AXIN2 c.2433G>A, p.(Glu811=)

Reasoning

Scope, as for PP3: NM_004655.4:c.2433G>A (NP_004646.3:p.(Glu811=)) is a synonymous substitution in AXIN2 exon 11, so BP4 is taken from the splice-prediction path (SpliceAI) only, never from the missense path - there is no amino-acid substitution to score and REVEL is recorded as not found, and a missing missense score is not evidence and does not license reaching across to the splice score in reverse. BayesDel is not used because no generic ACMG-strength calibration exists for it and no VCEP supplies an AXIN2 cutoff. No AXIN2 CSPEC/VCEP specification exists, so the generic published SpliceAI calibration applies and no gene-specific lookup pre-assigned a code. SpliceAI Lookup for this exact variant returns DS_AG 0.000, DS_AL 0.000, DS_DG 0.003, DS_DL 0.000 with a maximum delta score of 0.003 (DP_AG 6, DP_AL 483, DP_DG 396, DP_DL 101), with Pangolin concordant (splice gain 0.002, splice loss -0.002). That maximum delta is at or below the recommended SpliceAI no-impact cutoff of <= 0.1, which is the published BP4 (supporting) calibration for the splice path (Jaganathan et al. 2019, Cell; PMID:30661751). The result is consistent with the variant's location: c.2433 sits 27 nt into exon 11, outside the canonical splice region, with no canonical splice consensus involvement (canonical_splice_consensus = false), and all four SpliceAI component deltas are <= 0.003. The single SpliceAI/Pangolin no-impact result is credited once, here, and is not counted again under BP7; that restriction is noted because both criteria would otherwise rest on this identical prediction. BP4 is met at supporting strength.

Rule

Generic ACMG/AMP in-silico splice path: SpliceAI max delta <= 0.1 -> BP4 (supporting) (Jaganathan et al. 2019, Cell; PMID:30661751). Met: max delta 0.003.

Evidence considered

• Consequence type fixed first: NM_004655.4:c.2433G>A / NP_004646.3:p.(Glu811=) is synonymous, so BP4 is assessed on the SpliceAI splice path only; the REVEL missense path does not apply (REVEL not found for this variant) and BayesDel has no generic ACMG-strength calibration and no VCEP-supplied AXIN2 cutoff.
• SpliceAI Lookup for NM_004655.4:c.2433G>A / 17-63526193-C-T: maximum delta score 0.003 (DS_AG 0.000, DS_AL 0.000, DS_DG 0.003, DS_DL 0.000), DP_AG 6, DP_AL 483, DP_DG 396, DP_DL 101; Pangolin SG 0.002, SL -0.002; spliceai_available = true.
• Threshold applied verbatim from the supplied generic ACMG/ClinGen-SVI calibration: SpliceAI max delta <= 0.1 -> BP4 (supporting) (Jaganathan et al. 2019, Cell; PMID:30661751, cited here in prose because it is not a source_registry key for this case). 0.003 satisfies it.
• No governing VCEP/CSPEC specification for AXIN2 (ClinGen CSPEC not found; vcep_materials.json empty; final_classification_framework = generic_acmg_fallback, framework_complete = false), so the generic published splice-predictor calibration was applied and no gene-specific lookup spreadsheet pre-assigned BP4 for this variant.
• The variant is not in a canonical splice site (canonical_splice_consensus = false) and sits 27 nt inside exon 11, consistent with the null splice prediction; the prediction, not the position, is the basis for the criterion.
• The single SpliceAI/Pangolin result is the only benign-direction computational evidence for this variant and is credited once, under BP4; crediting it again under BP7 would double-count one prediction (see BP7 rationale).

Sources

Reasoning

PS3 requires well-established functional studies demonstrating a damaging effect on the gene or gene product (Richards et al. 2015, PMID:25741868). NM_004655.4:c.2433G>A in AXIN2 is a synonymous substitution (NP_004646.3:p.(Glu811=)), and no experimental assay of this specific allele exists anywhere in the case record. The 12 exact-match ClinVar submissions for VCV000182020 (no expert-panel submission) cite no functional study; the only criterion signal extracted from any submitter text is in-silico reasoning (GeneDx submission SCV000209811, PP3-type), and no submission invokes PS3. The papers cited by those submitters are standards and methodology papers (PMID:25741868, PMID:26467025, PMID:28492532) that do not name AXIN2 or this variant: the Karbassi/Quest scoring study's only synonymous-variant content is aggregate (96% of synonymous changes scored in the benign range, 1% in the pathogenic range, mostly predicted splice-affecting) and includes no AXIN2 assay. The AXIN2 germline functional literature retrieved for this case (pvs1_gene_context) assays other alleles only - TOPflash/Wnt-luciferase activation for p.Thr693Met and approximately 50% PBMC transcript reduction for c.1060-3T>C - and does not mention c.2433G>A or p.Glu811=. The SpliceAI max delta score of 0.003 available for this variant is a computational splice prediction, and under the case framework in-silico splice output feeds the PP3/BP4 path rather than PS3; generic ACMG/AMP requires at least one validated functional assay with appropriate controls and case/control and assay-performance caveats before PS3 may be applied. With no such assay, PS3 is not met. No AXIN2 CSPEC or VCEP specification exists (framework_complete=false), so the generic ACMG/AMP 2015 criterion definition governs.

Rule

PS3 (generic ACMG/AMP 2015, PMID:25741868): requires >=1 well-established functional assay showing a damaging effect on gene/gene product; no variant-specific assay exists.

Evidence considered

• Generic ACMG/AMP 2015 definition (PMID:25741868): PS3 applies when well-established functional studies show a damaging effect on the gene or gene product; the framework supplies no functional-assay evidence for any specific variant.
• Framework provenance: no AXIN2 CSPEC/VCEP or local gene framework was retrieved (vcep_materials.json records 'no cspec url available', empty files_found/summaries, framework_complete=false, source=generic_acmg_fallback), so the generic ACMG/AMP 2015 rule set governs PS3.
• ClinVar VCV000182020 (exact match, NM_004655.4(AXIN2):c.2433G>A p.Glu811=) has 12 submissions with no expert-panel submission; per-submission audit shows no functional-study evidence: criteria_found and evidence_signals are empty for 11 of 12 submissions, and the single populated signal is in-silico (GeneDx SCV000209811, PP3-type text). No submission cites PS3.
• Papers cited by the ClinVar submitters are standards/methodology works with no variant content: PMID:25741868 (ACMG/AMP 2015 guidelines), PMID:26467025 (laboratory-wide 7-point scoring system), PMID:28492532 (Sherloc refinement). PMID:26467025's only relevant content is aggregate synonymous-variant behaviour (96% of synonymous changes scored benign-range; 1% pathogenic-range, majority predicted splice-affecting) and it assays no AXIN2 variant.
• AXIN2 germline functional literature retrieved for this case (source file pvs1_gene_context; supporting papers PMID:21476993, PMID:23838596, PMID:31285513, PMID:34817745, PMID:41361419) assays other alleles: PMID:41361419 reports TOPflash/Wnt-luciferase activation for p.Thr693Met and approximately 50% AXIN2 transcript reduction by RT-qPCR in carrier PBMCs for c.1060-3T>C. A text search of these retrieved full texts for 'c.2433G>A', 'Glu811' and 'E811' returned no hits, so none provides PS3 evidence for this variant.
• In-silico splice output recorded for the variant: SpliceAI max delta score 0.003 (DS_AG 0.0, DS_AL 0.0, DS_DG 0.003, DS_DL 0.0; pangolin SG 0.002 / SL -0.002). This is a computational prediction, not an experimental functional result, and belongs to the PP3/BP4 in-silico path, so it cannot support PS3.
• OncoKB returned no variant-specific curated evidence for AXIN2 p.Glu811= (is_variant_specific=false; oncogenicity label 'Unknown Oncogenic Effect', no PMIDs), and COSMIC and cancer hotspots returned no records; none supplies a functional assay for this allele.

