-segregation likelihood ratio or pedigree-level data exists for NM_000059.4:c.8432A>G (p.Asp2811Gly; p.D2811G). The declared ENIGMA evidence tables were searched directly in their converted full text and the variant IS present, but with no segregation component: in HUMU-40-1557-s001 (Parsons et al. 2019, PMID:31131967) Supplementary Table 1 the row for c.8432A>G / p.(Asp2811Gly) has an EMPTY Segregation LR cell, with only Co-occurrence LR 1.0757 and Family History LR 0.3997 multiplying to a Combined LR of 0.4300 (posterior probability 0.0131, IARC class 'Likely Benign'); Supplementary Table 2 records the variant only in the 'Easton et al 2007 dataset' with the Co-Segregation Family Count cell blank and a Family History Score Family Count of 1. SupplementaryTables_V1.2 ST7 ('Reference Set Variants', used to derive frequency-based LRs) likewise lists c.8432A>G / p.(Asp2811Gly) as a '(Likely) Benign' reference-set variant with prior 0.03 and posterior 0.0131 (IARC class 2), with no co-segregation LR. There is consequently no Bayes score to compare against the PP1 thresholds, so PP1 cannot be assessed and is not met. Because this variant causes a common, genetically heterogeneous cancer, and the ENIGMA framework forbids deriving PP1 from proband/family-history likelihood ratios (those are handled under PP4), no generic-ACMG substitute is used.

Rule

Apply PP1 only from quantitative co-segregation analysis: PP1 LR >=2.08:1; PP1_Moderate LR >=4.3:1; PP1_Strong LR >=18.7:1; PP1_Very Strong LR >=350:1 (ENIGMA VCEP v1.2, Appendix I).

Evidence considered

• ENIGMA VCEP v1.2 requires quantitative co-segregation analysis for PP1 and defines LR thresholds of >=2.08 (supporting), >=4.3 (moderate), >=18.7 (strong) and >=350 (very strong).
• Appendix I (Co-segregation Analysis, PP1 and BS4) specifies a Bayesian analysis of age-specific penetrance as the preferred method, with COOL or CAL-Leiden recommended, and states the same LR thresholds for converting the Bayes score to PP1/BS4 strength.
• HUMU-40-1557-s001 SuppT1 row for BRCA2 c.8432A>G / p.(Asp2811Gly): Segregation LR blank; Co-occurrence LR 1.0756788212; Family History LR 0.39973152851; Combined LR 0.4299827393629182; prior 0.03; posterior 0.01312390780007919; IARC class 'Likely Benign'.
• HUMU-40-1557-s001 SuppT2 row: 'Easton et al 2007 dataset | BRCA2 | c.8432A>G | (Co-Segregation Family Count blank) | (Breast Tumour Count blank) | 1 (Family History Score Family Count)'. A co-segregation family count of zero means no pedigree contributed a segregation likelihood ratio for this variant.
• SupplementaryTables_V1.2 ST7 ('Reference Set Variants and associated population frequencies used to estimate frequency-based likelihood ratios') row for BRCA2 c.8432A>G / p.(Asp2811Gly): reference set '(Likely) Benign'; prior probability 0.03; posterior probability 0.0131239078; IARC class 2; source 'Parsons et al. 2019 (PMID 31131967)'.
• The consolidated case-evidence bundle was reviewed for family, proband, parental testing, segregation and meioses content; none is present for this variant, and the ClinVar record contains 11 single-submitter submissions (7 Uncertain significance, 4 Likely benign) with no expert-panel submission or segregation assertion.

Still missing

• No pedigree structure, affected-relative genotypes, unaffected-relative genotypes, parental genotypes, meioses or quantitative co-segregation LR for c.8432A>G is present in the case evidence or in any declared VCEP source.
• HUMU-40-1557-s001.xlsx was searched for c.8432A>G, p.(Asp2811Gly), 8660A>G, D2811G and 2811; the variant IS present, but its Segregation LR field is blank, so a PP1-supporting LR cannot be derived.
• SupplementaryTables_V1.2 ST7 was searched for c.8432A>G and p.(Asp2811Gly); the variant IS present as a '(Likely) Benign' reference-set variant (prior 0.03, posterior 0.0131, IARC class 2, source Parsons et al. 2019), but the entry contains no co-segregation likelihood ratio.
• No co-segregation analysis for this variant using the recommended COOL or CAL-Leiden tools was identified.

