CUX1 c.2598_2606dup, p.(Ser868_Gly870dup)

Reasoning

BA1 is met at stand-alone strength: the highest observed population allele frequency for NM_001202543.1:c.2598_2606dup (p.Ser868_Gly870dup) in CUX1 is 1.99% in African/African American individuals (1496/75044 alleles, 12 homozygotes; gnomAD v4.1), exceeding the lab's non-VCEP BA1 operating threshold of 1% applied to the maximum population frequency. The high AFR frequency is corroborated by gnomAD v2.1 (AFR AF 1.77%, 437/24626 alleles, 5 homozygotes), gnomAD-Canada v1.0 (AFR AF 2.55%, 26/1018 alleles), and by the gnomAD v4.1 joint grpmax filtering allele frequency of 1.91% (0.0190942). A 9-bp tandem duplication present at ~2% in an African-ancestry population, with 12 homozygotes observed in an unselected general-population cohort, is incompatible with a rare fully penetrant disease-causing allele. The GC-rich tandem-duplication artifact risk was considered but is not supported: frequencies are concordant across three independent datasets (exomes and genomes), gnomAD-Canada reports high mapping quality (MQ 250) with an inbreeding coefficient near 0, and multiple homozygotes are observed.

Rule

ACMG/AMP 2015 (PMID 25741868) BA1: allele frequency in a reference population is too high to be consistent with disease causation (stand-alone benign; lab non-VCEP operating threshold: highest observed population AF > 1%)

Evidence considered

• gnomAD v4.1 (GRCh38, joint exome+genome, 1,613,378 alleles): total AF 0.10463% (1688 alleles), highest observed frequency in African/African American at AF 1.99350% (1496/75044 alleles, 12 homozygotes); joint grpmax FAF 0.0190942 (1.91%); NFE AF 0.0014%, AMR AF 0.153%, MID AF 0.0495%, SAS AF 0.0011%, EAS/FIN/ASJ AF 0%.
• gnomAD v2.1 (GRCh37, exome, 280,600 alleles): total AF 0.17391% (488 alleles, 5 homozygotes), highest observed frequency in African/African American at AF 1.77455% (437/24626 alleles, 5 homozygotes); grpmax FAF 0.0158866 (1.59%).
• gnomAD-Canada v1.0 (HostSeq, 10,487 genomes, 18,418 alleles): overall AF 0.1466% (27 alleles, 0 homozygotes), highest frequency in African/African American at AF 2.554% (26/1018 alleles); grpmax faf95 0.0179 (1.79%).
• Lab non-VCEP gnomAD operating calibration for generic ACMG population criteria: BA1 > 1% (applied to the highest observed population frequency), BS1 > 0.3%, PM2 < 0.1%; the AFR maximum frequency (1.99% in gnomAD v4.1; 1.91% grpmax FAF) exceeds the 1% BA1 threshold.
• Threshold-artifact assessment: the 9-bp GC-rich tandem duplication is concordant across three independent datasets and both sequencing types (exome and genome), with 12 homozygotes in gnomAD v4.1 and an inbreeding coefficient near 0 in gnomAD-Canada, indicating a genuine population polymorphism rather than a recurrent alignment artifact.
• No ClinGen CSPEC/VCEP or local gene-specific framework exists for CUX1 (vcep_materials.json: 'no cspec url available'; final_classification_framework found: false; gene_vcep_dir: null), so generic ACMG/AMP 2015 (PMID 25741868) governs via generic_acmg_combination_rules.

Sources

Reasoning

Adjudicated under generic ACMG/AMP 2015 (Richards et al., PMID 25741868), BP3 = 'In-frame indels in a repeat region without known function' (supporting benign). All three elements are met. (1) In-frame: the variant is a 9-bp duplication of 3 codons, NM_001202543.1:c.2598_2606dupTGGCAGCGG, p.(Ser868_Gly870dup) (Mutalyzer and VariantValidator normalization; gnomAD VEP 'inframe_insertion'; ClinVar reports the same change on the MANE transcript NM_181552.4 as c.2565_2573dup, p.(Gly859_Gly860insSerGlyGly)); no frameshift, no premature stop, NMD not applicable, exon 18 of 23. (2) Repeat region: the duplicated unit SGG (residues 868-870) is a tandem repeat of the immediately upstream unit (residues 865-867; reference 865-870 = SGGSGG), and the variant inserts a third tandem SGG copy (verified by CDS simulation); the region is a Gly/Ser-rich low-complexity stretch (33.8% G+S in the +/-40 aa window, 70% GC DNA context) within the UniProt P39880 intrinsically disordered region 815-930 flanked by polar-residue compositional-bias annotations. (3) Without known function: no variant-specific functional evidence exists (OncoKB: no variant-specific reviewed functional evidence; no PMIDs or full-text papers for this variant per literature_pass.json; COSMIC lists a single somatic occurrence COSV52917296 with no functional characterization); the variant site is outside all CUT-repeat and homeodomain DNA-binding domains (UniProt P39880: CUT1 542-629, CUT2 934-1021, CUT3 1117-1204, homeodomain 1244-1303); SpliceAI predicts no splice impact (max delta 0.006), so no splicing-mediated mechanism is available. BP3 is met at supporting (benign) strength.

