FGFR2 c.346G>A, p.(Glu116Lys)

Reasoning

PM2 is met: the variant is absent from gnomAD v2.1 and gnomAD-Canada v1.0 and is present in gnomAD v4.1 at an extremely low allele frequency of 1.24e-06 (0.00012%, 2/1,614,194 alleles, 0 homozygotes), with the highest subpopulation AF only 1.69e-06 (NFE) and grpmax FAF 2.8e-07. This is well below the PM2 threshold of AF < 0.1% for non-VCEP genes (pipeline convention) and consistent with extreme rarity across all ancestry groups. Strength applied at supporting (PM2_Supporting), per current ClinGen SVI calibration of PM2; generic ACMG/AMP 2015 lists PM2 as moderate, so strength was down-weighted to supporting.

Rule

Generic ACMG/AMP 2015 (PMID 25741868) via generic_acmg_combination_rules; PM2 threshold AF < 0.1% in gnomAD for non-VCEP genes (pipeline convention); strength supporting per ClinGen SVI PM2_Supporting calibration.

Evidence considered

• gnomad_v4
• gnomad_v2
• gnomad_canada

Still missing

• None material. All subpopulations have AC=0 except NFE (AC=2); no ancestry-specific AF approaches any benign threshold.

Sources

Reasoning

Generic ACMG/AMP 2015 (PMID:25741868) BP4: multiple lines of computational evidence suggest no impact on the gene or gene product. Three independent calibrated computational lines are concordant in predicting no impact for NM_000141.4:c.346G>A (p.Glu116Lys): (1) missense path - REVEL 0.26, in the benign-supporting band (0.183-0.290) of the ClinGen SVI calibration verified against PMID 36413997 (supports no impact; value sits at the low end of the band, near the 0.290 neutral boundary); (2) missense path - BayesDel noAF -0.24425, in the benign-supporting tier (-0.36 to -0.18) of the pipeline SVCv4 ladder (placeholder/unverified calibration, corroborating line); (3) splice path - SpliceAI max delta 0.09 (DS_AG 0.0, DS_AL 0.09, DS_DG 0.09, DS_DL 0.07), below the 0.2 threshold for predicted splice-altering effect for this exonic variant (exon 3, 30 nt from the exon 3/intron 3 boundary, not in the splice consensus). All three lines agree on no impact, meeting the 'multiple lines' requirement; no single prediction is double-counted across PP3/BP4/BP7 (PP3 not met; BP7 not applicable - variant is missense, not synonymous). BP4 applied at supporting strength, the standard strength for this criterion under the generic framework. The splice-impact path (SpliceAI) and missense in-silico path (REVEL/BayesDel) are both owned by this group and both point to no impact.

Rule

Generic ACMG/AMP 2015 BP4 (Richards et al., PMID:25741868; source_registry key generic_acmg_combination_rules): 'Multiple lines of computational evidence suggest no impact on the gene or gene product (conservation, evolutionary, splicing impact, etc.).' Applied at supporting strength; three concordant benign-direction computational lines (REVEL, BayesDel, SpliceAI).

Evidence considered

• REVEL score 0.26 (local lookup, revel-v1.3) - benign-supporting band (0.183-0.290) per ClinGen SVI calibration verified against PMID 36413997; supports no impact on FGFR2
• BayesDel noAF score -0.24425 (local lookup, BayesDel_170824_noAF_chr10) - benign-supporting tier (-0.36 to -0.18) per pipeline SVCv4 ladder (placeholder calibration); supports no impact, concordant with REVEL
• SpliceAI max delta score 0.09 (DS_AG 0.0, DS_AL 0.09, DS_DG 0.09, DS_DL 0.07; variant_consequence=sequence_variant) - below the 0.2 splice-impact threshold; no predicted splice impact for this exonic missense
• Generic ACMG/AMP 2015 BP4 definition (PMID:25741868): 'Multiple lines of computational evidence suggest no impact on the gene or gene product (conservation, evolutionary, splicing impact, etc.)'

Still missing

• BayesDel noAF calibration is flagged placeholder/unverified in the pipeline SVCv4 in-silico threshold table (ACMG_V4/svc4/insilico.py); it is used only as a corroborating line, with REVEL (verified, PMID 36413997) and SpliceAI as the primary lines.
• REVEL 0.26 sits at the low end of the benign-supporting band (0.183-0.290), close to the 0.290 neutral boundary; the BP4 determination rests on the three-line concordance, not on REVEL alone.
• No RNA-based splicing assay or functional data available to experimentally confirm no impact; would strengthen but is not required for supporting-strength BP4.
• No gene-specific VCEP-calibrated BP4 thresholds exist for FGFR2; generic ACMG/AMP 2015 thresholds applied.