Still missing

• No variant-specific functional assay has been performed or reported for NM_004655.4:c.2433G>A - no minigene splicing assay, patient-tissue RT-PCR/cDNA sequencing, or AXIN2/Wnt reporter (e.g. TOPflash) experiment. Such an assay with validated positive/negative controls would be required to invoke PS3.
• AXIN2 has no ClinGen CSPEC/VCEP specification, so no gene-specific PS3 assay requirements, accepted assay list, or strength-calibration rules were available; the generic ACMG/AMP 2015 definition was used.

Sources

Reasoning

PS4 (Richards et al. 2015, PMID:25741868) requires that the prevalence of the variant in affected individuals be significantly increased compared with the prevalence in controls, i.e. a case-control or affected-cohort enrichment signal for this specific variant. No governing VCEP/CSPEC specification exists for AXIN2 (final_classification_framework.found = false; source = generic_acmg_fallback), so the generic ACMG/AMP 2015 definition applies. The generic definition carries no numeric threshold (no odds-ratio or likelihood-ratio cutoff is supplied by the governing source), so no numeric comparison operator could be applied or mis-directed here: unlike a VCEP case-control rule with a stated operator, there is no number to substitute into an inequality for this criterion. Substantively, no enrichment data exist. NM_004655.4:c.2433G>A (NP_004646.3:p.(Glu811=)) is not mentioned in any paper retrieved for this case: exact-string searches for c.2433 and E811/Glu811 returned 0 hits across the five AXIN2 germline papers (PMID:21476993, PMID:23838596, PMID:31285513, PMID:34817745, PMID:41361419) and across the three ClinVar-cited papers (PMID:25394175, PMID:25741868, PMID:26467025). ClinVar variation 182020 contains 12 submissions (11 per the API), all ordinary clinical laboratories, and none contributes case-control counts; the aggregated phenotype-prevalence signal is a Benign/Likely benign assertion, not an enrichment statistic. PMID:26467025 provides only variant-type-level statistics (1% of synonymous changes scored in the pathogenic range versus 96% in the benign range) and does not report this variant. Without any affected-versus-control comparison carrying this variant, PS4's enrichment requirement is not satisfied.

Rule

PS4 (ACMG/AMP 2015, PMID:25741868): the prevalence of the variant in affected individuals is significantly increased compared with the prevalence in controls. Assessed under the generic ACMG/AMP fallback because no governing AXIN2 VCEP/CSPEC specification was retrieved; no numeric threshold or comparison operator is defined by that framework, and no variant-level case-control data exist.

Evidence considered

• Framework check: no VCEP/CSPEC specification retrieved for AXIN2 (final_classification_framework: found=false, source=generic_acmg_fallback), so the generic ACMG/AMP 2015 PS4 definition governs; that definition states a requirement for significantly increased prevalence in affected individuals versus controls but supplies no numeric threshold or comparison operator, so no inequality could be written.
• Variant-specific literature check: exact-string searches for 'c.2433' and for 'E811'/'Glu811' returned zero hits in all five retrieved AXIN2 germline articles and all three ClinVar-cited articles, so no affected-case cohort in the retrieved literature reports this variant.
• ClinVar record 182020 (12 audited submissions; API reports 11 submitters, 'criteria provided, multiple submitters, no conflicts', 2 stars): all submissions are from ordinary clinical laboratories and none provides case-control counts or an enrichment statistic.
• PMID:26467025 (16,500 assessments / 11,894 variants / 338 genes) reports only aggregate variant-type distributions - 1% of synonymous changes in the pathogenic score range vs 96% in the benign range - and does not mention AXIN2 or this variant, so it cannot supply variant-specific enrichment.
• Population context only (not PS4 evidence): gnomAD v4.1 AF 0.000123279 (199/1,614,228 alleles, 0 homozygotes) and gnomAD v2.1 AF 9.90239e-05 (28/282,760 alleles, 0 homozygotes); no case-enrichment inference is drawn from these unaffected-population data.
• No numeric PS4 threshold was available from the governing framework or from a named calibration publication supplied for this case, so PS4 was adjudicated on presence versus absence of variant-level case-control enrichment rather than against any recalled number.

Still missing

• No case-control or affected-cohort study reporting the exact variant NM_004655.4:c.2433G>A was available; a curated affected/control comparison from a genotype-phenotype cohort would be required to satisfy PS4.
• No governing VCEP/CSPEC specification for AXIN2 was retrieved, so if a VCEP later publishes a numeric PS4 enrichment threshold (with a stated comparison operator), PS4 would need re-evaluation against that threshold.

Sources

Reasoning

PM2 at supporting strength is applied when the variant is absent from, or extremely rare in, a large general population cohort; the SVI-recommended supporting threshold is an allele frequency <= 0.0001, and the ClinGen practice is to judge rarity on the highest ancestry-group filtering (popmax/grpmax) frequency rather than the pooled total alone. gnomAD v4.1 is the larger and more current dataset: combined total AF 1.23279e-04 (199/1,614,228 alleles, homozygotes 0), exome AF 1.23129e-04, genome AF 1.24713e-04, European (non-Finnish) AF 1.53384e-04 (181/1,180,046), and joint grpmax FAF 1.3499e-04 - every one of these exceeds 0.0001. gnomAD v2.1 similarly gives grpmax FAF 1.4552e-04 and European (non-Finnish) AF 1.93699e-04, both above 0.0001, and its combined total AF is 9.90239e-05, the single measure that falls marginally below 0.0001. Applying the applicable exome-only/non-cancer subsets instead of the combined totals does not change the conclusion: gnomAD v2.1 non-cancer exomes AF 1.09785e-04 (26/236,826) and gnomAD v3.1 non-cancer genomes AF 1.14885e-04 (17/147,974) are both above 0.0001, with grpmax FAFs of 1.5286e-04 and 1.2398e-04 respectively. Because the variant is present at a consistent ~1.2e-04 frequency in over 1.6 million alleles and in the non-cancer subsets, with a grpmax FAF above the threshold in both gnomAD versions, PM2 is not met. Under the applicable generic default this criterion is also not met at any strength.

Rule

PM2: absent or extremely rare (<= 0.0001) in a large general population cohort; applied at supporting strength per ClinGen SVI 2020 recommendation.

Evidence considered

• gnomAD v4.1 (chr17-65530075-C-T): total AF 1.23279e-04 (199/1,614,228 alleles, homozygotes 0); exome AF 1.23129e-04 (180/1,461,878); genome AF 1.24713e-04 (19/152,350); European (non-Finnish) AF 1.53384e-04 (181/1,180,046, homozygotes 0); joint grpmax filtering AF 1.3499e-04.
• gnomAD v2.1 (17-63526193-C-T): total AF 9.90239e-05 (28/282,760 alleles, homozygotes 0); European (non-Finnish) AF 1.93699e-04 (25/129,066); grpmax filtering AF 1.4552e-04.
• Exome-only/non-cancer subsets confirm rather than remove the frequency: gnomAD v2.1 non-cancer exomes AF 1.09785e-04 (26/236,826 alleles, homozygotes 0, grpmax FAF 1.5286e-04) and gnomAD v3.1 non-cancer genomes AF 1.14885e-04 (17/147,974 alleles, homozygotes 0, grpmax FAF 1.2398e-04).
• PM2 was evaluated at the SVI-recommended supporting strength with a threshold of <= 0.0001 (ClinGen SVI 2020 downgrade of PM2 to supporting; Richards et al. 2015, PMID:25741868). The ClinGen-consistent measure is the highest ancestry-group filtering allele frequency (grpmax FAF), which is 1.3499e-04 in gnomAD v4.1 and 1.4552e-04 in gnomAD v2.1, both above 0.0001; only the pooled gnomAD v2.1 total (9.90239e-05) is marginally below.
• gnomAD-Canada v1.0 was queried for 17-65530075-C-T but returned HTTP 500 Internal Server Error; its recorded 'absent' status is a retrieval failure and was not treated as a confirmed absence.

Still missing

• No ClinGen CSPEC or AXIN2 VCEP specification exists, so no gene- or prevalence-derived PM2 threshold is available; the generic 1e-4 supporting threshold was used.
• gnomAD-Canada v1.0 could not be queried (HTTP 500 server error) and returned no frequency data, so absence from that cohort was not independently confirmed; this does not affect the conclusion because the gnomAD v2.1/v4.1 frequencies already exceed the threshold.

