POLD1 c.3218+4T>C, p.?

Reasoning

PM2 applies when a variant is absent from, or present at extremely low frequency in, population databases; for a gene without a gene-specific threshold the SVI recommendation is to apply it at supporting strength using a default frequency threshold of <= 0.0001 (supplied generic calibration; PMID:25741868). No POLD1 ClinGen CSPEC, gene VCEP specification, or local framework is available, and no specification mandates the non-cancer or exome-only gnomAD subsets, so the all-comers GNOMAD_V2_1 / GNOMAD_V4_1 entries were used as the default source. The variant is absent from gnomAD v2.1 (all-comers) and present in gnomAD v4.1 all-comers at total AF = 6.312549474606508e-06 (AC 10 / AN 1,584,146; exome 9/1,431,988; genome 1/152,158), i.e. approximately 16-fold below the 0.0001 default threshold, with zero homozygotes (total, exome, genome and every ancestry). Two features were weighed explicitly before calling this met. (1) Ancestry: the highest-frequency population is East Asian at AF 0.0002042112906153567 (9/44,072 alleles, 0 homozygotes) - 9 of the 10 total alleles derive from this one subpopulation, so the apparent enrichment rests on a single ancestry with no carrier genotype data supporting a common polymorphism elsewhere. (2) Threshold boundary: the grpmax filtering allele frequency is 0.00010566 (joint; exome-only 0.00010209), which marginally exceeds the 0.0001 default by about 6%. This is a borderline, single-subpopulation point estimate rather than a frequency incompatible with pathogenicity: the variant is still ultra-rare globally (approximately 1 allele per 158,000 alleles) and entirely absent from the older, all-comers gnomAD v2.1 dataset that the threshold was calibrated against. Applying PM2 at supporting strength is therefore justified, but the sub-threshold margin on grpmax FAF is close enough that a gene-specific (or non-cancer-cohort) threshold could flip the call, so this criterion is flagged for human review. Corroborating (not required): the ClinVar Labcorp submission asserted PM2 on the basis that the variant was 'not present in population databases' - an assertion made before the v4.1 record existed and now known to be inexact, since the variant is present at very low frequency in v4.1.

Rule

PM2 supporting: variant absent from or at extremely low frequency in population databases; default threshold frequency <= 0.0001 for genes without a specific specification (SVI recommendation to apply PM2 at supporting strength; PMID:25741868).

Evidence considered

• gnomAD v4.1 all-comers totals for chr19-50417273-T-C: AC 10 / AN 1,584,146, AF 6.312549474606508e-06, total homozygotes 0; exome 9/1,431,988 (AF 6.284968868454205e-06, hom 0); genome 1/152,158 (AF 6.572115826969334e-06, hom 0); grpmax FAF 0.00010566; exome grpmax FAF 0.00010209; genome grpmax FAF null.
• gnomAD v4.1 ancestry data: East Asian AF 0.0002042112906153567 (9/44,072 alleles, 0 homozygotes; EAS_XX 6/22,576; EAS_XY 3/21,496) is the only ancestry with observations; NFE 1/1,167,734 (AF 8.563594106191992e-07, hom 0); 0 alleles in AFR (AN 74,386), SAS (AN 88,898), AMR (AN 58,276), FIN (AN 53,284), ASJ (AN 29,154), MID (AN 5,972), AMI (AN 912) and remaining (AN 61,458); best_subpop = eas.
• gnomAD v2.1 all-comers query (GRCh37 19-50920530-T-C): search_status 'absent', found = false; the variant is not present in this older all-comers dataset.
• gnomAD-Canada v1.0 (HostSeq) API query: search_status 'absent' with error 'timed out' - no frequency data returned.
• Source selection: no ClinGen CSPEC (cspec found = false), no gene VCEP directory or user specification (gene_vcep_dir null, no gene VCEP files), and the final classification framework resolved to generic ACMG/AMP 2015 fallback; therefore no non-cancer or exome-only population source was mandated and the all-comers gnomAD v2.1/v4.1 totals were used as the default, consistent with the governing-specification precedence rule.
• Threshold applied verbatim from the supplied published calibration block: PM2 supporting at frequency <= 0.0001 (SVI recommendation to downgrade PM2 to supporting, ClinGen SVI 2020; PMID:25741868), used because no POLD1-specific threshold exists; the ACMG/AMP 2015 PM2 criterion definition is carried by the registry entry keyed to PMID:25741868.
• ClinVar record 647102: submission SCV000941306 (Labcorp Genetics/Invitae) lists PM2 among its criteria with the supporting sentence 'this variant is not present in population databases (gnomad no frequency)' (evidence_signals criterion PM2, confidence 0.8). This corroborates the direction of the frequency finding but was based on absence alone and predates the gnomAD v4.1 record; it was not used to determine the verdict.
• Literature screened for population-frequency content relevant to PM2: PMID:17576681, PMID:9536098 and PMID:25394175 report no POLD1 allele-frequency or carrier data; the only POLD1-relevant population fact in the case evidence is the gnomAD data above.

Still missing

• The grpmax filtering allele frequency (0.00010566) marginally exceeds the 0.0001 generic default, while the all-comers total AF (6.312549474606508e-06) is well below it; no POLD1-specific or non-cancer-cohort threshold exists to settle which metric governs, so PM2 was applied at supporting strength and flagged for human review.
• 9 of the 10 gnomAD v4.1 alleles are East Asian (AF 0.0002042112906153567 with 0 homozygotes); no ancestry-matched control frequency threshold exists for this gene, and no additional ancestry-specific cohort data were available.
• gnomAD v2.1 non-cancer and gnomAD v3.1 non-cancer (exome-only/genome-only) subsets were not retrieved for this case; because no governing specification mandates them, the all-comers GNOMAD_V2_1 and GNOMAD_V4_1 totals were treated as the default source.
• gnomAD-Canada v1.0 (HostSeq) returned no data (lookup timed out), so a second independent population dataset could not be added to the frequency estimate.

Sources

Reasoning

Scope: NM_002691.4:c.3218+4T>C is an intronic substitution at donor-site position +4, i.e. a splice-region variant, so BP4 is evaluated on the splice-prediction (SpliceAI) path only; the missense REVEL path is not applicable. SpliceAI Lookup returned a usable prediction with max delta score 0.025 (DS_DL 0.025 donor loss, DS_AG 0.002, DS_AL 0.002, DS_DG 0.000), which is at or below the generic SpliceAI BP4 threshold of <= 0.1 for supporting evidence (Jaganathan et al. 2019, Cell; PMID:30661751); the value is also well below the 0.2 PP3 supporting cutoff, so the same prediction does not simultaneously support PP3 and there is no double-counting between PP3 and BP4. Both SpliceAI and Pangolin in the same lookup agree on a negligible impact (Pangolin SG 0.01 / SL -0.014). BP4 is therefore met at supporting strength only. Corroborating in-silico result: the variant is extremely rare in gnomAD v4.1 (AF 6.31e-06) and absent from gnomAD v2.1, which is not BP4 evidence but does not undermine a benign computational call. The evidence is limited to computational prediction; no RNA-based or minigene assay of donor-site usage exists for this position, which is why BP4 is kept at supporting strength and flagged as prediction-only. No POLD1-specific VCEP/CSPEC or PP3/BP4 lookup spreadsheet exists to override or strengthen this call (ClinGen CSPEC not found; vcep_materials empty), so the generic calibrated cutoff was applied. PMID:17576681 was reviewed for position-specific aberrant-splice-site rates and found to contain no data on this variant or on position +4 specifically, so it is not cited as BP4 support.

Rule

SpliceAI max delta <= 0.1 (BP4 supporting) — Jaganathan et al. 2019, Cell (PMID:30661751); governing VCEP absent, generic calibration applied.

Evidence considered

• SpliceAI Lookup result for NM_002691.4:c.3218+4T>C (hg37): max delta score 0.025, DS_DL 0.025, DS_AG 0.002, DS_AL 0.002, DS_DG 0.000; Pangolin SG 0.01 / SL -0.014; scrape status result_found, screenshot captured at screenshots/spliceai.png.
• Threshold applied verbatim: SpliceAI max delta <= 0.1 = BP4 supporting (Jaganathan et al. 2019, Cell; PMID:30661751); 0.025 is below the cutoff, so BP4 is met at supporting strength.
• Path selection: because the variant is intronic/splice-region, only the SpliceAI path was used for BP4; the absent REVEL score (found = false) was not substituted, and the same SpliceAI value was not also counted toward PP3, which is separately not met.
• No POLD1 VCEP/CSPEC or gene-specific PP3/BP4 lookup material exists (ClinGen CSPEC not found; vcep_materials.json contains no summaries, files or index guidance), so there is no pre-assigned gene-specific BP4 code to apply and the generic cutoff governs.
• Population context only (not BP4 evidence): gnomAD v4.1 AF 6.31e-06 (10/1,584,146 alleles, 0 homozygotes; grpmax FAF 0.000106) and absent from gnomAD v2.1.
• PMID:17576681 (DBASS5) reports that point mutations at consensus positions +3, +4, +6 and -2 each accounted for <3% of single-nucleotide substitutions in its aberrant 5' splice-site dataset; the paper does not name this variant or POLD1, so it is background only and is not cited as BP4 support.