Sources

Reasoning

PM1 (Richards et al. 2015, PMID:25741868) requires the variant to lie in a mutational hot spot and/or a critical, well-established functional domain without benign variation. The hotspot prong is directly negative: the cancerhotspots.org query for FGFR2 E116 returned no result (evidence.json results.hotspots search_status=no_result), so p.Glu116 is not a statistically significant recurrent mutation hotspot; COSMIC lists the variant in somatic cancers (COSV60640640, n=3), which is recurrent but not a statistically significant hotspot and is somatic/cancer context rather than germline. No retrieved source places Glu116 (exon 3, N-terminal extracellular region of FGFR2) within a critical, well-established functional domain: OncoKB provides no variant-specific or domain-level functional annotation for E116K ('Unknown Oncogenic Effect', not_reviewed=true), and no domain/hotspot literature exists in the case bundle (literature_pass 0 PMIDs). The variant is present in gnomAD v4.1 at only 2/1,614,194 alleles (AF 1.24e-06, 0 homozygotes), which does not indicate benign common variation but also cannot establish hotspot/domain status. PM1 is not met on available evidence.

Rule

PM1 (Richards et al. 2015, PMID:25741868): located in a mutational hot spot and/or critical and well-established functional domain (e.g., active site of an enzyme) without benign variation

Evidence considered

• cancerhotspots.org query 'FGFR2 E116' (evidence.json results.hotspots, search_status=no_result): not a statistically significant hotspot (this source has no source_registry key; documented here as prose only)
• COSMIC COSV60640640 (evidence.json results.cosmic): somatic count 3; recurrent in cancer but not a statistically significant hotspot and not germline evidence (this source has no source_registry key; documented here as prose only)
• OncoKB FGFR2/E116K: 'Unknown Oncogenic Effect', no variant-specific reviewed functional evidence (not_reviewed=true)
• gnomAD v4.1: 2/1,614,194 alleles (AF 1.24e-06), 0 homozygotes - position is not a site of benign common variation, but no hotspot/domain signal either
• Variant located in exon 3 (prefetch.json variant_exonic_positions); no retrieved source maps Glu116 to a critical well-established functional domain

Still missing

• No curated FGFR2 domain annotation (e.g., UniProt/InterPro domain coordinates or a germline-specific hotspot map) was present in the evidence bundle to fully evaluate the 'critical well-established functional domain' prong at residue 116
• cancerhotspots.org (residue query FGFR2 E116) was the only hotspot resource consulted; no germline-specific hotspot resource was retrieved

Sources

Reasoning

Variant NM_000141.4:c.346G>A (FGFR2) is a missense substitution predicting NP_000132.3:p.(Glu116Lys), located in exon 3 (c.110-376) of NM_000141.4, 30 nt upstream of the exon 3/intron 3 boundary and not within the canonical splice consensus. No ClinGen CSPEC or VCEP specification exists for FGFR2 (cspec found=false; vcep_materials.json: 'no cspec url available'; gene_vcep_dir=null; framework_mode=generic_acmg; precedence_rank=3), so generic ACMG/AMP 2015 (Richards et al., PMID:25741868) governs. PP3 requires multiple lines of computational evidence supporting a deleterious effect on the gene or gene product. No calibrated computational line supports a deleterious effect: (1) splice path - SpliceAI max delta score 0.09 (DS_AG 0.0, DS_AL 0.09, DS_DG 0.09, DS_DL 0.07), below the 0.2 splice-impact threshold; (2) missense path - REVEL 0.26, below the 0.644 ClinGen SVI supporting-pathogenic threshold (verified calibration, PMID 36413997) and instead in the benign-supporting band (0.183-0.290); (3) missense path - BayesDel noAF -0.24425, negative, in the benign-supporting tier (-0.36 to -0.18) of the pipeline SVCv4 ladder (placeholder calibration). All three lines are neutral-to-benign direction; none predicts a deleterious effect, so PP3 is not met.

Rule

Generic ACMG/AMP 2015 PP3 (Richards et al., PMID:25741868; source_registry key generic_acmg_combination_rules): 'Multiple lines of computational evidence support a deleterious effect on the gene or gene product (conservation, evolutionary, splicing impact, etc.).' Requires concordant in-silico evidence of a deleterious effect; none present here (all three predictors neutral-to-benign direction).