Sources

Reasoning

Scope was settled before any predictor score was read. NM_004655.4:c.2433G>A normalises to NP_004646.3:p.(Glu811=) (prefetch.json Mutalyzer/VariantValidator; ClinVar variation 182020 name 'NM_004655.4(AXIN2):c.2433G>A (p.Glu811=)'), i.e. a synonymous substitution in AXIN2 exon 11 (c.2406-*1413; NC_000017.10:g.63526193C>T), not a missense change and not a frameshift, nonsense or other protein-length-changing variant. PP3 for this variant therefore has to come from the splice-prediction path (SpliceAI) only; the REVEL missense protein-impact path does not apply, because a synonymous change has no amino-acid substitution to score (REVEL is recorded as not found for this variant, and a missing or gray-zone missense score never licenses substituting the splice score, nor the reverse). BayesDel is likewise unusable: no generic ACMG-strength calibration exists for it and no VCEP supplies an AXIN2 cutoff (score also not found in this case). No ClinGen CSPEC and no AXIN2 VCEP/gene-specific specification exists (cspec found=false; final classification framework resolved to the generic ACMG/AMP 2015 fallback, framework_complete=false; vcep_materials.json carries no rules, domain tables or index guidance), so the generic published calibration governs. SpliceAI Lookup for this exact variant is available (spliceai_available=true) and reports DS_AG 0.000, DS_AL 0.000, DS_DG 0.003, DS_DL 0.000, maximum delta score 0.003 (DP_AG 6, DP_AL 483, DP_DG 396, DP_DL 101), with Pangolin concordant (splice gain 0.002, splice loss -0.002). A maximum delta of 0.003 is far below both published SpliceAI PP3 bands - supporting >= 0.2 and moderate >= 0.5 (Jaganathan et al. 2019, Cell; PMID:30661751) - so no computational evidence supports a deleterious splice effect. The null prediction is internally consistent with the variant's position: c.2433 lies 27 nt into exon 11, away from the intron 10 acceptor and outside the canonical splice region (canonical_splice_consensus = false). PP3 is not met.

Rule

Generic ACMG/AMP in-silico splice path: SpliceAI max delta >= 0.2 -> PP3 (supporting); >= 0.5 -> PP3 (moderate) (Jaganathan et al. 2019, Cell; PMID:30661751). Not met: max delta 0.003.

Evidence considered

• Variant identity and consequence type were fixed first: NM_004655.4:c.2433G>A = NP_004646.3:p.(Glu811=) (prefetch.json Mutalyzer/VariantValidator normalisation; ClinVar variation 182020 name 'NM_004655.4(AXIN2):c.2433G>A (p.Glu811=)'; pvs1_variant_assessment consequence_class 'synonymous'), so the applicable PP3 path is SpliceAI (splice impact), not REVEL (missense protein impact).
• SpliceAI Lookup (spliceailookup.broadinstitute.org, hg37, basic, distance 500) for NM_004655.4:c.2433G>A / 17-63526193-C-T: DS_AG 0.000, DS_AL 0.000, DS_DG 0.003, DS_DL 0.000; maximum delta score 0.003; DP_AG 6, DP_AL 483, DP_DG 396, DP_DL 101; Pangolin SG 0.002, SL -0.002; spliceai_available = true; variant_consequence 'sequence_variant'.
• Thresholds applied verbatim from the supplied generic ACMG/ClinGen-SVI calibration: SpliceAI max delta >= 0.5 -> PP3 (moderate); >= 0.2 -> PP3 (supporting) (Jaganathan et al. 2019, Cell; PMID:30661751, which is not a source_registry key for this case and is therefore cited in this prose entry rather than in evidence_refs). 0.003 reaches neither band.
• No governing VCEP/CSPEC specification exists for AXIN2 (ClinGen CSPEC not found; review_manifest gene_vcep_files = []; vcep_materials.json empty; final_classification_framework source = generic_acmg_fallback, framework_complete = false), so no gene-specific PP3 lookup code is available and the generic published calibration was applied.
• REVEL is not applicable to this variant (synonymous; no amino-acid substitution to score) and is recorded as not found; BayesDel is likewise unused because no generic ACMG-strength calibration exists for it and no VCEP supplies an AXIN2 cutoff. No other splice-relevant predictor output (VarSeak/CADD-type) exists in this case.
• Position corroborates the null prediction: c.2433 lies in exon 11 (c.2406-*1413 of NM_004655.4, i.e. 27 nt downstream of the intron 10 acceptor) and is not at a canonical splice consensus position (canonical_splice_consensus = false).

Sources

Reasoning

PP5 (Richards et al. 2015, PMID:25741868) requires a reputable source reporting the variant as pathogenic, with the evidence not independently available to the laboratory. Per the applied rule for this case, PP5 may be triggered only by an exact-variant ClinVar expert-panel (VCEP) classification of Pathogenic or Likely pathogenic; ordinary laboratory submissions, aggregated ClinVar labels, and non-expert assertions must never trigger it. Here the condition is not met on the expert-panel gate: ClinVar variation 182020 has 12 audited submissions and 0 expert-panel submissions (submission_audit counts: expert_panel_submissions = 0; the ClinVar API 'expert_panels' object is empty), all submitters are ordinary clinical laboratories (GeneDx, Ambry Genetics, Invitae/Labcorp, Myriad Genetics, Quest Diagnostics, CeGaT, PreventionGenetics, Sema4, Erasmus MC, Academic Medical Center, Rigshospitalet) with 'is_expert_panel' false on every audited submission. Every submitted classification is Benign or Likely benign; not a single submission asserts Pathogenic or Likely pathogenic, so there is no pathogenic reputable-source assertion of any tier on which PP5 could rest. No AXIN2 VCEP/CSPEC specification was retrieved (final_classification_framework: found = false, official_clingen = false), so no panel label exists to consult. The triaged candidate paper for PP5, PMID:25394175, was reviewed and contains no AXIN2 content, no mention of c.2433G>A, and no classifier assertion about this variant. PP5 involves no numeric threshold, so no inequality was constructed; the criterion fails on the categorical expert-panel requirement. Because the failure is categorical and the expert-panel count is exactly zero across all 12 submissions, confidence is high.

Rule

PP5 (ACMG/AMP 2015, PMID:25741868): reputable source reports the variant as pathogenic, with evidence unavailable to the laboratory; applied here only as an exact-variant ClinVar expert-panel (VCEP) Pathogenic/Likely pathogenic classification. Ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions do not trigger PP5.

Evidence considered

• PP5 definition consulted in the generic ACMG/AMP 2015 framework (PMID:25741868): a reputable source reporting the variant as pathogenic where the laboratory cannot access the underlying evidence; per the criterion rule applied here, the reputable source must be an exact-variant ClinVar expert-panel (VCEP) classification of Pathogenic/Likely pathogenic. PP5 carries no numeric threshold, so no comparison operator applies.
• ClinVar submission audit for variation 182020: 12 submissions, expert_panel_submissions = 0; the ClinVar API expert_panels field is empty and is_expert_panel = false for every audited submission (GeneDx, Ambry Genetics, Labcorp/Invitae, Myriad Genetics, Quest Diagnostics, CeGaT Tuebingen, PreventionGenetics, Sema4, Erasmus MC, Academic Medical Center, Rigshospitalet, and one additional submitter).
• Classification direction check: all submitted classifications are Benign (n = 2 clinical laboratories) or Likely benign (n = 9 clinical laboratories); no Pathogenic or Likely pathogenic submission exists, expert or otherwise.
• Framework check: no AXIN2 CSPEC/VCEP specification was retrieved (final_classification_framework.found = false, source = generic_acmg_fallback), so there is no expert-panel label for this variant to consult.
• PMID:25394175 was reviewed as the triaged PP5 candidate paper: it contains no AXIN2 content, does not mention c.2433G>A or p.(Glu811=), and makes no assertion about this variant.
• The aggregated ClinVar record label (Benign/Likely benign, 2-star, 'criteria provided, multiple submitters, no conflicts') was not treated as a reputable-source assertion for PP5, since aggregate and non-expert labels must not trigger this criterion.