Sources

Reasoning

The ENIGMA BRCA2 VCEP specification v1.2 defines BS4 strictly as 'Lack of segregation in affected members of a family ... as measured by a quantitative co-segregation analysis method' (Appendix I), with weights assigned from the Bayes score: BS4_Supporting LR <=0.48:1, BS4_Moderate LR <=0.23:1, BS4 LR <=0.05:1, BS4_Very Strong LR <=0.00285:1. BS4 therefore requires a co-segregation likelihood ratio from pedigree analysis; no such value exists for NM_000059.4:c.8432A>G (p.Asp2811Gly; p.D2811G). The variant IS present in the two declared ENIGMA evidence tables, but neither entry supplies a segregation component: HUMU-40-1557-s001 (Parsons et al. 2019, PMID:31131967) SuppT1 gives an EMPTY Segregation LR, with the Combined LR of 0.4300 built solely from Co-occurrence LR 1.0757 and Family History LR 0.3997 (posterior probability 0.0131, IARC class 'Likely Benign'); SuppT2 shows the Co-Segregation Family Count is blank (0 families). SupplementaryTables_V1.2 ST7 ('Reference Set Variants ... used to estimate frequency-based likelihood ratios') lists c.8432A>G / p.(Asp2811Gly) as a '(Likely) Benign' reference-set variant, prior 0.03, posterior 0.0131, IARC class 2, source Parsons et al. 2019. That combined multifactorial clinical likelihood ratio is not a co-segregation measure: ENIGMA v1.2 Table 1 assigns 'combined LR against pathogenicity, based on multifactorial likelihood clinical data' to BP5 (and, in the pathogenic direction, to PP4), and Table 7 demonstrates this explicitly (its BP5 examples combine co-occurrence and family-history LRs, whereas its BS4 example is a co-segregation LR of 0.234). Applying the Parsons combined LR/posterior probability to BS4 would both misuse a non-segregation evidence type and double-count evidence that belongs to the BP5 assessment. There is consequently no segregation or non-segregation likelihood ratio to compare against the BS4 thresholds, so BS4 cannot be assessed and is not met. The ENIGMA stipulation to assess bi-linearity/phenocopies before assigning Benign Very Strong could not be evaluated because no family-level data exist.

Rule

Apply BS4 only from quantitative co-segregation analysis: BS4_Supporting LR <=0.48:1; BS4_Moderate LR <=0.23:1; BS4 LR <=0.05:1; BS4_Very Strong LR <=0.00285:1 (ENIGMA VCEP v1.2, Appendix I).

Evidence considered

• ENIGMA VCEP v1.2 requires quantitative co-segregation analysis for BS4 and defines LR thresholds of <=0.48 (supporting), <=0.23 (moderate), <=0.05 (strong) and <=0.00285 (very strong), with a stipulation to assess bi-linearity before assigning Benign Very Strong.
• HUMU-40-1557-s001 SuppT1 row for BRCA2 c.8432A>G / p.(Asp2811Gly): Segregation LR blank; Co-occurrence LR 1.0756788212; Family History LR 0.39973152851; product 1.0756788212 x 0.39973152851 = 0.42998, matching the reported Combined LR 0.4299827393629182; posterior 0.01312390780007919; IARC class 'Likely Benign'.
• HUMU-40-1557-s001 SuppT2 row for c.8432A>G lists the variant only in the 'Easton et al 2007 dataset' with Co-Segregation Family Count blank and Family History Score Family Count 1.
• SupplementaryTables_V1.2 ST7 row for BRCA2 c.8432A>G / p.(Asp2811Gly): reference set '(Likely) Benign'; prior 0.03; posterior 0.0131239078; IARC class 2; source Parsons et al. 2019 (PMID 31131967); the sheet is described as the reference set of variants and population frequencies used to estimate frequency-based likelihood ratios.
• ENIGMA v1.2 Table 1 assigns combined multifactorial-likelihood clinical data to PP4 (pathogenic direction) and BP5 (benign direction), not BS4; Table 7 of the same specification illustrates BP5 using co-occurrence plus family-history LRs and BS4 using a dedicated co-segregation LR.
• The consolidated case-evidence bundle was reviewed for affected-relative non-carriers, parental testing, unaffected relatives, pedigree and meioses content; none is present for this variant.

Still missing

• No affected-relative non-carrier genotypes, pedigree structure, meioses or quantitative co-segregation LR for c.8432A>G is present in the case evidence or in any declared VCEP source.
• HUMU-40-1557-s001.xlsx was searched for c.8432A>G, p.(Asp2811Gly), 8660A>G, D2811G and 2811; the variant IS present but with a blank Segregation LR field and a co-segregation family count of 0, so no BS4-supporting LR can be derived.
• SupplementaryTables_V1.2 ST7 was searched for c.8432A>G and p.(Asp2811Gly); the variant IS present as a '(Likely) Benign' reference-set variant (prior 0.03, posterior 0.0131, IARC class 2), but that entry is a multifactorial/frequency-based result with no segregation component and per ENIGMA v1.2 Table 1 and Table 7 corresponds to BP5 rather than BS4, so it is not applied here.
• The ENIGMA stipulation to assess the possibility of bi-linearity to explain negative co-segregation could not be evaluated, as no family-level genotype data were available.

Sources