Evidence considered

• SpliceAI max delta score 0.09 (DS_AG 0.0, DS_AL 0.09, DS_DG 0.09, DS_DL 0.07; variant_consequence=sequence_variant) - below the 0.2 splice-impact threshold; no predicted splice impact, does not support PP3
• REVEL score 0.26 (local lookup, revel-v1.3) - below the 0.644 ClinGen SVI supporting-pathogenic threshold (verified, PMID 36413997); no deleterious prediction, does not support PP3
• BayesDel noAF score -0.24425 (local lookup, BayesDel_170824_noAF_chr10) - negative, below the 0.06 supporting-pathogenic tier of the pipeline SVCv4 ladder (placeholder calibration); no deleterious prediction, does not support PP3
• Generic ACMG/AMP 2015 PP3 definition (PMID:25741868): 'Multiple lines of computational evidence support a deleterious effect on the gene or gene product (conservation, evolutionary, splicing impact, etc.)'

Still missing

• No gene-specific VCEP-calibrated PP3 thresholds exist for FGFR2; generic ACMG/AMP 2015 thresholds applied. Not outcome-relevant: no predictor returned a score in any pathogenic band.
• aGVGD and HCI prior not available for this case (HCI gene_not_supported, source_file null); not required for the not_met determination.
• No RNA-based experimental splicing data; not required since no splice impact is predicted and the missense predictors are concordantly non-pathogenic.

Sources

Reasoning

BA1 requires a population allele frequency above the benign threshold (generic ACMG/AMP 2015, PMID 25741868; pipeline gnomAD non-VCEP convention: BA1 AF > 1%). The highest observed allele frequency for FGFR2 c.346G>A is 1.69e-06 (0.00017%) in European (non-Finnish) gnomAD v4.1 and the global AF is 1.24e-06 (0.00012%, 2/1,614,194 alleles); the variant is absent from gnomAD v2.1 and gnomAD-Canada v1.0. Observed AF is ~4 orders of magnitude below the 1% BA1 threshold, so BA1 is not met.

Rule

Generic ACMG/AMP 2015 (PMID 25741868) via generic_acmg_combination_rules; BA1 threshold AF > 1% in gnomAD for non-VCEP genes (pipeline convention).

Evidence considered

• gnomad_v2
• gnomad_v4
• gnomad_canada

Still missing

• None. gnomAD v2.1, v4.1, and gnomAD-Canada are all consistent; no population approaches the BA1 threshold.

Sources

Reasoning

BS1 requires an allele frequency greater than expected for the disorder (pipeline gnomAD non-VCEP convention: BS1 AF > 0.3%). FGFR2-related craniosynostosis syndromes (Apert, Crouzon, Pfeiffer) are rare autosomal dominant congenital disorders with high penetrance, so the expected disease allele frequency is extremely low. The observed AF is 1.24e-06 (0.00012%) globally and 1.69e-06 (0.00017%) in the highest subpopulation (NFE) - more than 3 orders of magnitude below the 0.3% BS1 threshold and far below any plausible expected AF for these syndromes. BS1 is not met.

Rule

Generic ACMG/AMP 2015 (PMID 25741868) via generic_acmg_combination_rules; BS1 threshold AF > 0.3% in gnomAD for non-VCEP genes (pipeline convention).

Evidence considered

• gnomad_v2
• gnomad_v4
• gnomad_canada

Still missing

• No formal disease-prevalence/penetrance model for FGFR2 craniosynostosis was retrieved in this case bundle; however, observed AF is orders of magnitude below any plausible expected AF, so this does not affect the determination.

Sources

Reasoning

BS2 requires observation of the variant in healthy adults - homozygotes for recessive disorders or unaffected heterozygous adults for fully penetrant dominant disorders. gnomAD v4.1 shows only 2 heterozygous alleles (0 homozygotes) among 1,614,194 alleles, and the variant is absent from gnomAD v2.1 and gnomAD-Canada. gnomAD is a general-population resource (not a healthy-adult cohort), and for a fully penetrant congenital autosomal dominant disorder such as Apert/Crouzon syndrome, population heterozygote counts cannot be interpreted as evidence of unaffected adult carriers. Two alleles in ~1.6M do not constitute the multiple documented healthy-carrier observations BS2 requires. BS2 is not met.