Sources

Reasoning

BA1 (stand-alone benign) requires an allele frequency >= 0.05 (ACMG/AMP 2015 stand-alone definition). No dataset or ancestry group reaches this. gnomAD v2.1 (17-63526193-C-T): combined total AF 9.90239e-05 (28/282,760 alleles, homozygotes 0); exome AF 1.0741e-04 (27/251,364); genome AF 3.185e-05 (1/31,396); highest observed ancestry frequency is European (non-Finnish) AF 1.93699e-04 (25/129,066 alleles, homozygotes 0); grpmax FAF 1.4552e-04. gnomAD v4.1 (chr17-65530075-C-T): total AF 1.23279e-04 (199/1,614,228 alleles, homozygotes 0), exome 1.23129e-04, genome 1.24713e-04, European (non-Finnish) AF 1.53384e-04 (181/1,180,046), joint grpmax FAF 1.3499e-04. The maximum value anywhere (1.94e-04) is roughly 260-fold below the 0.05 threshold, so the variant is nowhere near the stand-alone benign frequency required. AXIN2 has no ClinGen specification, so the generic 0.05 stand-alone threshold applies rather than any gene- or disease-specific derivation.

Rule

BA1: allele frequency >= 0.05 in a general population database, stand-alone benign evidence (ACMG/AMP 2015).

Evidence considered

• gnomAD v2.1 all-comers totals for 17-63526193-C-T: AC 28/AN 282,760, total AF 9.90239e-05, homozygotes 0; exome-only AF 1.0741e-04 (27/251,364) and genome-only AF 3.185e-05 (1/31,396).
• gnomAD v2.1 highest ancestry group: European (non-Finnish) AF 1.93699e-04 (25/129,066 alleles, homozygotes 0); grpmax filtering allele frequency 1.4552e-04.
• gnomAD v4.1 all-comers totals for chr17-65530075-C-T: AC 199/AN 1,614,228, total AF 1.23279e-04, homozygotes 0; European (non-Finnish) AF 1.53384e-04 (181/1,180,046); joint grpmax FAF 1.3499e-04.
• Maximum allele frequency observed for this variant across both gnomAD versions and all ancestry groups is 1.94e-04, about 260-fold below the 0.05 ACMG/AMP 2015 stand-alone benign threshold (Richards et al. 2015, PMID:25741868), so BA1 is not met.
• No ClinGen CSPEC or VCEP specification was found for AXIN2, so the generic ACMG/AMP stand-alone threshold was applied rather than a gene-specific frequency derivation.

Sources

Reasoning

BS1 (strong benign on allele frequency) is applied at a generic default of >= 0.01 because no AXIN2-specific ClinGen specification exists to supply a per-gene threshold (per-gene derivation would be preferred where available). The variant is present at a combined gnomAD v2.1 allele frequency of 9.90239e-05 (28/282,760 alleles) and a combined gnomAD v4.1 allele frequency of 1.23279e-04 (199/1,614,228 alleles); the highest ancestry-group frequency is European (non-Finnish) 1.93699e-04 in v2.1 and 1.53384e-04 in v4.1; grpmax filtering allele frequencies are 1.4552e-04 (v2.1) and 1.3499e-04 (v4.1). The most frequent observation anywhere (1.94e-04, roughly 1 in 5,165 alleles) remains about 50-fold below the 1% (0.01) threshold, so BS1 is not met. The variant is clearly rare in the general population, but not at a frequency that qualifies as strong benign evidence on frequency alone.

Rule

BS1: allele frequency >= 0.01 (generic SVI default), strong benign evidence; per-gene derivation preferred where a specification exists.

Evidence considered

• gnomAD v2.1 combined allele frequency 9.90239e-05 (28/282,760 alleles, homozygotes 0); European (non-Finnish) 1.93699e-04 (25/129,066); grpmax FAF 1.4552e-04.
• gnomAD v4.1 combined allele frequency 1.23279e-04 (199/1,614,228 alleles, homozygotes 0); European (non-Finnish) 1.53384e-04 (181/1,180,046); joint grpmax FAF 1.3499e-04.
• All observed values are at least ~50-fold below the generic BS1 threshold of 0.01 (1%); the generic SVI default threshold derives from Whiffin et al. 2017 (PMID:28518168), and ACMG/AMP 2015 defines BS1 as a frequency greater than expected for the disorder (Richards et al. 2015, PMID:25741868).
• No AXIN2 gene-specific or disease-prevalence-derived BS1 threshold is available in this case's evidence, so the generic default was applied as the more conservative standard.

Still missing

• No gene-specific BS1 threshold is available: no ClinGen CSPEC or AXIN2 VCEP specification was found, so the generic 1% default was used rather than a prevalence- or penetrance-derived AXIN2 threshold.

Sources

Reasoning

BS2 requires observation of the variant in a healthy adult individual for a disorder in which full penetrance is expected at an early age (recessive homozygous, dominant heterozygous, or X-linked hemizygous). Homozygote counts are 0 in every gnomAD dataset examined: v2.1 combined total_hom 0 (exome 0, genome 0) and v4.1 combined total_hom 0 (exome 0, genome 0); the non-cancer subsets likewise report 0 homozygotes. The variant is therefore seen only as heterozygotes in adult population reference cohorts. AXIN2-associated disease (oligodontia-colorectal cancer syndrome / adenomatous polyposis and colorectal cancer predisposition) is an autosomal dominant, adult-onset, incompletely penetrant phenotype, so heterozygous occurrence in adult gnomAD participants - a cohort not phenotyped for dental agenesis or colonic polyposis - does not establish health in a fully penetrant early-onset disorder. No homozygote or healthy-adult-with-full-penetrance observation exists, so BS2 is not met. Given the adult and typically mild phenotype, gnomAD participants cannot be treated as screened healthy adults for this gene.

Rule

BS2: 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).

Evidence considered

• gnomAD v2.1: total_hom 0 (exome_hom 0, genome_hom 0); every ancestry group also reports 0 homozygotes.
• gnomAD v4.1: total_hom 0 (exome_hom 0, genome_hom 0); every ancestry group also reports 0 homozygotes.
• gnomAD v2.1 non-cancer exomes report homozygotes 0 and gnomAD v3.1 non-cancer genomes report homozygotes 0, so the zero-homozygote observation is not an artifact of including cancer cohorts.
• BS2 as defined requires a fully penetrant, early-onset disorder and a well-documented healthy adult carrier (Richards et al. 2015, PMID:25741868); AXIN2-associated phenotypes are autosomal dominant with reduced, age-dependent penetrance and adult onset, so heterozygous adult carriers in a general population cohort are expected and do not meet BS2.
• gnomAD-Canada v1.0 returned an HTTP 500 error (not a genuine absence), so it provided no independent homozygote or control-cohort information for this criterion.

Still missing

• No AXIN2-specific penetrance estimate or phenotype-screened control cohort is available in this case's evidence; the assessment rests on gnomAD homozygote counts being zero and on the autosomal dominant, incompletely penetrant, adult-onset nature of AXIN2-associated disease.
• gnomAD participants are not phenotyped for oligodontia or colonic polyposis, so their status as unaffected adults cannot be verified in the sense BS2 requires.

Sources

Reasoning

BS3 requires well-established functional studies demonstrating no damaging effect on the gene or gene product (Richards et al. 2015, PMID:25741868). For NM_004655.4:c.2433G>A (NP_004646.3:p.(Glu811=)) no functional study of any kind - minigene splicing assay, patient-tissue RT-PCR/cDNA analysis, or AXIN2/Wnt reporter assay - has been reported, so there is no experimental demonstration of normal function or normal splicing to convert this criterion. The evidence commonly mistaken for BS3 here is not assay evidence: the ClinVar consensus of Benign/Likely benign across 11 submitters for VCV000182020 (2 stars, no expert panel) carries no functional data and no criterion statement; no submission cites BS3, and the criterion signal extracted from submitter text is in-silico only (GeneDx SCV000209811). Likewise the SpliceAI max delta of 0.003 - which indicates no predicted splice impact - is a computational prediction and, for a synonymous variant, belongs to the BP4 in-silico path, not to BS3; per the case framework the in-silico splice score may not be re-labelled as experimental functional evidence. The standards and scoring papers cited by submitters (PMID:25741868, PMID:26467025, PMID:28492532) report no assay of this variant, and the AXIN2 functional literature retrieved for this case assays other alleles (p.Thr693Met Wnt-reporter activation; c.1060-3T>C transcript reduction). BS3 is therefore not met: benign functional behaviour has not been experimentally demonstrated, only assumed. Generic ACMG/AMP 2015 governs, as no AXIN2 CSPEC or VCEP specification exists (framework_complete=false).