Evidence considered

• Generic ACMG/AMP 2015 BP3 (Richards et al., PMID 25741868): 'In-frame indels in a repeat region without known function' (supporting benign). No CSPEC/VCEP exists for CUX1 (vcep_materials.json: generic_acmg_fallback), so the generic framework governs.
• In-frame: 9-bp duplication of 3 codons, NM_001202543.1:c.2598_2606dupTGGCAGCGG, p.(Ser868_Gly870dup) (Mutalyzer and VariantValidator normalization in prefetch.json; gnomAD-Canada VEP most_severe_consequence 'inframe_insertion'); protein length 1517 to 1520 aa; no frameshift, no premature stop, NMD not applicable; exon 18 of 23 in NM_001202543.1. ClinVar (variation 3041850) reports the same change on the MANE transcript NM_181552.4 as c.2565_2573dup, p.(Gly859_Gly860insSerGlyGly), confirming an in-frame Ser-Gly-Gly insertion across transcripts (ClinVar entry has no assertion criteria and is used here only for transcript/protein notation).
• Repeat region: the duplicated unit SGG (residues 868-870) is a perfect tandem repeat of the immediately upstream unit (residues 865-867; reference 865-870 = SGGSGG); the variant creates three tandem SGG copies (...GKEKGSGGSGGSGGGSQPR...), verified by CDS-level simulation using the NCBI NM_001202543.1 sequence. The region is Gly/Ser-rich low complexity (33.8% G+S in the +/-40 aa window; 70% GC DNA in c.2538-2667 with adjacent homopolymer runs c.2405 CCCCC, c.2418 CCCCC, c.2464 AAAAA). UniProt P39880 (CUX1) annotates the region as intrinsically disordered (residues 815-930) with flanking compositional-bias 'polar residues' stretches (868-877, 887-911); UniProt is not a source_registry key, so this is narrative context.
• Without known function: OncoKB (source_registry key 'oncokb') reports no variant-specific reviewed functional evidence for CUX1 S868_G870dup (classification 'Unknown Oncogenic Effect', not_reviewed = true); literature_pass.json found 0 PMIDs for this variant; COSMIC (evidence.json results.cosmic, COSV52917296, somatic count 1) lists a single uncharacterized somatic occurrence; the variant site is outside all CUX1 DNA-binding domains (UniProt P39880 CUT1 542-629, CUT2 934-1021, CUT3 1117-1204, homeodomain 1244-1303).
• SpliceAI Lookup (source_registry key 'spliceai') predicts no splice impact (max delta score 0.006; DS_AG 0.006, DS_AL 0.001, DS_DG 0.000, DS_DL 0.002), so no splicing-mediated mechanism exists; the only consequence is the in-frame repeat insertion, consistent with BP3.
• Context only (not primary BP3 evidence): gnomAD v4.1 overall AF 0.00105 with AFR AF 0.0199 and 12 homozygotes (gnomAD v2.1 AF 0.00174, 5 homozygotes) and the ClinVar Benign submission (1 laboratory, no assertion criteria, not usable per submission audit) are consistent with a tolerated polymorphic in-frame repeat insertion; population-frequency-based criteria (BA1/BS1/PM2) are adjudicated by the population group.
• CUX1 germline disease mechanism is loss of function/haploinsufficiency (pvs1_gene_context.json, lof_mechanism_supported = true); an in-frame insertion of 3 residues in a non-domain low-complexity region is not expected to produce loss of function, consistent with the 'without known function' assessment for BP3.