Still missing

• BP4 rests on computational prediction alone; no RNA/cDNA or minigene assay of donor-site usage at c.3218+4T>C was available to confirm the absence of a splicing effect.
• No POLD1-specific (VCEP/CSPEC) splice-prediction calibration exists, so a generic SpliceAI cutoff was used; a gene-specific specification could change the strength assigned.
• BayesDel and REVEL results are unavailable, but neither is part of the splice path for BP4 and neither was used.

Sources

Reasoning

NM_002691.4:c.3218+4T>C is an intronic single-nucleotide substitution located 4 nucleotides into the intron 26 donor region (VariantValidator: 26i, MANE Select/RefSeq Select transcript NM_002691.4, predicted protein consequence NP_002682.2:p.?). The generic ClinGen SVI PVS1 decision tree (Abou Tayoun et al. 2018, PMC6185798) assigns PVS1 to null variants of the class nonsense / frameshift / canonical +/-1,2 splice site / initiation codon / single- or multi-exon deletion. A +4 donor position is not part of the invariant GT dinucleotide at +1,+2 and does not sit in any of those default null-variant buckets, so full-strength PVS1 is not available for this variant; the variant-level PVS1 scaffold likewise returns canonical_splice_consensus = false and apply_generic_pvs1_framework = false. The predicted-splice-impact route also fails: SpliceAI max delta is 0.025 (DS_DL 0.025, DS_DG 0.000, DS_AG 0.002, DS_AL 0.002; Pangolin SG 0.010, SL -0.014), far below the 0.2 supporting and 0.5 moderate SpliceAI cutoffs (Jaganathan et al. 2019, PMID:30661751), so no gain or loss of a donor/acceptor is predicted. No RNA/cDNA/minigene data exist for this transcript, so an aberrant transcript cannot be demonstrated either. POLD1 germline loss-of-function is an established disease mechanism (gene-level gate eligible, literature-only), but that only makes PVS1 assessable in principle; it does not supply the variant-level null consequence PVS1 requires. No strength tier is assigned because the criterion is not met and no named SVI/VCEP downgrade rule was reached. Exon-importance and NMD considerations were therefore not reached: they become decisive only after a splice/null consequence is established, which is not the case here.

Rule

ClinGen SVI PVS1 decision tree (Abou Tayoun et al. 2018, PMC6185798), generic ACMG/AMP 2015 criterion PVS1 (PMID:25741868); variant is outside the default null-variant classes, so no strength tier applies.

Evidence considered

• VariantValidator normalisation places NM_002691.4:c.3218+4T>C in intron 26 (variant_exonic_positions start_exon/end_exon = 26i) on the MANE Select / RefSeq Select POLD1 transcript, with predicted protein consequence NP_002682.2:p.? and no protein-level annotation.
• The generic ClinGen SVI PVS1 decision tree (Abou Tayoun et al. 2018, PMC6185798) restricts full-strength PVS1 to nonsense, frameshift, canonical +/-1,2 splice-site, initiation-codon and single/multi-exon deletion variants; a donor-site +4 position lies outside the invariant GT dinucleotide and outside all of those default buckets.
• The variant-level PVS1 scaffold classifies this change as variant_bucket 'other' with canonical_splice_consensus = false, apply_generic_pvs1_framework = false and no suggested default strength.
• SpliceAI max delta 0.025 (DS_DL 0.025, DS_DG 0.000, DS_AG 0.002, DS_AL 0.002; Pangolin SG 0.010, SL -0.014) is far below the 0.2 supporting and 0.5 moderate SpliceAI thresholds (Jaganathan et al. 2019, Cell, PMID:30661751), so no altered donor or acceptor usage is predicted.
• No RNA/cDNA, RT-PCR or minigene data are available for this transcript, so a predicted-splice PVS1 downgrade path (PVS1_Strong/PVS1_Moderate requiring supporting splice evidence) is not reached either.
• POLD1 germline loss of function is an established disease mechanism in the gene-level context (pvs1_gene_gate = eligible; literature-only, no official CSPEC/VCEP), which makes PVS1 assessable but supplies no variant-level null consequence.
• Buratti et al. (PMID:17576681) compiled mutation-activated aberrant 5' splice sites and show that substitutions within the 9-nt donor consensus can weaken the authentic site and activate cryptic/de novo sites, but report that positions +3, +4, +6 and -2 each accounted for <3% of single-nucleotide substitutions, giving no position-specific support for a +4T>C change.
• Zhang (PMID:9536098) defines the human 5' splice-site consensus as AG|GTRAGT, in which +4 is the degenerate R position rather than an invariant nucleotide, consistent with a +4 substitution not being a canonical splice-site null allele.
• ClinVar (ClinVarID 647102, two single-submitter germline entries, both Uncertain significance, no expert-panel submission) was reviewed only for the variant's transcript/exon annotation and cited PMID trail; no PVS1-relevant evidence and no RNA data were present, and the classification itself was not used as evidence.

Still missing

• No RNA/cDNA or minigene data demonstrate or exclude an aberrant splice product for c.3218+4T>C; the only splice evidence is the SpliceAI prediction (max delta 0.025).
• No POLD1-specific ClinGen CSPEC/VCEP decision tree exists, so the generic SVI PVS1 framework was applied.

Sources

Reasoning

PS3 requires well-established functional studies demonstrating a damaging effect of this specific variant. No such study exists for NM_002691.4(POLD1):c.3218+4T>C. ClinVar VCV000647102 (ClinVarID 647102) holds exactly two germline submissions for this variant - Ambry Genetics SCV001180631 and Labcorp Genetics (formerly Invitae) SCV000941306 - both Uncertain significance, both 'criteria provided, single submitter', with no expert-panel review. Neither submission claims PS3 and neither reports an assay result; the only submitter criterion signal captured from those records is PM2 (absent from population databases; gnomAD v2.1 absent, gnomAD v4.1 total AF 6.31e-06). The complete citation trail attached to those submissions was retrieved and read in full text: PMID:17576681 (DBASS5 compilation of mutation-induced aberrant 5' splice sites), PMID:9536098 (statistical survey of human exon/flanking-intron features) and PMID:25394175 (ACMG/NSGC cancer-predisposition referral guideline). None of the three mentions POLD1 or c.3218+4T>C, none contains a minigene, RT-PCR, cDNA/RNA-seq or protein-activity assay of any POLD1 allele, and none reports a patient carrying this substitution; the DBASS5 paper supplies only mechanism-level background (point mutations in the 9-nt 5' splice-site consensus, with +4 among the least frequently mutated positions, each of +3/+4/+6/-2 accounting for <3% of substitutions). A broader gene-level POLD1 germline sweep run in this case returned only gene/disease-context papers (e.g. PPAP case reports and reviews) and likewise no variant-specific functional data. SpliceAI max delta 0.025 (Pangolin SG 0.01, SL -0.014) is a computational prediction, not the assay evidence PS3 demands, so it cannot support PS3 at any strength. The criterion's condition is therefore unsatisfied: PS3 is not met, and the variant's intronic donor +4 position means RNA-based donor-site usage data (minigene or patient RNA) would be the appropriate evidence to look for if it ever becomes available. No strength is assigned because the criterion is unmet and no calibration decision arises.

Rule

PS3 (ACMG/AMP 2015, PMID:25741868): well-established functional studies show a damaging effect on the gene or gene product. No POLD1-specific VCEP/CSPEC specification was available, so the generic definition was applied; condition not satisfied because no variant-specific functional study exists.