Rule

BS3 (generic ACMG/AMP 2015, PMID:25741868): requires >=1 well-established functional assay showing no damaging effect on gene/gene product; no variant-specific assay exists.

Evidence considered

• Generic ACMG/AMP 2015 definition (PMID:25741868): BS3 requires well-established functional studies demonstrating no damaging effect on the gene or gene product, with assay controls, reproducibility and case/control rigour considered; the framework supplies no variant-specific assay data.
• Framework provenance: no AXIN2 CSPEC/VCEP or local gene framework was retrieved (vcep_materials.json: 'no cspec url available', empty files_found/summaries, framework_complete=false, source=generic_acmg_fallback), so the generic ACMG/AMP 2015 rule set governs BS3.
• ClinVar VCV000182020 exact-match record includes 12 submissions (11 submitters), 8 usable for classification, 0 expert-panel submissions, overall Benign/Likely benign. Per-submission audit shows no submission cites BS3 and no submission reports a functional assay; the only extracted criterion signal is in-silico (GeneDx SCV000209811). The classifications therefore rest on criteria belonging to other evidence groups and cannot be used as functional evidence here.
• Papers cited by the ClinVar submitters contain no assay of this variant: PMID:25741868 (ACMG/AMP 2015 guidelines), PMID:26467025 (laboratory-wide 7-point scoring system; aggregate synonymous-variant statistics only), PMID:28492532 (Sherloc refinement, full text not retrieved in this case).
• AXIN2 germline functional literature retrieved for this case (source file pvs1_gene_context; supporting papers PMID:21476993, PMID:23838596, PMID:31285513, PMID:34817745, PMID:41361419) covers other alleles only - TOPflash/Wnt-luciferase activation for p.Thr693Met and approximately 50% transcript reduction for c.1060-3T>C in PMID:41361419. A text search of these retrieved texts for 'c.2433G>A', 'Glu811' and 'E811' returned no hits, so no benign functional result exists for this variant.
• In-silico splice output for the variant: SpliceAI max delta score 0.003, i.e. no predicted splice impact. This is a computational prediction used on the PP3/BP4 path (non-missense variants use SpliceAI) and is not experimental evidence of preserved function, so it cannot support BS3.
• OncoKB recorded no variant-specific curated content for AXIN2 p.Glu811= (is_variant_specific=false, 'Unknown Oncogenic Effect', no PMIDs); COSMIC and cancer hotspots returned no records. Neither provides functional assay data for BS3.

Still missing

• No functional study demonstrating normal function or normal splicing has been reported for NM_004655.4:c.2433G>A - e.g. a minigene assay or patient-tissue RT-PCR showing wild-type splicing, or an AXIN2/Wnt reporter showing normal regulatory activity. BS3 cannot be converted without such an assay with validated controls.
• AXIN2 has no ClinGen CSPEC/VCEP specification, so no gene-specific BS3 assay requirements or accepted-assay list were available; the generic ACMG/AMP 2015 definition was used.

Sources

Reasoning

BP6 (Richards et al. 2015, PMID:25741868) requires a reputable source reporting the variant as benign, with the evidence unavailable to the laboratory for independent assessment. Per the applied rule for this case, BP6 may be triggered only by an exact-variant ClinVar expert-panel (VCEP) classification of Benign or Likely benign; ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions must never trigger it. The expert-panel gate is not met: ClinVar variation 182020 has 12 audited submissions and 0 expert-panel submissions (submission_audit counts: expert_panel_submissions = 0; the ClinVar API 'expert_panels' object is empty), and every audited submission carries is_expert_panel = false. Although the submissions are uniformly benign-direction (Likely benign in 9 clinical laboratories, Benign in 2 clinical laboratories) and the aggregate API label is 'Benign/Likely benign' with 'criteria provided, multiple submitters, no conflicts' (2 stars), those are ordinary clinical-laboratory assertions and an aggregate record label, both of which are excluded by BP6's expert-panel requirement here. No AXIN2 VCEP/CSPEC specification was retrieved (final_classification_framework: found = false, official_clingen = false), so no panel classification of this variant exists. The triaged candidate paper for this criterion family, PMID:25394175, was reviewed and contains no AXIN2 or variant-specific content. BP6 involves no numeric threshold, so no inequality was constructed; the criterion fails on the categorical expert-panel requirement, and because the expert-panel count is exactly zero across all 12 submissions the confidence is high. Note that the benign-direction signal itself is captured elsewhere in this interpretation (population-frequency and in-silico criteria); this entry records only that the expert-panel condition for BP6 is unmet.

Rule

BP6 (ACMG/AMP 2015, PMID:25741868): reputable source reports the variant as benign, with evidence unavailable to the laboratory; applied here only as an exact-variant ClinVar expert-panel (VCEP) Benign/Likely benign classification. Ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions do not trigger BP6.

Evidence considered

• BP6 definition consulted in the generic ACMG/AMP 2015 framework (PMID:25741868): a reputable source reporting the variant as benign where evidence is unavailable to the laboratory; per the criterion rule applied here the reputable source must be an exact-variant ClinVar expert-panel (VCEP) classification of Benign/Likely benign. BP6 carries no numeric threshold, so no comparison operator applies.
• ClinVar submission audit for variation 182020: 12 submissions, expert_panel_submissions = 0; the ClinVar API expert_panels field is empty and is_expert_panel = false for every audited submission, all of which are ordinary clinical laboratories (GeneDx, Ambry Genetics, Labcorp/Invitae, Myriad Genetics, Quest Diagnostics, CeGaT Tuebingen, PreventionGenetics, Sema4, Erasmus MC, Academic Medical Center, Rigshospitalet, and one additional submitter).
• Classification direction check: the benign-direction signal is present (Benign in 2 clinical laboratories, Likely benign in 9; aggregate API label 'Benign/Likely benign', review status 'criteria provided, multiple submitters, no conflicts', 2 stars) but it originates from non-expert submissions and an aggregate record label, which this rule excludes from triggering BP6.
• Framework check: no AXIN2 CSPEC/VCEP specification was retrieved (final_classification_framework.found = false, source = generic_acmg_fallback, official_clingen = false), so no expert-panel classification of this exact variant exists to support BP6.
• PMID:25394175 was reviewed as a triaged candidate for this criterion family and contains no AXIN2 content, no mention of c.2433G>A, and no benign-source assertion for this variant.

Sources

Reasoning

Scope: BP7 applies only to a synonymous variant, and NM_004655.4:c.2433G>A (NP_004646.3:p.(Glu811=)) is synonymous, so the criterion is in scope and was evaluated rather than dismissed. BP7 as defined in the ACMG/AMP 2015 framework (Richards et al., PMID:25741868) requires a synonymous (silent) variant for which splicing prediction algorithms predict no impact to the splice consensus sequence nor the creation of a new splice site, and for which the nucleotide is not highly conserved. The first element is supported by the only splice prediction in the case: SpliceAI maximum delta score 0.003 (DS_AG 0.000, DS_AL 0.000, DS_DG 0.003, DS_DL 0.000; Pangolin splice gain 0.002, splice loss -0.002), i.e. no predicted splice impact, consistent with c.2433 lying 27 nt into exon 11 away from the canonical splice region (canonical_splice_consensus = false). That identical prediction, however, is the evidence already credited to BP4 at supporting strength for this variant (SpliceAI max delta <= 0.1). The ACMG/AMP framework treats in-silico predictions as a single piece of evidence rather than independent pieces of evidence, and this group's scope forbids double-counting one prediction across PP3/BP4/BP7; crediting the same SpliceAI/Pangolin result under both BP4 and BP7 would count one prediction twice and inflate the benign evidence for this variant. The second element cannot be established either: no nucleotide-conservation metric for c.2433 (phyloP, GERP, phastCons or equivalent) is present anywhere in the case evidence, so the 'nucleotide is not highly conserved' requirement of BP7 is unverified. BP7 is therefore not counted as an independent benign criterion here; the prediction that would support it has already been applied once, under BP4, and no gene-specific (VCEP/CSPEC) rule exists for AXIN2 to resolve which of the two synonymous-variant codes is preferred.