Rule

Generic ACMG/AMP 2015 (PMID 25741868) via generic_acmg_combination_rules; BS2 requires observation in healthy adults (homozygous for recessive; unaffected adult heterozygous for fully penetrant dominant), not merely presence in a general population database.

Evidence considered

• gnomad_v4
• gnomad_v2
• gnomad_canada

Still missing

• No dedicated healthy-adult carrier cohort (e.g., phenotyped unaffected adult carriers) is available in this case bundle; gnomAD alone is insufficient to satisfy BS2.

Sources

Reasoning

BP1 (Richards et al. 2015, PMID:25741868) applies to a missense variant in a gene for which primarily truncating variants are known to cause disease. FGFR2 is the opposite: the established germline disease mechanism is heterozygous activating missense (gain-of-function) variants causing autosomal dominant craniosynostosis syndromes (Apert, Crouzon, Pfeiffer, Jackson-Weiss, Beare-Stevenson, Saethre-Chotzen), per retrieved gene-context sources (pvs1_gene_context.json embedded literature, e.g., PMID:38265560 'Heterozygous activating non-mosaic germline variants in FGFR2 have been linked to numerous autosomal dominantly inherited disorders including several craniosynostoses and skeletal dysplasia syndromes'; OncoKB gene background). Missense variants, not truncating variants, are the primary disease mechanism, so BP1 is not met.

Rule

BP1 (Richards et al. 2015, PMID:25741868): missense variant in a gene for which primarily truncating variants are known to cause disease

Evidence considered

• pvs1_gene_context.json embedded literature (PMID:38265560): 'Heterozygous activating non-mosaic germline variants in FGFR2 have been linked to numerous autosomal dominantly inherited disorders including several craniosynostoses and skeletal dysplasia syndromes'
• pvs1_gene_context.json embedded literature (PMID:41572294): 'Dominant missense mutations in FGFR1, FGFR2, and FGFR3 have been associated with at least 20 congenital disorders characterized by skeletal abnormalities'
• OncoKB gene background: germline FGFR2 mutations cause craniosynostosis syndromes; FGFR2 is a receptor tyrosine kinase altered by mutation/rearrangement/amplification in cancer - no truncation-predominant germline mechanism

Still missing

• None material to the determination; retrieved gene-context sources consistently describe FGFR2 germline disease as activating-missense driven rather than truncation-predominant

Sources

Reasoning

PS1 (Richards et al. 2015, PMID:25741868) requires the same amino acid change as a previously established pathogenic variant, regardless of nucleotide change (e.g., Val>Leu caused by G>C and G>T at the same position). No established pathogenic variant producing p.Glu116Lys is documented anywhere in the case bundle: the variant is absent from ClinVar across all six query formulations (gene+variant name, exact transcript HGVS NM_000141.4:c.346G>A, genomic HGVS NC_000010.10:g.123324982C>T), literature_pass.json retrieved 0 PMIDs, fulltext_map.json is empty, and the automated same-residue comparator harvest was skipped (pm5_candidates.json: pm5_mode 'unknown', 0 candidates). Sequence-level analysis additionally shows that p.Glu116Lys arises from the GAG->AAG change at c.346 (first base of codon 116): no alternative single-nucleotide change at c.346-348 can produce Lys, so the classic PS1 alternate-nucleotide scenario cannot occur for this exact amino acid change. Nevertheless, because an established-pathogenic record of p.Glu116Lys was never affirmatively ruled out by a dedicated search (harvest skipped, no variant-specific literature), the evidence is insufficient to either meet or definitively refute PS1; the criterion is not assessed on current evidence rather than being affirmatively not met.

Rule

PS1 (Richards et al. 2015, PMID:25741868): same amino acid change as an established pathogenic variant regardless of nucleotide change (e.g., Val>Leu caused by G>C and G>T at the same position)

Evidence considered

• ClinVar: no record for NM_000141.4:c.346G>A across 6 query formulations (gene+variant name, exact transcript HGVS, genomic HGVS); no submissions or classifications
• pm5_candidates.json: same-residue comparator harvesting skipped (pm5_mode='unknown'), 0 candidates collected
• literature_pass.json: 0 PMIDs retrieved; fulltext_map.json empty
• Sequence analysis (prefetch.json mutalyzer output): codon 116 = GAG; only c.346G>A (GAG->AAG) yields Lys, so no alternate single-nucleotide change produces p.Glu116Lys