Evidence considered

• ClinVar VCV000647102 (ClinVarID 647102) records NM_002691.4(POLD1):c.3218+4T>C as Uncertain significance from 2 clinical laboratories (Ambry Genetics SCV001180631; Labcorp/Invitae SCV000941306), each 'criteria provided, single submitter', no expert panel; classifications are Uncertain significance for 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome'.
• The only criterion signal extracted from either ClinVar submission is PM2 ('this variant is not present in population databases (gnomad no frequency)'); no submission asserts PS3, BS3, or any functional assay result, so the ClinVar record carries no functional evidence in either direction.
• Full texts of the three PMIDs cited by the ClinVar submitters were reviewed (PMID:17576681, PMID:9536098, PMID:25394175) and none mentions POLD1 or c.3218+4T>C nor contains a minigene, RT-PCR, RNA-seq or protein-activity assay of any POLD1 allele; the fourth triaged PMID (28492532, Sherloc) was not retrievable as full text and is an ACMG-criteria refinement paper, not a functional study.
• PMID:17576681 (DBASS5) provides only mechanism-level context relevant to a +4 donor substitution: point mutations within the 9-nt 5' splice-site consensus can inactivate the authentic donor and activate a nearby cryptic/de novo site, while +4 is among the least frequently mutated positions ('positions +3, +4, +6 and -2 each accounted for <3% of all the single-nucleotide substitutions'). This is background on splicing biology, not an assay of this variant.
• SpliceAI Lookup returned max delta 0.025 (DS_DL 0.025, DS_AG 0.002, DS_AL 0.002, DS_DG 0.000; Pangolin SG 0.01 / SL -0.014), i.e. no predicted splice impact; as in-silico prediction this is excluded from PS3 and is handled on the in-silico path (BP4), not here.
• The governing specification for this case is the generic ACMG/AMP 2015 framework (PMID:25741868): cspec lookup returned found=false, vcep_materials.json is empty (no summaries, files, user files or index guidance), and final_classification_framework resolved to source 'generic_acmg_fallback' with official_clingen=false, so the generic PS3 definition ('well-established functional studies show a damaging effect on the gene or gene product') was applied unmodified.
• The gene-level POLD1 literature sweep performed for this case (5 supporting papers on POLD1/POLD3 germline disease, e.g. PMID:36291559, PMID:41487566) documents loss-of-function/proofreading-deficiency as the POLD1 disease mechanism and contains no functional data on this variant.
• No case file in the evidence bundle (evidence.json, literature_pass.json, paper_extracts.json, vcep_materials.json, case_summary.json, prefetch.json) reports a functional assay of c.3218+4T>C; the triaged literature themes were 'splicing_rna' and 'case_observation' only.

Still missing

• No RNA-based (RT-PCR/cDNA/RNA-seq) or minigene assay measuring donor-site usage at POLD1 c.3218+4T>C was found in this case; a study demonstrating aberrant splicing at the exon ending at c.3218 would be required to meet PS3.
• The case's functional-evidence universe is the ClinVar citation trail (4 PMIDs triaged, 3 full texts retrieved) plus a gene-level POLD1 germline sweep; a dedicated search for POLD1 intronic splice-assay literature was not performed, so the absence of assay evidence is established only against the retrieved sources.
• No POLD1 protein-level readout (e.g. proofreading-fidelity or mutator-phenotype assays used for PPAP missense alleles) is applicable or documented for this intronic change, since its protein consequence is reported as p.? (unknown).

Sources

Reasoning

PS4 requires case-control or cohort enrichment showing that this variant (or the same-variant class) occurs significantly more often in affected individuals than controls, with the enrichment statistic meeting or exceeding the governing threshold (>= if the framework defines an odds ratio / rate ratio). No governing specification exists for POLD1: no ClinGen CSPEC was found and the case falls back to generic ACMG/AMP 2015, which supplies no numeric PS4 cutoff; therefore no numeric PS4 threshold is available to compare against. No numeric enrichment value was extracted from any source either, so no operator/direction comparison was performed (none is possible without a value). Evidence reviewed: ClinVar VCV000647102 (NM_002691.4(POLD1):c.3218+4T>C) carries two clinical-laboratory submissions, both classified Uncertain significance (Ambry Genetics SCV001180631; Labcorp Genetics/formerly Invitae SCV000941306), with no case counts, no unaffected-control data, and no case-control series. The literature assembled for this case (three full texts) contains no report of this variant in any patient, so there is no proband cohort to enrich. gnomAD v4.1 shows the variant at total AF 6.3125e-06 (10/1,584,146 alleles; grpmax FAF 1.0656e-04) with zero homozygotes, i.e. an ultra-rare variant, which is consistent with rarity but provides no case-control enrichment signal. Conclusion: PS4 is not met because no case-control/cohort enrichment evidence exists for this variant.

Rule

Generic ACMG/AMP 2015 PS4: variant prevalence in affected individuals is significantly increased compared with controls; requires an enrichment statistic (e.g. odds ratio) meeting or exceeding the governing threshold, evaluated with the operator read from the governing specification.

Evidence considered

• ClinVar VCV000647102 lists exactly this variant (NM_002691.4(POLD1):c.3218+4T>C) as Uncertain significance with two germline clinical-laboratory submissions (Ambry Genetics SCV001180631; Labcorp Genetics/formerly Invitae SCV000941306); neither submission provides case counts, affected/unaffected ascertainment, or an enrichment statistic.
• ClinVar submission audit reports expert_panel_submissions = 0 and no case-control or cohort content in either submission.
• gnomAD v4.1 reports total AF 6.3125e-06 (10/1,584,146 alleles; exome 9/1,431,988, genome 1/152,158), zero homozygotes, and grpmax FAF 1.0656e-04 (East Asian AF 2.042e-04, 9/44,072) - an ultra-rare observation that cannot be converted into an enrichment statistic without an affected cohort.
• gnomAD v2.1 reports the variant absent, and gnomAD-Canada v1.0 returned no result (query timed out); neither adds case-control information.
• The one literature source triaged as a candidate for PS4 (the ACMG/NSGC cancer-referral practice guideline, PMID:25394175) was screened for POLD1 and for this variant and contains neither, so it supplies no enrichment evidence.
• No governing POLD1 VCEP or ClinGen CSPEC was available (no cspec URL, no gene VCEP directory, generic ACMG fallback framework), so no framework-specific PS4 numeric threshold exists; consequently no threshold comparison was made.

Still missing

• No case-control, case-cohort, or enrichment study (odds ratio / rate ratio) for NM_002691.4:c.3218+4T>C in POLD1-associated polyposis/colorectal cancer was identified in any source.
• No proband-level case series reporting this exact variant was found, so no affected-vs-control counts can be assembled.
• No POLD1-specific VCEP/CSPEC defines a numeric PS4 threshold; only generic ACMG 2015 PS4 semantics are available.

Sources

Reasoning

PM4 (ACMG/AMP 2015, PMID:25741868) requires a protein length change caused by an in-frame deletion/insertion in a non-repeat region or by a stop-loss variant. NM_002691.4:c.3218+4T>C is a single-nucleotide intronic substitution at the exon 26 donor +4 position with predicted protein consequence NP_002682.2:p.? - it is not itself an in-frame indel and not a stop-loss. The only route by which such a variant could produce a length change is aberrant in-frame exon skipping, and that route is not supported: SpliceAI max delta is 0.025, well below the 0.2 supporting threshold (Jaganathan et al. 2019, PMID:30661751), and there is no RNA/cDNA, RT-PCR or protein evidence demonstrating an aberrant transcript or an altered protein length. Protein length change is therefore neither present nor predicted, and no strength tier is assigned.

Rule

ACMG/AMP 2015 PM4 (protein length change: in-frame indel in a non-repeat region or stop-loss variant) (PMID:25741868)

Evidence considered

• PM4 requires a protein length change from an in-frame deletion/insertion in a non-repeat region or from a stop-loss variant; NM_002691.4:c.3218+4T>C is an intronic single-nucleotide substitution and is neither an in-frame indel nor a stop-loss.
• VariantValidator reports the predicted protein consequence as NP_002682.2:p.? with no length change annotated, and the variant-level assessment records consequence_class 'unknown' / variant_bucket 'other'.
• SpliceAI max delta 0.025 (DS_DL 0.025; Pangolin SG 0.010, SL -0.014) predicts no exon skipping that could generate an aberrant in-frame transcript; the 0.2 supporting and 0.5 moderate SpliceAI cutoffs are from Jaganathan et al. 2019, Cell (PMID:30661751).
• No RNA, cDNA, RT-PCR or protein data are available for this variant, so an in-frame length-altering consequence could not be demonstrated.
• ClinVar (ClinVarID 647102) contains two single-submitter Uncertain significance germline entries with no PM4-level protein-length evidence; its classification was not used as evidence.

Still missing

• No RNA/cDNA or protein-length assay is available to exclude an in-frame consequence generated by altered splicing at this donor site.