Rule

BP7 (Richards et al. 2015, PMID:25741868): synonymous variant with no predicted impact on the splice consensus sequence or creation of a new splice site, and the nucleotide not highly conserved. Not credited here because the only supporting prediction is already counted under BP4 and the conservation element is unverified.

Evidence considered

• Consequence type confirmed as synonymous before scoring: NM_004655.4:c.2433G>A = NP_004646.3:p.(Glu811=) (prefetch.json Mutalyzer/VariantValidator normalisation; ClinVar variation 182020), so BP7 is the criterion scoped to this variant type.
• BP7 definition taken from the ACMG/AMP 2015 standards (Richards et al., PMID:25741868): a synonymous (silent) variant with no predicted impact on the splice consensus sequence or creation of a new splice site, and a nucleotide that is not highly conserved.
• SpliceAI/Pangolin result for this exact variant (max delta 0.003; DS_AG 0.000, DS_AL 0.000, DS_DG 0.003, DS_DL 0.000; Pangolin SG 0.002, SL -0.002; DP_AG 6, DP_AL 483, DP_DG 396, DP_DL 101) predicts no splice impact, but it is the same single prediction already used for BP4 at supporting strength; in-silico predictions are one piece of evidence and this group prohibits counting one prediction across PP3/BP4/BP7.
• Richards et al. (PMID:25741868) warn that computational/in-silico predictions should be treated as a single piece of evidence rather than independent evidence and should not be the sole basis of a clinical assertion - the basis for declining to credit this prediction twice.
• No conservation data (phyloP/GERP/phastCons or equivalent) for c.2433 appears in any case file, so the second element of BP7 ('not highly conserved') could not be checked.
• No CSPEC/VCEP specification for AXIN2 exists (cspec found=false; framework generic_acmg_fallback, framework_complete=false) to specify gene-level BP7 handling, and no gene-specific lookup table pre-assigned a BP7 code for c.2433.
• Background context reviewed: Karbassi et al. (PMID:26467025) report that 96% of synonymous changes scored in the benign range and that most of the rare pathogenic non-amino-acid-altering variants were predicted to affect splicing; neither that paper nor the ACMG/NSGC referral guideline (PMID:25394175) supplies a splice prediction or conservation metric for AXIN2 c.2433.

Still missing

• No nucleotide-conservation metric for AXIN2 c.2433 (phyloP, GERP, phastCons or equivalent) appears in any case file (case_summary.json, prefetch.json, evidence.json, paper_extracts.json all lack conservation data), so the 'nucleotide is not highly conserved' element of BP7 could not be verified.
• BP7 and BP4 would rest on the same single in-silico prediction for this variant (SpliceAI max delta 0.003; Pangolin SG 0.002 / SL -0.002); only one of the two may be credited, and BP4 was selected because it carries an explicit published calibration (SpliceAI <= 0.1) in this case's governing generic framework.
• No gene-specific (VCEP/CSPEC) BP7 rule for AXIN2 exists to resolve which of the two synonymous-variant codes is preferred (ClinGen CSPEC not found; generic ACMG/AMP 2015 fallback, framework_complete = false).

Sources

Reasoning

PS2 requires a variant observed to have arisen de novo, with both parental samples shown by identity testing to be the biological parents, in a patient whose phenotype matches the gene's disease association with reasonable specificity and whose family history is consistent with de novo inheritance. No proband-level clinical data, no parental genotypes, and no phenotype description exist for this case: the ClinVar record for NM_004655.4(AXIN2):c.2433G>A (variation 182020) is an aggregate of 12 clinical-testing submissions (9 laboratories reporting Likely benign, 2 reporting Benign) with no expert-panel review, and not one submission reports a de novo occurrence, parental testing, or any family data. All four PMIDs screened for this case are gene-general or framework papers (ACMG/AMP 2015 guidance; a cancer-predisposition referral guideline; a laboratory scoring-system paper; Sherloc) and none mentions AXIN2 c.2433G>A, p.(Glu811=), or any de novo event at this locus. The prerequisite observations for PS2 cannot be established from the available evidence, so the criterion is not assessed rather than met or refuted.

Rule

PS2 (ACMG/AMP 2015, Richards et al.): de novo (both maternity and paternity confirmed) in a patient with the disease and no family history. Conditions: both parental samples shown by identity testing to be biological parents; family history consistent with de novo inheritance; patient phenotype matches the gene's disease association with reasonable specificity.

Evidence considered

• PS2 requires a de novo observation with parentage confirmed by identity testing, a phenotype matching the gene-disease association, and a family history consistent with de novo inheritance; PS2 is not supported by an assumed (unconfirmed) de novo event, which is the PM6 situation instead.
• The ACMG/AMP 2015 guideline text states that 'A variant observed to have arisen de novo (parental samples test negative), is considered strong support for pathogenicity if the following conditions are met: Both parental samples were shown through identity testing to be the biological parents of the patient. Note: PM6 applies if identity is assumed but not confirmed.'
• ClinVar variation 182020 aggregates 12 submissions from clinical laboratories, all classified Likely benign (9 laboratories) or Benign (2 laboratories), with no expert-panel submission; none of the submissions reports de novo status, parental testing, or family history for the proband.
• The ClinVar submission audit for this record lists only PP3-type in-silico reasoning among the extracted evidence signals and no de novo, segregation, or family-based assertion.
• The Karbassi et al. laboratory scoring study describes de novo identification as one of the family-data types that can move a variant score upward, but reports no AXIN2 variant and no de novo event for c.2433G>A; it is methodological background only.
• No source consulted for this case reports AXIN2 c.2433G>A (p.Glu811=) in a proband with parental testing, so the de novo observation PS2 depends on is absent.

Still missing

• No proband-level clinical information (phenotype, age at onset, disease status) for the individual in whom c.2433G>A was identified.
• No parental testing results and no confirmation of biological parentage, so a de novo occurrence cannot be confirmed or excluded.
• No report of a de novo occurrence of AXIN2 c.2433G>A in any consulted source (ClinVar submissions, literature).

Sources

Reasoning

No AXIN2-specific ClinGen CSPEC or VCEP specification is available for this case (cspec found=false; final classification framework source=generic_acmg_fallback; framework_complete=false), so the generic ACMG/AMP 2015 criteria govern (Richards et al., PMID:25741868). Generic PM3 is met only for a recessive disorder when the variant being evaluated is demonstrated in trans with a pathogenic (or likely pathogenic) variant in an affected individual; it therefore requires (a) proband-level genotyping that shows a second pathogenic allele and (b) an established or strongly supported phase, plus a recessive disease context. None of these is present for NM_004655.4:c.2433G>A (NP_004646.3:p.(Glu811=)) in AXIN2. The case evidence contains no proband genotype, no second AXIN2 variant, no experimental or pedigree-based phase determination, no affected-proband observation, and no family-segregation data. ClinVar (VCV000182020, 11 submissions) supplies only Benign/Likely-benign assertions with no expert-panel review, and no submission reports an allelic configuration or an in-trans partner variant for this record; the ClinVar-listed conditions (AXIN2-related disorder, hereditary cancer-predisposing syndrome, oligodontia-cancer predisposition syndrome) are described with autosomal dominant cancer/tooth-agenesis inheritance, not an established recessive disease context. Population data provide no recessive-type support either: gnomAD v4.1 reports 199/1,614,228 alleles and gnomAD v2.1 28/282,760 alleles, both with zero homozygotes. PM3 therefore cannot be assessed rather than applied.

Rule

Generic PM3: pathogenic or likely pathogenic variant observed in trans with the variant under evaluation in an affected individual with a recessive disorder; proband genotype and phase must be established or sufficiently supported.