Still missing

• No established pathogenic variant producing p.Glu116Lys identified in any retrieved source; a dedicated ClinVar/literature search of FGFR2 p.Glu116 variants was not performed (same-residue harvesting skipped, pm5_candidates pm5_mode='unknown')
• ClinVar contains no record for c.346G>A and literature_pass retrieved 0 PMIDs, so the established-pathogenic-variant comparator space is unevaluated; a manual ClinVar review of other variants at residue 116 (including alternate transcripts/isoforms such as NM_022970.4) would close this gap

Sources

Reasoning

No applicable CSPEC/VCEP exists for FGFR2 (vcep_materials.json: cspec found=false, final_classification_framework source=generic_acmg_fallback, scoring_model=ACMG/AMP 2015), so the generic ACMG/AMP 2015 definition governs: de novo occurrence with confirmed paternity and maternity. The consolidated case bundle contains no proband genotype data, no parental testing results, no trio/quad data, and no report of de novo occurrence anywhere (full-bundle search for 'de novo', 'proband', 'parent', 'maternal', 'paternal', 'meios', 'segregat', 'PS2' returned zero hits). ClinVar has no entry for this variant (variation_ids=[], no submissions, no expert panels), the literature pass identified 0 PMIDs, and no full-text papers were retrieved (publications/ empty, fulltext_map.json {}). Without any proband/parental genotype evidence the criterion cannot be evaluated; absence of data is not scored as met or not_met.

Rule

PS2: De novo (paternity and maternity confirmed) - ACMG/AMP 2015, Richards et al., PMID 25741868 (generic fallback, no FGFR2 CSPEC/VCEP)

Still missing

• No proband phenotype/genotype record in the case bundle from which a de novo occurrence could be established
• No parental (maternity/paternity-confirmed) testing results available
• No variant-specific literature reporting a de novo occurrence of NM_000141.4:c.346G>A (p.Glu116Lys)

Reasoning

PS3 (ACMG/AMP 2015, PMID:25741868: 'Well-established functional studies show a deleterious effect') cannot be evaluated because no well-established functional study of NM_000141.4:c.346G>A (p.Glu116Lys) exists anywhere in the retrieved evidence base. literature_pass.json returned 0 PMIDs, fulltext_map.json is empty, the publications directory contains no full texts, and no PVS1-pass article mentions this variant. ClinVar has no entry for the variant. OncoKB reports 'Unknown Oncogenic Effect' with no variant-specific reviewed functional evidence (not_reviewed=true, no PMIDs), and COSMIC (COSV60640640) lists only somatic occurrence (n=3) with no functional readout. REVEL (0.26), BayesDel (-0.24425), and SpliceAI (max delta 0.09) are in silico predictors, not validated assays, and cannot satisfy PS3. Biological context notes that FGFR2 germline disease is autosomal dominant and caused by activating/gain-of-function missense variants, and p.Glu116Lys falls in the extracellular ligand-binding (Ig-like) region where a signaling/activation assay would be the relevant functional readout (e.g., FGFR2 autophosphorylation, MAPK pathway activation, proliferation/transformation), but no such assay was retrieved. No FGFR2 CSPEC/VCEP exists (framework_mode=generic_acmg), so generic ACMG/AMP 2015 calibration applies. Evidence is insufficient in the damaging direction; because the absence of retrieved functional evidence reflects a retrieval gap (0 PMIDs) rather than proof that no study exists, PS3 is not_assessed and flagged for human review.

Rule

PS3 (ACMG/AMP 2015, PMID:25741868): Well-established functional studies show a deleterious effect; requires validated assay evidence with appropriate controls, not in silico prediction

Evidence considered

• oncokb
• spliceai
• revel
• bayesdel
• clinvar

Still missing

• No validated functional assay evidence for p.Glu116Lys (e.g., FGFR2 autophosphorylation/signaling, MAPK activation, proliferation/transformation assays with appropriate controls) was retrieved
• literature_pass.json retrieved 0 PMIDs and fulltext_map.json is empty; variant-specific functional literature may exist but was not surfaced by automated retrieval
• No FGFR2 CSPEC/VCEP specification available to calibrate PS3 strength; generic ACMG/AMP 2015 fallback in use

Sources

Reasoning

No ClinGen CSPEC/VCEP exists for FGFR2, so the generic ACMG/AMP 2015 framework (PMID 25741868) governs. PS4 requires a statistically significant increase in variant prevalence in affected individuals versus controls from case-control studies. The evidence bundle contains no case-control study, no affected-individual prevalence data, and no proband cohort for this exact variant (literature_pass.json reports 0 PMIDs across all sources; fulltext_map.json is empty). The only variant-occurrence datum is a somatic cancer count in COSMIC (COSV60640640, n=3), which cannot substitute for germline case-control enrichment in a Mendelian disease context and is not a source_registry key. Absence from gnomAD v2.1 and near-absence in gnomAD v4.1 (AF 1.24e-06, 2/1,614,194 alleles) is control-population frequency context handled by other groups (BA1/BS1/PM2), not PS4 evidence. Criterion not assessed due to absent evidence; a dedicated case-control or affected-cohort study reporting this exact variant is required before PS4 can be applied.