Sources

Reasoning

Scope: the variant is NM_002691.4:c.3218+4T>C (NC_000019.9:g.50920530T>C) in POLD1, an intronic substitution at the +4 position of the donor (5') splice site of the exon ending at c.3218. It is not a missense variant and not a truncating/protein-length-changing variant, so PP3 is evaluated on the splice-prediction (SpliceAI) path only; REVEL is not applicable and was not consulted. SpliceAI Lookup returned an available result with max delta score 0.025, composed of DS_DL 0.025 (donor loss), DS_AG 0.002, DS_AL 0.002 and DS_DG 0.000, with the highest delta localised 2 bp from the donor site (DP_DL -2). Against the generic SpliceAI calibration supplied for this review, PP3 requires max delta >= 0.5 (moderate) or >= 0.2 (supporting); 0.025 is far below both. SpliceAI therefore does not predict a splice-altering effect and PP3 is not met at any strength. No other PP3-eligible computational evidence exists for a non-missense variant: REVEL was not found (not applicable to an intronic change), and BayesDel has no established generic ACMG-strength calibration and is not available in this case, so it cannot be substituted. No governing VCEP/CSPEC or gene-specific PP3/BP4 specification exists for POLD1 (ClinGen CSPEC not found; vcep_materials carries no summaries, files or index guidance), so the generic calibrated thresholds were used and no pre-assigned gene-specific PP3 code was available to apply. The two splicing-themed papers retrieved (PMID:17576681, PMID:9536098) are background on aberrant 5' splice sites and splice-site consensus statistics; neither names this variant and neither is used as PP3 support.

Rule

SpliceAI max delta >= 0.5 (PP3 moderate) or >= 0.2 (PP3 supporting) — Jaganathan et al. 2019, Cell (PMID:30661751); governing VCEP absent, generic calibration applied.

Evidence considered

• Variant consequence for scope: intronic c.3218+4T>C at the donor splice site +4 position of POLD1 (genomic NC_000019.9:g.50920530T>C; GRCh38 19-50417273-T-C); no amino-acid change (NP_002682.2:p.?). Therefore the splice-prediction path, not the missense REVEL path, governs PP3/BP4.
• SpliceAI Lookup (spliceailookup.broadinstitute.org, transcript NM_002691.4, hg37, basic, distance 500) returned spliceai_available = true with max delta score 0.025: DS_DL 0.025, DS_AG 0.002, DS_AL 0.002, DS_DG 0.000 (DP_DL -2, DP_AG 121, DP_AL -225, DP_DG -4); Pangolin SG 0.01 / SL -0.014.
• Thresholds applied verbatim: SpliceAI PP3 moderate at max delta >= 0.5 and PP3 supporting at max delta >= 0.2 (Jaganathan et al. 2019, Cell; PMID:30661751). 0.025 does not reach either cutoff, so PP3 is not met.
• REVEL score not available in this case (found = false) and in any case out of path for an intronic variant; BayesDel score not available (found = false) and has no established generic ACMG-strength PP3 calibration, so neither can support PP3.
• No POLD1 VCEP/CSPEC specification exists (ClinGen CSPEC not found; vcep_materials.json holds no summaries, files or index guidance), so no gene-specific pre-assigned PP3 code was available and generic calibration was used.
• PMIDs 17576681 (DBASS5 compilation of aberrant 5' splice sites) and 9536098 (statistical features of human exons and flanking regions) were reviewed via the paper-extract layer: neither names c.3218+4T>C or POLD1, so neither is cited as PP3 evidence; they supply only general splice-site background.

Still missing

• No RNA/minigene functional data specifically measuring donor-site usage for c.3218+4T>C, so the prediction-based PP3 negative cannot be confirmed experimentally.
• No POLD1-specific (VCEP/CSPEC) SpliceAI or PP3 calibration was available; generic SpliceAI cutoffs were applied.

Sources

Reasoning

PP5 requires a reputable, independent source (in current practice, an exact-variant ClinGen/VCEP expert-panel classification) that the variant is pathogenic. The governing PP5 rule applied here is the framework's own: only an exact-variant expert-panel Pathogenic/Likely pathogenic classification triggers PP5, and ordinary laboratory submissions, aggregate ClinVar labels, or non-expert assertions must never trigger it. ClinVar VCV000647102 (NM_002691.4(POLD1):c.3218+4T>C) carries exactly two germline submissions, both from clinical laboratories and both classified Uncertain significance: Ambry Genetics (SCV001180631, 'criteria provided, single submitter') and Labcorp Genetics/formerly Invitae (SCV000941306, 'criteria provided, single submitter'). The submission audit records expert_panel_submissions = 0, is_expert_panel = false for both submissions, and the ClinVar record's expert_panels object is empty; the aggregate review status is 'criteria provided, multiple submitters, no conflicts' (2 stars), which is an ordinary multi-laboratory VUS label, not an expert-panel assertion. No ClinGen CSPEC or POLD1 VCEP exists for this gene (no cspec URL; final classification framework falls back to generic ACMG), so no expert-panel classification path exists for this variant at all. Because no exact-variant expert-panel Pathogenic/Likely pathogenic classification exists, PP5 is not met. No numeric threshold is involved in PP5, so no operator/direction comparison applies.

Rule

PP5 (as governed for this case): a reputable independent source must assert the exact variant is pathogenic; only an exact-variant ClinGen expert-panel Pathogenic/Likely pathogenic classification triggers PP5, and ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions must never trigger it.

Evidence considered

• ClinVar VCV000647102 (NM_002691.4(POLD1):c.3218+4T>C) has exactly two submissions, both classified Uncertain significance: Ambry Genetics SCV001180631 (review status 'criteria provided, single submitter') and Labcorp Genetics/formerly Invitae SCV000941306 (review status 'criteria provided, single submitter'); both are clinical testing, germline, non-expert-panel submissions.
• ClinVar submission audit records counts: expert_panel_submissions = 0, and each submission has is_expert_panel = false; the record's expert_panels object is empty.
• ClinVar aggregate review status is 'criteria provided, multiple submitters, no conflicts' (2 stars), which is a multi-laboratory VUS label rather than an expert-panel classification.
• No ClinGen CSPEC and no POLD1 gene-specific VCEP material was found; the final classification framework resolves to generic ACMG fallback, so no expert-panel assertion path exists for this variant.

Still missing

• No ClinGen/VCEP expert-panel submission (or any expert-panel Pathogenic/Likely pathogenic assertion) exists for NM_002691.4:c.3218+4T>C in ClinVar.
• No POLD1 VCEP/CSPEC exists, so no expert-panel pathway is currently available for this gene.

Sources

Reasoning

BA1 (stand-alone benign) requires an allele frequency >= 0.05 in a population dataset that is appropriate for the gene-disease context (ACMG/AMP 2015 stand-alone benign frequency, generic default supplied for this gene with no specification). No POLD1-specific framework exists: ClinGen CSPEC was not found, no gene VCEP directory or user-supplied specification is present, and the final classification framework resolved to generic ACMG/AMP 2015, so the generic default applies and no non-cancer/exome-only subset is mandated. In gnomAD v4.1 all-comers totals the variant is present at total AF = 6.312549474606508e-06 (AC = 10, AN = 1,584,146; exomes 9/1,431,988; genomes 1/152,158) with zero homozygotes (total_hom = 0; exome_hom = 0; genome_hom = 0). It is absent from gnomAD v2.1 (all-comers). The highest ancestry-specific frequency, East Asian (EAS) AF = 0.0002042112906153567 (9/44,072 alleles, 0 homozygotes), and the grpmax filtering allele frequency of 0.00010566, are still roughly three to five orders of magnitude below the 0.05 stand-alone threshold. Intra-population data are not consistent with a benign standing polymorphism for this gene: the variant is rare in every ancestry surveyed (0/74,386 AFR, 0/88,898 SAS, 0/58,276 AMR, 0/53,284 FIN, 1/1,167,734 NFE, 0/29,154 ASJ, 0/5,972 MID, 0/912 AMI). A redundant gnomAD-Canada v1.0 (HostSeq) lookup returned no result (request timed out), leaving no additional cohort data that could raise the frequency toward the threshold. BA1 is therefore not met, at any strength.

Rule

BA1 stand-alone benign: allele frequency >= 0.05 in an appropriate population dataset (ACMG/AMP 2015, PMID:25741868; generic default, no POLD1 specification available).