Evidence considered

• Generic PM3 definition requires the variant in trans with a pathogenic or likely pathogenic variant in an affected individual with a recessive disorder; phase must be established or sufficiently supported.
• The case documents only the variant description, database results and computational predictions; it contains no affected-proband observation, no second AXIN2 allele, no phase result and no family-segregation data for c.2433G>A.
• ClinVar VCV000182020 (11 submissions; Benign/Likely benign; zero expert-panel submissions; conditions listed as AXIN2-related disorder, hereditary cancer-predisposing syndrome, oligodontia-cancer predisposition syndrome) reports no allelic configuration and provides no in-trans observation for this variant.
• gnomAD v4.1: 199 alternate alleles in 1,614,228 with zero homozygotes; gnomAD v2.1: 28 alternate alleles in 282,760 with zero homozygotes - no homozygote is present to support a recessive mechanism.
• No AXIN2-specific CSPEC or VCEP specification was retrieved, so the generic ACMG/AMP 2015 framework applies.

Still missing

• Affected-proband genotype and phenotype context.
• A second pathogenic or likely pathogenic variant in the same individual.
• Experimental or pedigree-based cis/trans phase determination.
• An established recessive inheritance context for AXIN2 (ClinVar-listed conditions are dominant).

Sources

Reasoning

PM6 requires an assumed de novo occurrence, i.e. a proband carrying the variant with unaffected parents but without confirmation of maternity and paternity by identity testing, together with a phenotype consistent with the gene's disease association. This case contains no proband genotype context, no parental genotypes, and no phenotype description. The ClinVar record for NM_004655.4(AXIN2):c.2433G>A (variation 182020) comprises 12 clinical-testing submissions (Likely benign in 9 laboratories, Benign in 2), none of which asserts de novo status, parental status, or family history, and no expert panel has reviewed it. The screened literature (ACMG/AMP 2015 guidance, a cancer-referral guideline, a laboratory scoring-system paper, Sherloc) does not mention this variant or an assumed de novo event at this locus. With neither the proband observation nor the parental data available, PM6 cannot be evaluated and is not assessed; the same missing data also prevents PS2.

Rule

PM6 (ACMG/AMP 2015, Richards et al.): assumed de novo, but without confirmation of paternity and maternity (applies when an apparent de novo event in a patient with a consistent phenotype has no identity-tested parental confirmation).

Evidence considered

• PM6 covers the situation where a variant appears to have arisen de novo but parentage was not confirmed; the ACMG/AMP 2015 text states explicitly 'Note: PM6 applies if identity is assumed but not confirmed.'
• No consulted source reports c.2433G>A in a proband with unaffected parents, with or without identity testing; ClinVar variation 182020 records 12 submissions, all Likely benign or Benign, none including family or de novo data.
• ClinVar submissions for this record carry condition labels such as 'AXIN2-related condition' or 'not provided' and provide no phenotypic description that could be matched against the gene's disease association, which PM6 also requires.
• The Karbassi et al. study notes that family segregation data, 'especially the identification of de novo variants', commonly moved variant scores toward pathogenicity, but provides no proband-level observation for this variant.
• Because the proband and parental data needed for an assumed de novo call are wholly absent, the criterion is returned as not assessed with the gap documented rather than as met or refuted.

Still missing

• No proband-level record stating that c.2433G>A was observed in an affected individual with unaffected parents.
• No parental genotypes or parentage information, so neither a confirmed (PS2) nor an assumed (PM6) de novo event can be documented.
• No de novo assertion of any kind in the 12 ClinVar submissions or the literature screened for this case.

Sources

Reasoning

PP1 requires co-segregation of the variant with disease in multiple affected members of a family, supported by genotyped affected and unaffected relatives and counted meioses. This case provides no pedigree, no affected or unaffected relative genotypes, and no meiosis counts. ClinVar variation 182020 comprises 12 clinical-testing submissions for AXIN2 c.2433G>A (Likely benign in 9 laboratories, Benign in 2), all of which report the variant as a single-proband finding with no family-based assertion, and none of the screened PMIDs (ACMG/AMP 2015 guidance; ACMG/NSGC cancer-referral guideline; Karbassi et al. laboratory scoring system; Sherloc) reports segregation of this variant or any family carrying it. The Karbassi et al. paper is relevant only as methodology, describing how pedigree segregation is scored from LOD estimates based on meiotic events, and explicitly deals with no AXIN2 family. With no family data at all, co-segregation cannot be evaluated and PP1 is not assessed.

Rule

PP1 (ACMG/AMP 2015, Richards et al.): co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease (strength may be increased with more segregating meioses).

Evidence considered

• PP1 requires co-segregation with disease in multiple affected family members; the ACMG/AMP 2015 framework treats family segregation as a distinct evidence class that must be based on genotyped relatives and counted meioses, not on population or in-silico data.
• ClinVar variation 182020 aggregates 12 submissions from clinical laboratories with no expert-panel review; none of the submissions reports a family study, segregation analysis, or relative genotypes for c.2433G>A.
• The Karbassi et al. scoring study describes its segregation methodology: segregation in family pedigrees was analysed by estimating LOD scores or by statistical association when family data were incomplete, with the LOD score based on the number of meiotic events; it names no AXIN2 variant and provides no segregation data for c.2433G>A.
• The other screened sources (ACMG/AMP 2015 guidance, an ACMG/NSGC cancer-predisposition referral guideline, and Sherloc) contain no variant-specific or family-specific segregation evidence for this variant.
• No pedigree-level data exist in this case, so co-segregation evidence for PP1 is absent rather than merely weak.

Still missing

• No pedigree or family history for the proband carrying c.2433G>A.
• No genotypes for affected or unaffected relatives, so co-segregation and meiosis counts cannot be computed.
• No publication reports a family in which AXIN2 c.2433G>A segregates with disease.

Sources

Reasoning

PP4 (Richards et al. 2015, PMID:25741868) requires that the patient's phenotype or family history be highly specific for a disease with a single genetic etiology. This criterion is intrinsically proband-level: it can only be evaluated against the clinical phenotype of the individual in whom the variant was found. No such data exist in this case. The case inputs are variant-centric only - prefetch.json, case_summary.json and the evidence bundle contain no HPO terms, no clinical description, no age or sex fields, and no family history for a proband, and none of the retrieved sources supplies a patient phenotype tied to NM_004655.4:c.2433G>A. The ClinVar submitters list condition labels ('AXIN2-related disorder', 'Hereditary cancer-predisposing syndrome', 'Oligodontia-cancer predisposition syndrome', 'not provided', 'not specified'), but these are submitter-level condition annotations on the variant record, not the proband's own phenotype, and they do not establish phenotype specificity for this individual. Because no numeric threshold applies to PP4 (the criterion is qualitative: phenotype 'highly specific for a disease with a single genetic etiology'), no inequality was constructed; the blocker is the absence of the required patient-level input, not a comparison result. PP4 is therefore not assessed rather than failed, and the gap is flagged for a human reviewer who may hold the clinical record.

Rule

PP4 (ACMG/AMP 2015, PMID:25741868): patient's phenotype or family history is highly specific for a disease with a single genetic etiology. Qualitative criterion (no numeric threshold); requires proband-level clinical data that are absent from this case.

Evidence considered

• PP4 definition consulted in the generic ACMG/AMP 2015 framework (PMID:25741868): 'Patient's phenotype or family history is highly specific for a disease with a single genetic etiology' - a qualitative, proband-level criterion with no numeric threshold, so no comparison operator applies.
• Case-input review: prefetch.json and case_summary.json contain no phenotype, HPO, age, sex, or family-history fields for a tested individual; the case is variant-centric (variant normalization, database lookups, ClinVar submission audit) with no clinical record.
• ClinVar variation 182020 condition labels were inspected and are submitter-level condition annotations, not individual-level phenotype data.

Still missing

• No proband clinical phenotype, HPO terms, age/sex, or family history was available in the case inputs (prefetch.json, case_summary.json, evidence bundle), so PP4's requirement for a phenotype highly specific for a single-gene disease could not be evaluated.
• ClinVar condition labels on variation 182020 (AXIN2-related disorder; Hereditary cancer-predisposing syndrome; Oligodontia-cancer predisposition syndrome) are submitter-level annotations, not the tested individual's phenotype, and are insufficient for PP4.