Rule

Generic ACMG/AMP 2015 (PMID 25741868): PS4 - prevalence of variant in affected individuals is significantly increased compared with controls (case-control studies). No such study exists for this variant.

Evidence considered

• clinvar: variant absent from ClinVar (no affected-individual submissions to draw prevalence from)
• gnomad_v4: AF 1.23901e-06 (2/1,614,194 alleles), grpmax FAF 2.8e-07 - control population frequency, not case-control enrichment
• gnomad_v2: absent from gnomAD v2.1 - control population frequency context only

Still missing

• No case-control or affected-vs-control enrichment study for FGFR2 c.346G>A (p.Glu116Lys) is available
• No affected-individual prevalence data (cohort series, patient registries) mentioning this exact variant
• No publications identified: literature_pass.json total_pmids = 0

Sources

Reasoning

PM5 (Richards et al. 2015, PMID:25741868) requires a novel missense change at an amino acid residue where a different missense change determined to be pathogenic has been seen before. No different pathogenic missense at FGFR2 p.Glu116 (e.g., p.Glu116Val, p.Glu116Gly) is documented in the case bundle: pm5_candidates.json collected 0 same-residue candidates and explicitly reports 'Unable to confirm classic same-residue PM5 semantics safely; skipping automatic candidate harvesting' with recommendation 'do not apply PM5 from this artifact alone'; ClinVar has no record for this variant and no same-residue submissions; literature_pass.json retrieved 0 PMIDs. Because the automated same-residue comparator search was skipped rather than performed, the evidence is insufficient to either meet or definitively refute PM5; the criterion is not assessed on current evidence.

Rule

PM5 (Richards et al. 2015, PMID:25741868): novel missense change at an amino acid residue where a different missense change determined to be pathogenic has been seen before

Evidence considered

• pm5_candidates.json: found=false, 0 same-residue candidates, recommendation level 'not_applicable' ('do not apply PM5 from this artifact alone')
• ClinVar: variant absent, no submissions, no linked PMIDs
• literature_pass.json: 0 PMIDs; no literature reporting a pathogenic missense at residue 116

Still missing

• Same-residue comparator harvesting was skipped (pm5_candidates.json pm5_mode='unknown', total_candidates_seen 0); no different pathogenic missense at p.Glu116 was identified in ClinVar or literature
• A manual ClinVar/literature review of other FGFR2 p.Glu116* variants (e.g., p.Glu116Val, p.Glu116Gly) is required to close this gap before PM5 could be applied

Sources

Reasoning

Generic ACMG/AMP 2015 definition governs (no FGFR2 CSPEC/VCEP): assumed de novo occurrence without confirmation of paternity and maternity. The case bundle contains no evidence of a de novo occurrence of c.346G>A in any proband: no trio/quad genotypes, no parental testing, no case report, and no variant-specific literature (literature_pass.json total_pmids=0; ClinVar entry absent; fulltext_map.json empty). PM6 cannot be assessed from the available evidence; it is not scored as met or not_met.

Rule

PM6: Assumed de novo, but without confirmation of paternity and maternity - ACMG/AMP 2015, Richards et al., PMID 25741868 (generic fallback, no FGFR2 CSPEC/VCEP)

Still missing

• No de novo occurrence reported for this variant in any proband
• No parental samples/genotypes (unconfirmed-parentage de novo cannot be inferred)
• No variant-specific publications available to screen for de novo case reports

Reasoning

Generic ACMG/AMP 2015 definition governs (no FGFR2 CSPEC/VCEP): co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease. The case bundle contains no family/pedigree data of any kind: no affected relatives tested, no segregation counts, no meioses, no inheritance pattern information (full-bundle search for 'segregat', 'family', 'sibling', 'affected relative', 'inheritance' returned zero hits). No variant-specific literature exists to supply segregation data (literature_pass.json total_pmids=0). The only segregation-type report surfaced in the case (PMID 41572294, FGFR2 c.2241G>C p.Q747H in a Kirner's deformity pedigree) concerns a different variant, was retrieved only for PVS1 gene-context purposes, is not a source_registry citation key, and per variant-grounded citation policy cannot be used for this variant. PP1 is not assessed due to absent evidence.