Evidence considered

• gnomAD v4.1 (GRCh38 chr19-50417273-T-C) all-comers totals: total AC 10 / AN 1,584,146, total AF 6.312549474606508e-06, total homozygotes 0; exome AC 9 / AN 1,431,988 (AF 6.284968868454205e-06, hom 0); genome AC 1 / AN 152,158 (AF 6.572115826969334e-06, hom 0).
• gnomAD v4.1 ancestry breakdown: EAS 9/44,072 (AF 0.0002042112906153567, hom 0; EAS_XX 6/22,576, EAS_XY 3/21,496), NFE 1/1,167,734 (AF 8.563594106191992e-07, hom 0), and 0 alleles in AFR (AN 74,386), SAS (AN 88,898), AMR (AN 58,276), FIN (AN 53,284), ASJ (AN 29,154), MID (AN 5,972), AMI (AN 912), and remaining (AN 61,458); grpmax FAF 0.00010566 (joint), exome grpmax FAF 0.00010209.
• gnomAD v2.1 (GRCh37 19-50920530-T-C) all-comers: search status 'absent', found = false - no observed alleles in the v2.1 exome dataset.
• gnomAD-Canada v1.0 (HostSeq genomes) API query for 19-50417273-T-C returned search_status 'absent' with error 'timed out' - no usable frequency data.
• Threshold applied verbatim from the supplied published calibration block: BA1 stand-alone benign at AF >= 0.05 (ACMG/AMP 2015 criteria, PMID:25741868); no POLD1 specification exists to override it. The generic ACMG/AMP 2015 criteria definitions and default frequency thresholds were taken from the supplied calibration data, keyed here to the registry entry that carries ACMG/AMP 2015 (PMID:25741868).
• Disease context considered for threshold appropriateness: POLD1 germline disease is an adult-onset, dominantly inherited cancer predisposition (ClinVar conditions for this variant: 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome'), so a stand-alone benign frequency call would require a frequency far above any plausible gene-specific threshold; the observed ~6e-06 is orders of magnitude below even the most conservative estimate.
• Literature screened for population-frequency content relevant to BA1: PMID:17576681 (DBASS5 aberrant 5' splice-site compilation), PMID:9536098 (statistical survey of human exon/flanking sequence features), PMID:25394175 (ACMG/NSGC cancer-predisposition referral guideline). None reports allele frequency, cohort prevalence, or homozygote data for POLD1 or for c.3218+4T>C; no population evidence for BA1 was obtained from them.

Still missing

• No POLD1-specific ClinGen CSPEC, gene VCEP specification, or local gene framework exists, so BA1 could not be evaluated against a gene-calibrated stand-alone frequency threshold; the generic ACMG/AMP 2015 default (>= 0.05) was used.
• gnomAD-Canada v1.0 (HostSeq) returned no data for this variant (lookup timed out), so an additional genome cohort could not corroborate the frequency estimate.
• The observed frequency is ~8,000-fold below the 0.05 threshold, so none of these gaps could plausibly change the BA1 verdict.

Sources

Reasoning

BS1 (strong benign) requires an allele frequency greater than expected for the disorder, evaluated at the generic SVI default of AF >= 0.01 (Whiffin et al. 2017, PMID:28518168) because no POLD1-specific specification exists to supply a gene-calibrated threshold (ClinGen CSPEC absent; no gene VCEP specification; final classification framework resolved to generic ACMG/AMP 2015). The relevant observations are the all-comers totals, since no specification mandates a non-cancer or exome-only source for POLD1. gnomAD v4.1 all-comers total AF is 6.312549474606508e-06 (10/1,584,146 alleles, 0 homozygotes), and the variant is absent from gnomAD v2.1. The most frequent ancestry, East Asian, reaches only AF 0.0002042112906153567 (9/44,072, 0 homozygotes), and the grpmax FAF is 0.00010566; even that highest subpopulation value is more than 45-fold below 0.01 (the total frequency is ~1,585-fold below it). Applying BS1 at 0.01 (or at any gene-specific threshold derived from the low prevalence and reduced penetrance of POLD1-related polymerase proofreading-associated polyposis, which would be lower still) is not supported. BS1 is not met.

Rule

BS1 strong benign: allele frequency greater than expected for the disorder, generic SVI default AF >= 0.01 (Whiffin et al. 2017, PMID:28518168; per-gene derivation preferred but no POLD1 specification available).

Evidence considered

• gnomAD v4.1 all-comers totals for chr19-50417273-T-C: AC 10 / AN 1,584,146, AF 6.312549474606508e-06, homozygotes 0; exome AF 6.284968868454205e-06 (9/1,431,988); genome AF 6.572115826969334e-06 (1/152,158).
• Highest-ancestry values in gnomAD v4.1: East Asian AF 0.0002042112906153567 (9/44,072 alleles, 0 homozygotes), EAS_XX 0.0002657689581856839 (6/22,576), EAS_XY 0.00013956084852995905 (3/21,496); grpmax FAF 0.00010566; NFE AF 8.563594106191992e-07 (1/1,167,734).
• gnomAD v2.1 all-comers query: 'Absent from gnomAD v2.1' (found = false), i.e. no controls with the allele in the older exome dataset.
• Threshold applied verbatim from the supplied published calibration block: BS1 strong benign at AF >= 0.01, generic SVI default with per-gene derivation preferred (Whiffin et al. 2017, PMID:28518168); no POLD1 specification exists to override it. The ACMG/AMP 2015 BS1 criterion definition itself is carried by the registry entry keyed to PMID:25741868.
• Disease context: POLD1 germline disease is an adult-onset autosomal-dominant cancer predisposition (ClinVar conditions 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome'); with reduced penetrance a Whiffin-style gene-specific BS1 threshold would sit far below the generic 0.01, and the observed frequency remains orders of magnitude below either value.
• Literature screened for frequency or prevalence content relevant to BS1: PMID:17576681, PMID:9536098, and PMID:25394175 contain no allele-frequency, cohort-prevalence, or POLD1 population data bearing on BS1.

Still missing

• No POLD1-specific CSPEC/VCEP or local framework was available, so BS1 was evaluated with the generic SVI default (>= 0.01) rather than a threshold derived from POLD1 PPAP prevalence and penetrance; no prevalence/penetrance figure for POLD1-related disease was present in the case evidence to derive one.
• gnomAD-Canada v1.0 (HostSeq) returned no data (lookup timed out), so an independent cohort could not be added.
• Observed frequency is >45-fold below the generic threshold even at the most enriched ancestry, so a gene-specific threshold would not change the BS1 verdict.

Sources

Reasoning

BS2 requires the variant to have been observed in a healthy adult individual, with full penetrance expected at an early age for the associated disorder (recessive homozygous, dominant heterozygous, or X-linked hemizygous). Neither precondition is satisfied here. First, no homozygous carrier exists in any dataset: gnomAD v4.1 reports total_hom = 0 and 0 homozygotes in every ancestry, including the East Asian group that carries 9 of the 10 total alleles (9/44,072, homozygote_count 0), and the variant is absent from gnomAD v2.1 altogether. Second, the POLD1 disease context is an adult-onset, incompletely penetrant, dominantly acting cancer predisposition (ClinVar conditions for this variant: 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome'), not a disorder for which full penetrance is expected at an early age, so the BS2 premise does not hold for this gene-disease pair. Additionally, at 10 alleles in 1,584,146 (carrier frequency ~1 in 79,000 individuals) the variant is far too rare for a healthy-adult observation to be expected. The blood/tissue origin of the gnomAD observations (presumed unaffected adult controls) adds nothing to BS2 because the condition is late-onset. BS2 is not met.

Rule

BS2 strong benign: observed in a healthy adult individual for a recessive (homozygous), dominant (heterozygous), or X-linked (hemizygous) disorder with full penetrance expected at an early age (ACMG/AMP 2015, PMID:25741868).

Evidence considered

• gnomAD v4.1 all-comers data for chr19-50417273-T-C: total AC 10 / AN 1,584,146, total_hom = 0; exome_hom = 0; genome_hom = 0; grpmax/popmax homozygote counts 0, including East Asian (9/44,072, homozygote_count 0), EAS_XX (6/22,576, hom 0) and EAS_XY (3/21,496, hom 0); all other ancestries 0 alleles.
• gnomAD v2.1 all-comers query: 'Absent from gnomAD v2.1' (found = false) - no carrier at all, homozygous or heterozygous.
• ClinVar record 647102 for NM_002691.4(POLD1):c.3218+4T>C lists conditions 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome' with germline origin in all submissions, establishing an adult-onset dominant cancer-predisposition disease context rather than a fully penetrant early-onset disorder; the ClinVar classification itself (Uncertain significance) was not used as evidence.
• Threshold/premise applied from the supplied published calibration block and the ACMG/AMP 2015 BS2 definition (PMID:25741868, carried by the registry entry for generic ACMG/AMP combination rules): BS2 requires observed homozygosity/heterozygosity in a healthy adult for a disorder with full penetrance expected at an early age.
• Literature screened for early-onset, fully penetrant POLD1 disease or healthy-adult carrier observations: PMID:17576681, PMID:9536098, and PMID:25394175 contain no POLD1 content, no homozygous carrier, and no penetrance data relevant to BS2.