Still missing

• No direct functional assays for this specific in-frame insertion are available in the case bundle; 'without known function' is supported by the absence of variant-specific functional evidence (OncoKB), absence of variant-specific literature (literature_pass.json total_pmids = 0), and location outside all annotated CUX1 DNA-binding domains (CUT repeats, homeodomain). A functional study or a formal repeat-region annotation (RepeatMasker/Tandem Repeat Finder) would further firm up the assessment but is not required for the supporting-level application of BP3.

Sources

Reasoning

PM2 is not met: the variant is not rare in population databases. gnomAD v4.1 overall AF is 0.10463% (1688/1,613,378 alleles), which already exceeds the lab's non-VCEP PM2 ceiling of 0.1%; gnomAD v2.1 overall AF is 0.17391% (488/280,600 alleles); and the highest population frequency is 1.99% (African/African American, gnomAD v4.1). A variant present at >0.1% overall and ~2% in an African-ancestry population, with 12 homozygotes, is far too common to support PM2 (absent or extremely low frequency).

Rule

ACMG/AMP 2015 (PMID 25741868) PM2: absent from controls or at extremely low frequency in large population cohorts (lab non-VCEP operating ceiling: AF < 0.1%)

Evidence considered

• gnomAD v4.1 total AF 0.00104625 (0.10463%, 1688/1,613,378 alleles) exceeds the 0.1% PM2 ceiling; AFR AF 0.019935 (1.99%) is ~20-fold above the ceiling; 12 homozygotes observed.
• gnomAD v2.1 total AF 0.00173913 (0.17391%, 488/280,600 alleles) exceeds the 0.1% ceiling; AFR AF 0.017745 (1.77%); 5 homozygotes observed.
• gnomAD-Canada v1.0 overall AF 0.001466 (0.1466%, 27/18,418 alleles) also exceeds the 0.1% ceiling.
• Lab non-VCEP gnomAD operating calibration: PM2 < 0.1%; every dataset (total and maximum population AF) exceeds this ceiling, so PM2 is not met.
• No ClinGen CSPEC/VCEP or local gene-specific framework exists for CUX1, so generic ACMG/AMP 2015 (PMID 25741868) governs via generic_acmg_combination_rules.

Sources

Reasoning

PM3 requires (1) a recessive disorder and (2) the variant detected in trans with a pathogenic variant in an affected proband. Neither precondition is satisfied for CUX1 c.2598_2606dup (p.(Ser868_Gly870dup)). The case bundle contains no proband-level, familial, or phased (cis/trans) observations of this variant whatsoever: literature_pass.json reports 0 PMIDs, evidence.json/prefetch.json contain no segregation or phase data, and the single ClinVar submission (SCV004739178, 0-star, no assertion criteria, no PMIDs) reports no co-occurring pathogenic variant. The germline CUX1 disease context retrieved for this case (pvs1_gene_context) describes heterozygous/dominant transmission (e.g., a heterozygous truncating CUX1 variant segregating in a proband and affected father with GDD/ID; a de novo CUX1 variant) and a loss-of-function/haploinsufficiency mechanism, with no established recessive CUX1 phenotype. Population data further argue against a recessive pathogenic model: the variant is present in gnomAD v4.1 at AF 0.105% with 12 homozygotes, in gnomAD v2.1 at AF 0.174% with 5 homozygotes, and in gnomAD-Canada at 27/18,418 alleles with 0 homozygotes, i.e., homozygous carriers exist in general-population cohorts. In the absence of any trans observation with a pathogenic variant and any recessive disease model, PM3 is not met.

Rule

PM3: For recessive disorders, detected in trans with a pathogenic variant (ACMG/AMP 2015, Richards et al., PMID 25741868)