Rule

PP1: Co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease - ACMG/AMP 2015, Richards et al., PMID 25741868 (generic fallback, no FGFR2 CSPEC/VCEP)

Still missing

• No affected family members tested for c.346G>A in the case bundle
• No segregation/meioses count or LOD-style segregation data
• No variant-specific publications reporting segregation of c.346G>A in FGFR2

Reasoning

PP2 (Richards et al. 2015, PMID:25741868) requires a missense variant in a gene that has a low rate of benign missense variation AND in which missense variants are a common mechanism of disease. The second prong is supported by retrieved gene-context evidence: FGFR2 germline disease consists of autosomal dominant craniosynostosis/skeletal dysplasia syndromes caused by heterozygous activating missense variants (pvs1_gene_context.json embedded literature, e.g., PMID:38265560). The first prong ('low rate of benign missense variation') cannot be verified because gene-level missense constraint metrics for FGFR2 (gnomAD missense Z-score / o/e ratio) are absent from the evidence bundle (content search across the case output returned 0 matches for constraint/missense_z/oe_mis/z_score). With one of the two required prongs unevaluable, PP2 is not assessed on current evidence.

Rule

PP2 (Richards et al. 2015, PMID:25741868): missense variant in a gene that has a low rate of benign missense variation and in which missense variants are a common mechanism of disease

Evidence considered

• pvs1_gene_context.json: germline FGFR2 disease is caused by heterozygous activating non-mosaic missense variants (craniosynostosis syndromes) - supports the 'missense is a common mechanism of disease' prong
• OncoKB gene background: germline FGFR2 mutations cause craniosynostosis syndromes; no truncation-predominant mechanism
• No gnomAD missense constraint metrics (Z-score/o/e) present anywhere in the case output (verified by content search: 0 matches for constraint/missense_z/oe_mis/z_score)

Still missing

• gnomAD missense constraint metrics for FGFR2 (missense Z-score / o/e ratio) are absent from the evidence bundle; the 'low rate of benign missense variation' prong of PP2 cannot be verified
• No FGFR2 CSPEC/VCEP specification exists to confirm whether PP2 is applicable/enabled for this gene under a gene-specific framework

Sources

Reasoning

Generic ACMG/AMP 2015 (PMID 25741868) governs (no CSPEC/VCEP for FGFR2). PP4 requires the proband's phenotype or family history to be highly specific for a disease with a single genetic etiology matching the variant's gene. FGFR2 germline missense variants cause recognizable craniosynostosis/skeletal syndromes (Apert, Crouzon, Pfeiffer spectrum), and gene-level germline context is documented (pvs1_gene_context.json), but the bundle contains NO proband phenotype information, clinical description, or family history for this case. Without the patient's phenotype, phenotype specificity cannot be evaluated, and no FGFR2 phenotype can be asserted as matching. Criterion not assessed due to missing clinical data.

Rule

Generic ACMG/AMP 2015 (PMID 25741868): PP4 - variant's phenotype or family history is highly specific for a disease with a single genetic etiology. Cannot be evaluated without proband phenotype data.

Evidence considered

• pvs1_gene_context: confirms FGFR2 germline disease mechanism is literature-supported (craniosynostosis/neurocutaneous syndromes), establishing the gene-phenotype link but not the proband's phenotype

Still missing

• No proband phenotype/clinical description in the evidence bundle
• No family history information
• No clinical indication or phenotypic specificity assessment for this case

Sources

Reasoning

BS3 (ACMG/AMP 2015, PMID:25741868: 'Well-established functional studies show no damaging effect on protein function or splicing') cannot be evaluated because no well-established functional study demonstrating absence of a damaging effect for p.Glu116Lys was retrieved. SpliceAI predicts no significant splice impact (max delta score 0.09), but this is an in silico prediction and per ACMG/AMP does not qualify as BS3, which requires a validated functional assay with appropriate controls. REVEL (0.26) and BayesDel (-0.24425) are likewise in silico missense predictors and were excluded. OncoKB provides no variant-specific functional determination (Unknown Oncogenic Effect), COSMIC (COSV60640640) provides occurrence-only somatic data (n=3) with no functional readout, and the variant is absent from ClinVar with no literature PMIDs retrieved (literature_pass.json total_pmids=0, fulltext_map.json empty, publications directory empty). With zero functional assay evidence in either direction, BS3 cannot be assessed; the criterion must not be applied on the basis of in silico predictors alone, and the absence of evidence reflects a retrieval gap, so the criterion is not_assessed and flagged for human review.