Still missing

• No homozygous carrier is reported and no healthy-adult homozygous or hemizygous observation is documented in any available dataset; gnomAD-Canada v1.0 returned no data (lookup timed out), so one additional cohort could not be screened.
• No POLD1-specific penetrance or age-of-onset quantification was available in the case evidence; the assessment that full penetrance at an early age is not expected rests on the ClinVar disease labels (adult colorectal cancer susceptibility / hereditary cancer-predisposing syndrome) plus the absence of any reported homozygote, which is sufficient because BS2 requires an actual healthy-adult observation that does not exist here.

Sources

Reasoning

BS3 requires well-established functional studies demonstrating no damaging effect on protein function or splicing. No functional study of NM_002691.4(POLD1):c.3218+4T>C exists in either direction: the three full texts reachable from this case's literature trail (PMID:17576681, PMID:9536098, PMID:25394175) contain no POLD1 sequence and no variant-specific assay, and ClinVar VCV000647102 carries no benign-direction functional evidence - both submitters (Ambry Genetics SCV001180631, Labcorp/Invitae SCV000941306) classify the variant Uncertain significance with a single-submitter 'criteria provided' status and no assay data, the only captured submitter criterion being PM2. The only evidence pointing toward a normal consequence is computational: SpliceAI max delta 0.025 (Pangolin SG 0.01, SL -0.014) predicts no significant splice impact, but ACMG/AMP 2015 reserves that kind of result for the in-silico criteria (BP4), which is assessed separately for this variant, and BP4 does not authorize BS3 at any strength. Because the criterion's requirement of a well-established functional assay showing a neutral effect is not met, BS3 is not met; because the variant sits at donor-site position +4, a minigene or patient-RNA donor-usage assay showing normal splicing is the evidence that would satisfy it if it existed.

Rule

BS3 (ACMG/AMP 2015, PMID:25741868): well-established functional studies show no damaging effect on protein function or splicing. No POLD1-specific VCEP/CSPEC specification was available, so the generic definition was applied; condition not satisfied because no variant-specific functional study exists in either direction.

Evidence considered

• No source retrieved for this case reports a functional assay of NM_002691.4(POLD1):c.3218+4T>C demonstrating a neutral or normal effect, so the BS3 condition (well-established functional studies showing no damaging effect) is unsatisfied.
• ClinVar VCV000647102: two germline submissions (Ambry Genetics SCV001180631; Labcorp/Invitae SCV000941306), both Uncertain significance, 'criteria provided, single submitter', no expert panel, no functional-assay evidence; the record's only extracted criterion signal is PM2 (absence from population databases).
• Full texts of PMID:17576681, PMID:9536098 and PMID:25394175 were reviewed for negative or neutral functional data on this variant; none mentions POLD1 or c.3218+4T>C and none contains any POLD1 functional assay, so they provide no BS3 support.
• SpliceAI Lookup gives max delta 0.025 with Pangolin SG 0.01 / SL -0.014 (no predicted splice impact) - a benign-leaning but purely computational result. Under ACMG/AMP 2015 (PMID:25741868) this maps to the in-silico criteria (BP4 supporting, <= 0.1), not to BS3, and the two must not be double-counted.
• Governing specification: cspec found=false, vcep_materials.json empty, final_classification_framework source 'generic_acmg_fallback' (official_clingen=false), so the unmodified generic ACMG/AMP 2015 BS3 definition was applied.
• The gene-level POLD1 sweep in this case addressed POLD1 loss-of-function/proofreading disease mechanisms (e.g. PMID:36291559, PMID:41487566) and contained no variant-specific functional or negative data for c.3218+4T>C.

Still missing

• No functional assay (minigene, RT-PCR/cDNA, RNA-seq, or POLD1 proofreading/activity assay) reporting normal behaviour of c.3218+4T>C was found, so BS3 cannot be met even though computational predictions are benign-leaning.
• The SpliceAI max delta of 0.025 (Pangolin SG 0.01, SL -0.014) is the only direction-of-effect evidence available; it is an in-silico predictor assessed under BP4 and is explicitly not usable as BS3, so a reviewer should not treat the BP4-supporting prediction as functional neutral evidence.
• Absence of BS3 rests on the retrieved ClinVar-citation literature trail (4 PMIDs triaged, 3 full texts retrieved) plus a gene-level POLD1 sweep, not on an exhaustive search for negative splice-assay data.

Sources

Reasoning

BP2 requires a positive phase observation: the variant 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. Neither observation exists for NM_002691.4:c.3218+4T>C. This case contains no affected proband, no pedigree, no parental genotypes and no phase-resolved data at all, so the variant has not been shown in cis or in trans with any pathogenic POLD1 variant. The ClinVar record (VCV000647102) carries only two germline submissions, both Uncertain significance (Ambry Genetics; Labcorp/Invitae), with no accompanying pathogenic allele reported in cis, and the variant has not been published in any case with a phased companion pathogenic variant. The three full-text papers available for this case (PMID:17576681, PMID:9536098, PMID:25394175) contain no POLD1 content and do not mention c.3218+4T>C, so they contribute no cis/trans information. With the required observation absent, BP2 is not met; the criterion is benign evidence and cannot be inferred from the mere absence of a reported second variant. No ClinGen POLD1 VCEP/CSPEC specification exists, so generic ACMG/AMP 2015 criteria govern.

Rule

BP2 (ACMG/AMP 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.

Evidence considered

• ClinVar VCV000647102 lists two germline submissions, both classified Uncertain significance (Ambry Genetics SCV001180631; Labcorp Genetics/formerly Invitae SCV000941306), with no pathogenic POLD1 allele reported in cis or in trans with c.3218+4T>C.
• No affected proband, pedigree, parental genotype, or phase-resolved data are present in this case, so the cis/trans observation that BP2 requires has not been made.
• The three full-text publications available for this case (PMID:17576681, PMID:9536098, PMID:25394175) mention neither POLD1 nor c.3218+4T>C and provide no phase or co-occurrence data.
• No ClinGen POLD1 VCEP or CSPEC specification was found, so the generic ACMG/AMP 2015 BP2 criterion governs this assessment.

Still missing

• No proband or phase-resolved genotype data are available, so co-occurrence of c.3218+4T>C in cis with a pathogenic POLD1 variant could not be directly examined; BP2 is therefore unsupported rather than formally excluded.
• ClinVar submitters for this variant report no companion allele and no phasing information, and no primary publication describes this variant in any individual.

Sources

Reasoning

BP6 requires a reputable, independent source asserting the variant is benign - in current practice an exact-variant ClinGen/VCEP expert-panel Benign/Likely benign classification, with ordinary laboratory submissions and aggregate ClinVar labels explicitly excluded from triggering it. ClinVar VCV000647102 has exactly two submissions, both from clinical laboratories, both classified Uncertain significance (Ambry Genetics SCV001180631; Labcorp Genetics/formerly Invitae SCV000941306); the ClinVar record has no Benign or Likely benign submission, its expert_panels object is empty, and the submission audit records expert_panel_submissions = 0. No ClinGen CSPEC or POLD1 VCEP exists for this gene. Population evidence does not supply benign support either: gnomAD v4.1 total AF is 6.3125e-06 (10/1,584,146 alleles) with grpmax FAF 1.0656e-04 and zero homozygotes - far below any benign-frequency threshold (BA1 >= 0.05; BS1 >= 0.01) - so the variant cannot be called benign on frequency grounds. With no exact-variant expert-panel Benign/Likely benign classification and no benign-direction evidence, BP6 is not met. No numeric BP6 threshold exists in the governing framework, so no operator/direction comparison applies.

Rule

BP6 (as governed for this case): a reputable independent source must assert the exact variant is benign; only an exact-variant ClinGen expert-panel Benign/Likely benign classification triggers BP6, and ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions must never trigger it.

Evidence considered

• ClinVar VCV000647102 contains no Benign or Likely benign submission for this exact variant: both submissions (Ambry Genetics SCV001180631; Labcorp Genetics/formerly Invitae SCV000941306) are classified Uncertain significance, and each is an ordinary clinical-laboratory submission with is_expert_panel = false.
• ClinVar submission audit records expert_panel_submissions = 0 and the record's expert_panels object is empty.
• No ClinGen CSPEC and no POLD1 VCEP material was found; the final classification framework resolves to generic ACMG fallback.
• Population frequency provides no benign-direction support: gnomAD v4.1 total AF 6.3125e-06 (10/1,584,146 alleles), grpmax FAF 1.0656e-04, zero homozygotes - well below benign frequency thresholds (BA1 >= 0.05, BS1 >= 0.01).