Evidence considered

• PM3 definition per generic ACMG/AMP 2015 (Richards et al., PMID 25741868): 'For recessive disorders, detected in trans with a pathogenic variant'; no CUX1 CSPEC/VCEP exists, so generic ACMG applies (vcep_materials.json: found=false, framework_mode=generic_acmg, scoring_model=ACMG/AMP 2015).
• No trans/cis or proband segregation data exists anywhere in the case bundle (literature_pass.json total_pmids=0; evidence.json and prefetch.json contain no phase, family, or second-variant observations).
• gnomAD v4.1 reports 1688/1,613,378 alleles (AF 0.105%) with 12 homozygotes; gnomAD v2.1 reports 488/280,600 alleles (AF 0.174%) with 5 homozygotes; gnomAD-Canada v1.0 reports 27/18,418 alleles with 0 homozygotes. Homozygotes in general-population cohorts provide no support for a recessive pathogenic effect and no trans observation.
• pvs1_gene_context germline disease context for CUX1 indicates heterozygous/dominant transmission (heterozygous truncating CUX1 variant in a proband and affected father with GDD/ID; a de novo CUX1 variant) and a loss-of-function/haploinsufficiency mechanism; inheritance array is empty and no recessive CUX1 phenotype is documented.
• ClinVar VCV003041850 (single submission SCV004739178, PreventionGenetics, 'no assertion criteria provided', no PMIDs, 0 stars) was checked for phase/family data; none is present, and the Benign label is not used as criterion evidence per the no-inference-from-final-classification rule.
• PM3 cannot be applied at any strength because the required trans observation with a pathogenic variant in affected probands of a recessive disorder is entirely absent.

Still missing

• No familial segregation or phase (cis/trans) data for this variant; no second CUX1 variant reported in any proband
• Recessive inheritance for any CUX1-related germline phenotype is not established; available context indicates heterozygous/dominant transmission
• No affected-proband observations of this exact variant in the literature (literature_pass total_pmids = 0)

Sources

Reasoning

Variant is a 9-bp in-frame duplication (c.2598_2606dupTGGCAGCGG, p.(Ser868_Gly870dup)) in exon 18 of CUX1; adjudicated under generic ACMG/AMP 2015 (Richards et al. 2015, PMID 25741868) because no CSPEC/VCEP specification exists for CUX1. The only applicable calibrated computational path for this non-SNV variant is the splice-impact path: SpliceAI Lookup returned a max delta score of 0.006 (DS_AG 0.006, DS_AL 0.001, DS_DG 0.000, DS_DL 0.002; predicted altered splice-site positions 297-497 nt from the variant), far below the 0.2 cutoff conventionally used to flag splice-altering potential in the SpliceAI publication (Jaganathan et al. 2019, PMID 30661751). No predicted splice impact. The missense in-silico path (REVEL/BayesDel/aGVGD) is not applicable: REVEL and BayesDel were skipped because the variant is not an SNV, and BayesDel has no verified published calibration available to this pipeline in any case. No computational evidence supports a deleterious effect on the gene or gene product, so PP3 is not met.

Evidence considered

• Adjudicated under generic ACMG/AMP 2015 PP3 ('multiple lines of computational evidence support a deleterious effect', Richards et al. 2015, PMID 25741868); vcep_materials.json confirms generic_acmg_fallback (no CSPEC for CUX1, precedence_rank 3).
• Variant is a 9-bp in-frame duplication, NM_001202543.1:c.2598_2606dupTGGCAGCGG, p.(Ser868_Gly870dup), within exon 18; gnomAD VEP most severe consequence 'inframe_insertion'.
• Missense in-silico predictors are not applicable to this variant class: prefetch.json records revel skipped ('not snv or no coords') and bayesdel skipped ('not snv or no coords'). BayesDel additionally has no verified published threshold available to this pipeline.
• SpliceAI Lookup (source_registry key 'spliceai') returned max delta score 0.006 (DS_AG 0.006, DS_AL 0.001, DS_DG 0.000, DS_DL 0.002; predicted altered splice-site distances -422/-490/+297/+497 nt). The SpliceAI algorithm (Jaganathan et al. 2019, PMID 30661751) reports delta scores in [0,1] with values >=0.2 conventionally used to indicate splice-altering potential; 0.006 is at noise level and predicts no splice impact, so the splice path provides no support for PP3.
• No other computational evidence (conservation scores, additional splice predictors) is present in the case evidence bundle.

Sources

Reasoning

PP5 is restricted by group policy to an exact-variant ClinVar expert-panel classification of Pathogenic or Likely pathogenic (3-star expert panel assertion). The ClinVar audit for VCV003041850 (VariationID 3041850) shows exactly one submission: PreventionGenetics (clinical testing laboratory), classified Benign, review status 'no assertion criteria provided', 0 review stars; expert_panel_submissions = 0. No expert-panel classification exists for this variant, so PP5 cannot be met. The aggregate Benign label from an ordinary laboratory is explicitly excluded from PP5.