Sources

Reasoning

BS4 requires a demonstrated lack of segregation of the variant in affected members of a family, i.e. genotyped affected relatives who do not carry it. No family, pedigree, or relative genotypes are present for this case. ClinVar variation 182020 contains 12 clinical-testing submissions for AXIN2 c.2433G>A, all classified Likely benign (9 laboratories) or Benign (2 laboratories); none reports a family study or a non-segregating affected relative, and no expert panel has reviewed the record. The screened literature, including the ACMG/AMP 2015 guideline and the Karbassi et al. segregation-methodology paper, contains no family data for this variant. Notably, the benign classifications in ClinVar rest on other evidence categories (no criteria, or PP3-type in-silico reasoning) and cannot be converted into BS4; BS4 specifically requires the genetic-epidemiology observation of an affected relative who lacks the variant, which is absent, so the criterion is not assessed rather than met.

Rule

BS4 (ACMG/AMP 2015, Richards et al.): lack of segregation in affected members of a family (the variant is absent in affected relatives who would be expected to carry it).

Evidence considered

• BS4 requires lack of segregation in affected members of a family, an observation that depends on genotyped affected relatives; it is a family-based criterion and cannot be inferred from in-silico predictions, population frequency, or a submitter's overall classification.
• ClinVar variation 182020 aggregates 12 submissions (Likely benign from 9 laboratories, Benign from 2) with no expert-panel review; the submission audit extracts only an in-silico PP3-type signal (GeneDx describing a conservative, poorly conserved change) and no segregation or family-based assertion.
• The Karbassi et al. paper states that family segregation data most often moved variant scores upward toward pathogenicity, the opposite direction to a lack-of-segregation argument, and reports no family for AXIN2 c.2433G>A.
• The remaining screened sources (ACMG/AMP 2015 guidance, ACMG/NSGC cancer-referral guideline, Sherloc) provide no segregation or non-segregation data for this variant.
• Because no affected relative has been genotyped, the observation BS4 requires is not merely unsupported but untestable with the evidence available.

Still missing

• No pedigree and no genotyped affected family members who lack c.2433G>A.
• No family study or non-segregation observation for this variant in any consulted source.
• ClinVar classifications for this record are Likely benign/Benign but are not accompanied by segregation-based assertions that could be re-used as BS4 evidence.

Sources

Reasoning

No AXIN2-specific ClinGen CSPEC or VCEP specification is available (cspec found=false; final classification framework source=generic_acmg_fallback), so generic ACMG/AMP 2015 applies (Richards et al., PMID:25741868). Generic BP2 is met only when the variant is observed in trans with a pathogenic variant for a fully penetrant dominant gene/disorder, or observed in cis with a pathogenic variant in any inheritance pattern; both readings require a second (pathogenic) variant in the same individual plus an established allelic configuration. The case contains no second variant, no proband genotype, no experimental or pedigree-based cis/trans phase determination, no family data, and no inheritance or penetrance context that could place c.2433G>A on a defined parental allele. No ClinVar submission for VCV000182020 (11 submissions, all Benign/Likely benign, zero expert-panel, none reporting an allelic configuration or co-occurring pathogenic variant) supplies this information, and the population observations are unphased counts - gnomAD v4.1 199/1,614,228 alleles and v2.1 28/282,760 alleles, zero homozygotes - which cannot establish cis or trans configuration. The consequence-level data (splice prediction max delta 0.003; no REVEL score available for this synonymous change) are likewise uninformative for allelic configuration. BP2 therefore cannot be assessed; the decisive gap is the absence of phase information with a pathogenic partner variant.

Rule

Generic BP2: 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; phase and the applicable inheritance/penetrance context must be established.

Evidence considered

• Generic BP2 definition requires observation of the variant in trans with a pathogenic variant for a fully penetrant dominant gene/disorder, or in cis with a pathogenic variant in any inheritance pattern; both require an established phase.
• The case evidence documents only the variant description, database records and computational predictions; no second variant, phase result, affected-proband genotype or family data are present.
• ClinVar VCV000182020 (11 submissions; Benign/Likely benign; zero expert-panel) reports no co-occurring pathogenic variant and no allelic configuration for this record.
• gnomAD v4.1 reports 199/1,614,228 alleles and gnomAD v2.1 28/282,760 alleles (zero homozygotes in both); unphased population counts cannot demonstrate cis or trans configuration with a pathogenic variant.
• SpliceAI max delta 0.003 and the absence of a REVEL score provide no allelic-configuration evidence.
• No AXIN2-specific CSPEC or VCEP specification was retrieved, so the generic ACMG/AMP 2015 framework applies.

Still missing

• A second variant identified in the same individual or family.
• Experimental or pedigree-based cis/trans phase determination.
• Observation of a pathogenic or likely pathogenic partner variant (in cis or in trans).
• Relevant inheritance and penetrance context for the associated condition.

Sources

Reasoning

BP5 (Richards et al. 2015, PMID:25741868) requires that the variant be found in a case with an alternate molecular basis for the disease - i.e. the tested individual carries a separate, established genetic cause of the phenotype, making this variant's contribution unnecessary. Like PP4, BP5 is a case-level criterion: it can only be evaluated against the clinical and molecular findings of the individual in whom the variant was detected. No such information exists in this case. The case inputs (prefetch.json, case_summary.json, the evidence bundle) contain no proband record, no phenotype, no second-variant or co-occurring-finding data, and no indication of an alternate molecular diagnosis; the only molecular data available are lookups for this variant itself (gnomAD frequencies, SpliceAI, ClinVar, COSMIC, OncoKB, hotspot status), none of which addresses an alternate cause of disease in a patient. No retrieved paper reports the exact variant at all (zero hits for c.2433 and E811/Glu811 in the five AXIN2 germline articles and the three ClinVar-cited articles), so no published case-level alternate-diagnosis observation is available either. Some VCEP specifications define BP5 with a likelihood-ratio-style numeric value and a stated operator; no such AXIN2 VCEP/CSPEC specification was retrieved here (final_classification_framework: found = false), so no numeric threshold exists for this criterion in this case and no inequality could be written - the blocker is missing case-level input, not an unmet numeric comparison. BP5 is therefore not assessed rather than failed.

Rule

BP5 (ACMG/AMP 2015, PMID:25741868): variant found in a case with an alternate molecular basis for disease. Case-level criterion; under the generic fallback no numeric threshold or comparison operator is defined, and no AXIN2 VCEP BP5 likelihood-ratio threshold was retrieved.

Evidence considered

• BP5 definition consulted in the generic ACMG/AMP 2015 framework (PMID:25741868): 'Variant found in a case with an alternate molecular basis for disease' - a case-level criterion requiring proband-specific molecular findings.
• Case-input review: prefetch.json and case_summary.json contain no proband record, no co-occurring molecular finding, and no alternate molecular diagnosis; all retrieved molecular data pertain to this variant alone (gnomAD v2.1/v4.1 frequencies, SpliceAI max delta 0.00, ClinVar variation 182020, COSMIC no result, OncoKB Unknown Oncogenic Effect, no cancer hotspot).
• Variant-specific literature check: exact-string searches for 'c.2433' and 'E811'/'Glu811' returned zero hits across the five retrieved AXIN2 germline articles and the three ClinVar-cited articles, so no published case-level report of this variant with an alternate molecular cause exists in the retrieved evidence.
• Framework check: no AXIN2 VCEP/CSPEC specification was retrieved (final_classification_framework.found = false, source = generic_acmg_fallback); the generic BP5 wording contains no numeric threshold, so no comparison operator could be applied and no inequality was constructed.

Still missing

• No proband-level molecular or clinical record (co-occurring pathogenic variants, alternate molecular diagnosis, or phenotype) was available, so BP5's requirement for a case with an alternate cause of disease could not be evaluated.
• No AXIN2 VCEP/CSPEC specification with a numeric (likelihood-ratio-style) BP5 threshold and comparison operator was retrieved; if one is later supplied, BP5 must be re-evaluated against that operator as written.

Sources