Still missing

• No ClinVar Benign or Likely benign submission, and no expert-panel assertion of any kind, exists for NM_002691.4:c.3218+4T>C.
• No POLD1 VCEP/CSPEC exists, so no expert-panel benign classification pathway is currently available for this gene.

Sources

Reasoning

PS2 requires a confirmed de novo occurrence of the variant in a patient with the relevant disease, with both parents tested and shown to lack the variant and with parentage/identity confirmed (ACMG/AMP 2015, PMID:25741868). This case is a variant-level query (POLD1 NM_002691.4:c.3218+4T>C, an intronic +4 splice-region change) and the evidence bundle contains no proband, no phenotype record, no parental testing result, and no trio or maternity/paternity confirmation of any kind. Neither ClinVar submission for variation 647102 asserts a de novo origin: Ambry Genetics (SCV001180631) and Labcorp/Invitae (SCV000941306) both submitted germline, clinical-testing-based Uncertain significance classifications, one of which (SCV001180631) provided no usable PMIDs and the other (SCV000941306) reported only PM2-type population-frequency reasoning. The three full-text papers retrieved via the ClinVar PMID trail (PMID:17576681, PMID:9536098, PMID:25394175) do not mention POLD1 c.3218+4T>C or any POLD1 family data, so no independent de novo observation is available either. With zero parental genotypes and zero proband ascertainment, PS2 is unassessable rather than refuted: the absence of a de novo claim is not evidence that the variant was inherited, and no de novo claim is not evidence that it arose de novo.

Evidence considered

• Case evidence bundle: the only case-level source for this variant is the ClinVar record (variation 647102); it contains two germline clinical-testing submissions, both Uncertain significance (Ambry Genetics SCV001180631; Labcorp Genetics/Invitae SCV000941306), with no expert-panel review (0 expert-panel submissions) and no de novo statement.
• ClinVar submission audit: both submissions are 'criteria provided, single submitter'; SCV000941306's only criterion lead is PM2 (population frequency, 'this variant is not present in population databases'), i.e. no PS2/PM6/PP1/BS4-type evidence was offered by either laboratory.
• ClinVar conditions attached to the record are 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome', consistent with an autosomal dominant cancer-predisposition context in which de novo criteria would be evaluable only with trio data, which are absent.
• Governing framework: no ClinGen CSPEC/VCEP or local gene-specific specification exists for POLD1 in this case (cspec found = false; vcep_materials lists no files, no gene_vcep_dir, final_classification_framework source = generic_acmg_fallback, official_clingen = false), so the ACMG/AMP 2015 standards (PMID:25741868) apply as the governing generic specification.
• Threshold note: PS2's requirement for confirmed parental non-carrier status plus confirmed parentage is the ACMG/AMP 2015 criterion definition (PMID:25741868, available in this case as the generic ACMG combination rules source); no numeric calibration from the supplied threshold block applies to PS2.
• Papers searched for this criterion: PMID:17576681 (DBASS5 aberrant 5' splice-site compilation), PMID:9536098 (statistical survey of exon/flanking sequence features) and PMID:25394175 (ACMG/NSGC cancer-predisposition referral guideline) were each screened for the exact variant and for any de novo/family observation; none mentions POLD1 or c.3218+4T>C.

Still missing

• No proband or affected-individual report for this variant: nothing documents that the variant was observed in a patient with a POLD1-associated phenotype (polymerase proofreading-associated polyposis / colorectal cancer susceptibility).
• No parental testing: neither parent has been genotyped, so confirmed de novo status (variant absent in both biological parents) cannot be established.
• No confirmation of parentage/identity (e.g. informative markers or trio-based relatedness checks) accompanying any parental result.
• No second, independent de novo occurrence in the literature; PMID:17576681, PMID:9536098 and PMID:25394175 do not mention this variant or POLD1 family data.

Sources

Reasoning

PM6 covers an assumed (phenotypically inferred but geno-typically unconfirmed) de novo occurrence, i.e. the variant is observed in a patient consistent with a de novo event but parental samples were unavailable or not tested (ACMG/AMP 2015, PMID:25741868). The minimum requirement still includes an affected proband in whom the variant was found and a clinical context of apparent de novo occurrence (typically absence of the variant in any affected relative and a sporadic presentation). This case supplies neither: there is no proband, no phenotype description, no family history statement, and no statement anywhere in the bundle that the variant occurred sporadically or was not inherited. ClinVar variation 647102 carries two single-submitter Uncertain significance classifications from clinical laboratories (Ambry Genetics SCV001180631; Labcorp/Invitae SCV000941306) with no de novo assertion, no family data, and no expert-panel review. The three retrieved full texts (PMID:17576681, PMID:9536098, PMID:25394175) contain no POLD1 case data on this variant. PM6 is therefore not assessable on the current evidence.

Evidence considered

• ClinVar variation 647102 is the only case-level record for NM_002691.4:c.3218+4T>C: two germline clinical-testing submissions, both Uncertain significance, both 'criteria provided, single submitter' (Ambry Genetics SCV001180631, date last evaluated 2024-12-16; Labcorp/Invitae SCV000941306, date last evaluated 2024-12-02), 0 expert-panel submissions, no de novo or family-history assertion.
• ClinVar submission audit listed SCV000941306 as the only criterion-lead submission, and its sole evidence signal was PM2 (population observation: 'this variant is not present in population databases'), not a de novo or segregation claim.
• No proband-level material of any kind is present in the case: the case is a variant query (gene POLD1, c.3218+4T>C) with no patient record, pedigree, or inheritance annotation anywhere in the reviewed case files (case summary, evidence, literature pass, full-text map, paper extracts).
• Governing framework: POLD1 has no ClinGen CSPEC/VCEP and no local gene-specific framework in this case, so generic ACMG/AMP 2015 (PMID:25741868) is the applicable specification for PM6; no numeric calibration from the supplied threshold block applies to PM6.
• Screening of the three full texts available for this case (PMID:17576681, PMID:9536098, PMID:25394175) found no mention of POLD1 c.3218+4T>C and no family-based observation that could support an assumed de novo event.

Still missing

• No affected proband documented for this variant, so the 'patient with the disease' prerequisite of PM6 is unmet.
• No information bearing on inheritance (no statement of sporadic occurrence, no parental sampling attempt, no record that parents were unavailable).
• No family history describing affection status in relatives, which would be required to infer an apparent de novo occurrence.
• Literature retrieved for this variant (PMID:17576681, PMID:9536098, PMID:25394175) contains no POLD1 case or family data on c.3218+4T>C.

Sources

Reasoning

PP1 requires co-segregation of the variant with disease in multiple affected family members, and its strength is scaled by the number of informative meioses observed (ACMG/AMP 2015, PMID:25741868). Applying PP1 requires, at minimum, a pedigree with genotype data on affected and unaffected relatives, and in an adult-onset, incompletely penetrant, autosomal dominant cancer-predisposition setting such as POLD1-related polymerase proofreading-associated polyposis, it further requires age-appropriate unaffected relatives (ideally older than the typical age of onset) for the segregation to be informative. This case contains no pedigree, no affected relative genotypes, no unaffected relative genotypes, and no statement of family history for the variant; consequently zero meioses can be counted. ClinVar variation 647102 lists only two single-submitter laboratory classifications (both Uncertain significance) with no segregation assertion and no expert-panel review, and none of the three retrieved full texts (PMID:17576681, PMID:9536098, PMID:25394175) reports family data for this variant. PP1 cannot be evaluated and is therefore not assessed rather than not met.

Evidence considered

• ClinVar variation 647102 contains no segregation evidence: both submissions are germline clinical-testing classifications of Uncertain significance by single submitters (SCV001180631, Ambry Genetics; SCV000941306, Labcorp/Invitae), with the only recorded criterion lead being PM2 for SCV000941306.
• No family or pedigree information exists anywhere in the reviewed case material: the case is a variant-level query (POLD1 NM_002691.4:c.3218+4T>C) whose bundle carries variant annotation, population, in-silico, ClinVar and literature sources only; no proband, no relatives, no meioses.
• Gene/disease context matters for informativeness: POLD1 germline variants are associated with an autosomal dominant adult-onset cancer-predisposition phenotype (ClinVar conditions for this record: 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome'), so segregation evidence would have to come from age-matched, ideally older unaffected relatives; none are available.
• Governing framework: no ClinGen CSPEC/VCEP or local specification exists for POLD1 in this case (cspec found = false; no VCEP files or gene_vcep_dir), so generic ACMG/AMP 2015 (PMID:25741868) governs, under which PP1 requires co-segregation in multiple affected family members with strength scaled by informative meioses. No numeric calibration from the supplied threshold block applies to PP1.
• Screening of the three available full texts (PMID:17576681, PMID:9536098, PMID:25394175) found no POLD1 c.3218+4T>C mention and no segregation data.