Rule

BS3 (ACMG/AMP 2015, PMID:25741868): Well-established functional studies show no damaging effect on protein function or splicing; in silico prediction does not substitute for a functional assay

Evidence considered

• oncokb
• spliceai
• revel
• bayesdel
• clinvar

Still missing

• No functional assay demonstrating no damaging effect for p.Glu116Lys (e.g., signaling assays showing wild-type-equivalent FGFR2 activity) was retrieved
• SpliceAI (max delta 0.09) and other in silico predictors are not acceptable BS3 evidence under ACMG/AMP; a validated assay with appropriate controls is required
• No FGFR2 CSPEC/VCEP specification available to calibrate BS3 strength; generic ACMG/AMP 2015 fallback in use

Sources

Reasoning

Generic ACMG/AMP 2015 definition governs (no FGFR2 CSPEC/VCEP): lack of segregation in affected family members (non-segregation). No non-segregation observations exist in the case bundle: there is no family/pedigree data, no affected relatives tested, and no variant-specific literature (literature_pass.json total_pmids=0; ClinVar entry absent). The complete absence of segregation data means BS4 can neither be met (observed non-segregation) nor refuted; per instructions, insufficient evidence returns not_assessed with the gap identified. Note: gnomAD v4.1 carrier alleles (2/1614194) reflect population frequency and are not family non-segregation evidence and are handled by the frequency group.

Rule

BS4: Lack of segregation in affected family members - ACMG/AMP 2015, Richards et al., PMID 25741868 (generic fallback, no FGFR2 CSPEC/VCEP)

Still missing

• No affected family members genotyped for c.346G>A (no non-segregation observations)
• No pedigree/segregation dataset for this variant
• No variant-specific publications reporting segregation or non-segregation of c.346G>A

Reasoning

BP2 (ACMG/AMP 2015: 'Observed in trans with a pathogenic variant for a fully dominant disease/disorder; or observed in cis with a pathogenic variant in any inheritance pattern') requires an observation of this variant with known phase relative to a pathogenic FGFR2 variant in a proband/family. No such observation is available: the variant is absent from ClinVar (no submissions, no PMIDs), literature_pass.json found 0 PMIDs across all evidence sources, fulltext_map.json is empty, the publications directory contains no full texts, and no proband or parental/phase data is present in the case bundle. Without any proband observation carrying phase information, BP2 cannot be evaluated; evidence is insufficient rather than refuting.

Rule

BP2: Observed in trans with a pathogenic variant for a fully dominant disease/disorder; or observed in cis with a pathogenic variant in any inheritance pattern (Richards et al. 2015, PMID:25741868).

Still missing

• No proband or family observations of NM_000141.4:c.346G>A with phase information (in trans with a pathogenic variant for dominant FGFR2 disease, or in cis with a pathogenic variant). Variant absent from ClinVar and no literature PMIDs were retrieved; a ClinVar/literature case-level search documenting phase relative to a pathogenic FGFR2 variant would be needed to assess BP2.

Sources

Reasoning

Generic ACMG/AMP 2015 (PMID 25741868) governs. BP5 requires the variant to be found in a case in which an alternate molecular basis for the disease has been established in the proband (e.g., a second clearly pathogenic variant or another genetic diagnosis explaining the phenotype). The bundle contains no proband-level genetic workup: no reported alternative pathogenic variant, no secondary diagnosis, no segregation or family testing data. The variant is absent from ClinVar, and literature_pass found no case reports. Criterion not assessed due to absence of any proband molecular data.

Rule

Generic ACMG/AMP 2015 (PMID 25741868): BP5 - variant found in a case with an alternate molecular basis for disease. No proband-level molecular data available to evaluate.

Evidence considered

• clinvar: absence of any ClinVar record confirms no reported alternate-diagnosis case context is retrievable from this source

Still missing

• No proband case data establishing or excluding an alternate molecular basis for disease
• No secondary/second genetic diagnosis reported for this case
• No case reports or literature mentioning this exact variant with an alternate etiology

Sources