Still missing

• No pedigree or family history for the variant, so the number of informative meioses is zero and no segregation analysis is possible.
• No genotypes for affected relatives and no genotypes for unaffected relatives (in particular no unaffected relatives older than the typical POLD1/PPAP age of onset, which this late-onset, incompletely penetrant autosomal dominant setting requires for an informative analysis).
• No segregation data reported by either ClinVar submitter (Ambry Genetics SCV001180631; Labcorp/Invitae SCV000941306); both remain Uncertain significance with a single submission each and no expert-panel review.
• No POLD1 family data on this variant in the retrieved literature (PMID:17576681, PMID:9536098, PMID:25394175).

Sources

Reasoning

PP4 requires that the patient's phenotype or family history be highly specific for a disease with a single genetic aetiology - a phenotype-first assertion that can only be judged from the proband's own clinical features (or from the phenotype of published carriers of the same variant). No proband phenotype, HPO terms, age at onset, tumour history, family history, or testing indication for this case appears anywhere in the case evidence; the case is variant-first and carries no clinical annotation. ClinVar records only the broad condition labels 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome' for VCV000647102, which are gene-level disease labels attached by the submitters, not proband-specific phenotypic findings, and neither submission reports a patient phenotype. The gene-level context does confirm POLD1-related polymerase proofreading-associated polyposis (germline exonuclease-domain variants predisposing to colorectal/endometrial cancer), but a gene-level disease association is not a phenotype-specificity observation about this patient and cannot substitute for one. Because the phenotype input required to evaluate PP4 is entirely absent, PP4 is not_assessed rather than met or not_met, and human review is flagged so the case's clinical phenotype can be supplied.

Rule

Generic ACMG/AMP 2015 PP4: patient's phenotype or family history is highly specific for a disease with a single genetic aetiology.

Evidence considered

• ClinVar VCV000647102 lists the exact variant with condition labels 'Colorectal cancer, susceptibility to, 10' and 'Hereditary cancer-predisposing syndrome'; these are gene-level disease labels attached by two clinical laboratories, not proband-specific phenotypic findings.
• Neither ClinVar submission description nor any case file records a patient phenotype, family history, age at onset, or testing indication for this case (variant-first case with no clinical annotation).
• POLD1 gene-level germline disease context (polymerase proofreading-associated polyposis; germline loss of function predisposing to colorectal/endometrial cancer) was reviewed but is a gene-level association only and does not establish that this patient's phenotype is highly specific for a single-gene aetiology.
• No POLD1 VCEP/CSPEC is available, so no framework-specific PP4 guidance exists and generic ACMG/AMP 2015 PP4 semantics apply.

Still missing

• No proband phenotype, HPO terms, tumour spectrum, age at onset, family history, or testing indication is present in the case evidence, so phenotype specificity cannot be evaluated.
• No published case report of NM_002691.4:c.3218+4T>C exists to supply an external carrier phenotype.
• ClinVar supplies only broad submitter-level condition labels ('Colorectal cancer, susceptibility to, 10'; 'Hereditary cancer-predisposing syndrome'), not proband-level phenotypic detail.

Sources

Reasoning

BS4 (lack of segregation with disease) requires genotype data showing that affected family members do not carry the variant while other affected members do, or an equivalent within-family discordance such as homozygosity in an unaffected or heterozygous loss in multiple affected relatives; the ACMG/AMP 2015 framework (PMID:25741868) grounds it in the affected-members-carry the-variant test done in the PP1 direction, so the same family data structure is required in both directions. This case has no pedigree and no relative genotypes at all: nothing establishes that any affected relative was genotyped, and nothing establishes that any affected relative lacked the variant. ClinVar variation 647102 contains only two single-submitter Uncertain significance classifications (Ambry Genetics SCV001180631; Labcorp/Invitae SCV000941306) with no segregation or non-segregation data and no expert-panel review, and the three retrieved full texts (PMID:17576681, PMID:9536098, PMID:25394175) report no POLD1 family data. BS4 is therefore unassessable on the available evidence; critically, the mere absence of segregation data is not non-segregation and must not be scored as support for a benign interpretation.

Evidence considered

• ClinVar variation 647102 reports no family data in either direction: two germline clinical-testing submissions, both Uncertain significance, both single-submitter (SCV001180631, Ambry Genetics; SCV000941306, Labcorp/Invitae), 0 expert-panel submissions, and the only criterion lead recorded is PM2 for SCV000941306.
• No relative genotypes, pedigree, or family-history records exist in the reviewed case material for this variant; the case is a variant-level query (POLD1 NM_002691.4:c.3218+4T>C) with variant annotation, population, in-silico, ClinVar and literature sources only.
• ClinVar conditions for this record ('Colorectal cancer, susceptibility to, 10'; 'Hereditary cancer-predisposing syndrome') place this in an adult-onset autosomal dominant cancer-predisposition context, where informative non-segregation analysis would require genotyped affected relatives (and consideration of reduced penetrance and late onset); no such data are available.
• Governing framework: no ClinGen CSPEC/VCEP or local gene-specific specification exists for POLD1 in this case (cspec found = false; no VCEP files, gene_vcep_dir null, framework source = generic_acmg_fallback), so generic ACMG/AMP 2015 (PMID:25741868) governs BS4. No numeric calibration from the supplied threshold block applies to BS4.
• Screening of the three full texts available for this case (PMID:17576681, PMID:9536098, PMID:25394175) found no mention of POLD1 c.3218+4T>C and no family segregation or non-segregation data.

Still missing

• No genotypes for any affected family members, so a variant-present/variant-absent discordance among affected relatives cannot be shown.
• No pedigree or family history at all, so no affected relative pair or branch comparison is possible.
• No non-segregation assertion in either ClinVar submission (Ambry Genetics SCV001180631; Labcorp/Invitae SCV000941306, both Uncertain significance, single submitters, no expert-panel review).
• No POLD1 family data on this variant in the retrieved literature (PMID:17576681, PMID:9536098, PMID:25394175).
• Absent segregation data must not be converted into BS4 support: uninformative family data is not evidence of non-segregation.

Sources

Reasoning

BP5 is a case-level benign criterion: it requires that the variant be found in a patient who has an alternate molecular basis for the disease phenotype being evaluated (or, under frameworks that define BP5 with a likelihood-ratio statistic, that the LR meet its own governing threshold in the direction written for BP5 - <= if the framework states BP5 strengthens as the value decreases). No BP5 semantics can be applied here because there is no case to which it can refer: the case carries no proband molecular findings, no co-occurring pathogenic variant, no second-hit or alternative-diagnosis data, and no phenotype at all. No numeric BP5 value (likelihood ratio or otherwise) was extracted from any source, and no POLD1-specific specification defines a BP5 threshold, so no operator/direction comparison was performed - none is possible without a value. What the case does show is only population rarity (gnomAD v4.1 total AF 6.3125e-06, grpmax FAF 1.0656e-04, zero homozygotes), which is not an alternate-molecular-basis observation. Because the required case-level input is absent, BP5 is not_assessed rather than not_met, with human review flagged.

Rule

Generic ACMG/AMP 2015 BP5: variant found in a case with an alternate molecular basis for disease; frameworks that define BP5 with a likelihood-ratio statistic must use that criterion's own written operator.

Evidence considered

• The case evidence contains no proband genotype, co-occurring variant, second-hit, or alternate-diagnosis data, and no phenotype, so no alternate molecular basis for disease is documented for BP5.
• ClinVar VCV000647102 reports only two ordinary clinical-laboratory Uncertain significance submissions for this exact variant, with no accompanying case-level molecular or phenotypic detail.
• No POLD1 VCEP/CSPEC exists and generic ACMG/AMP 2015 provides no numeric BP5 threshold; the only numeric observation available (gnomAD v4.1 AF 6.3125e-06, grpmax FAF 1.0656e-04) is a rarity statistic, not a BP5 quantity.

Still missing

• No proband molecular findings (co-occurring pathogenic variant, second hit, or alternative diagnosis) are recorded for this case, so BP5 cannot be evaluated.
• No phenotype is recorded, so there is no disease to which an 'alternate molecular basis' could apply.
• No POLD1-specific VCEP defines a BP5 likelihood-ratio threshold, and no numeric BP5 value exists in any source, so no threshold comparison was possible.

Sources