Rule

Policy-restricted PP5: exact-variant ClinVar expert-panel (3-star) Pathogenic/Likely pathogenic classification only; derived from ACMG/AMP 2015 PP5 (PMID 25741868)

Evidence considered

• ClinVar VCV003041850 submission audit: 1 submission (SCV004739178, PreventionGenetics, clinical testing), classification Benign, review status 'no assertion criteria provided', 0 stars, is_expert_panel=false, expert_panel_submissions=0
• Group policy: PP5 applies only for an exact-variant ClinVar expert-panel Pathogenic/LP classification; ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions never trigger PP5
• No reputable published source (PMID) asserting pathogenicity for this variant was found (literature_pass returned 0 PMIDs)

Still missing

• No ClinVar expert-panel (3-star) Pathogenic/Likely pathogenic classification exists for this variant to evaluate

Sources

Reasoning

BS1 (AF greater than expected for the disorder; lab non-VCEP operating threshold >0.3%) is not applied: the highest observed allele frequency (1.99% in African/African American, gnomAD v4.1) exceeds the BS1 threshold but also exceeds the BA1 threshold (1%), so the frequency band triggers BA1 instead. BS1 and BA1 are mutually exclusive in the ACMG/AMP framework, and BA1 subsumes this observation at stand-alone strength; BS1 adds no independent evidence.

Rule

ACMG/AMP 2015 (PMID 25741868) BS1: allele frequency is greater than expected for the disorder (lab non-VCEP operating threshold 0.3%); not applied when the frequency meets the BA1 threshold

Evidence considered

• Highest observed population AF: 1.99% AFR (gnomAD v4.1), 1.77% AFR (gnomAD v2.1), 2.55% AFR (gnomAD-Canada); all exceed the 0.3% BS1 operating threshold.
• The same frequency exceeds the 1% BA1 operating threshold; per ACMG/AMP 2015 practice, BS1 is not applied when BA1 is met (mutually exclusive criteria), so BS1 is recorded as not_met without independent application.
• No ClinGen CSPEC/VCEP or local gene-specific framework exists for CUX1, so generic ACMG/AMP 2015 (PMID 25741868) governs via generic_acmg_combination_rules.

Sources

Reasoning

BS2 is not formally met: the variant is observed in the homozygous state in general-population cohorts (12 homozygotes in gnomAD v4.1, 5 in gnomAD v2.1, all in African/African American individuals; 0 in gnomAD-Canada), but the criterion requires observation in healthy adults for a disorder with full penetrance expected at an early age (homozygous for a recessive disorder, or heterozygous for a fully penetrant early-onset dominant disorder). CUX1-related condition has no established germline inheritance model, disease prevalence, or penetrance (CUX1 is principally characterized as a somatic haploinsufficient tumor suppressor in myeloid malignancies), and gnomAD is an unselected general-population resource without phenotype-based exclusion of affected individuals. The homozygous observations nonetheless support the benign direction already established by BA1.

Rule

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

Evidence considered

• gnomAD v4.1: 1688/1,613,378 alleles with 12 homozygotes (all in AFR; AFR homozygote count 12); gnomAD v2.1: 488/280,600 alleles with 5 homozygotes (all in AFR); gnomAD-Canada v1.0: 27/18,418 alleles with 0 homozygotes.
• Homozygous carriers are present in two independent datasets (gnomAD v2.1 and v4.1), which is notable, but gnomAD cohorts are not phenotype-verified as unaffected for CUX1-related conditions.
• CUX1-related condition lacks an established inheritance model and penetrance: gene context describes CUX1 as a haploinsufficient tumor suppressor recurrently deleted in somatic myeloid malignancies, with no established germline fully penetrant early-onset disorder.
• Generic ACMG/AMP 2015 BS2 (PMID 25741868) requires observation in healthy adults with full penetrance expected at an early age; this premise cannot be verified for CUX1, so BS2 is not met.
• No ClinGen CSPEC/VCEP or local gene-specific framework exists for CUX1, so generic ACMG/AMP 2015 (PMID 25741868) governs via generic_acmg_combination_rules.

Still missing

• No established disease prevalence or penetrance estimate is available for the CUX1-related condition (ClinVar condition label only), so the full-penetrance-expected-at-early-age premise of BS2 cannot be verified.
• gnomAD v2.1/v4.1 and gnomAD-Canada are unselected general-population cohorts; carrier and homozygote unaffected status for any CUX1-related phenotype is not phenotype-confirmed.
• No phenotyped case-control or family-based cohort data for this variant were available.

Sources

Reasoning

BP2 requires the variant to be 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. No such observation exists for CUX1 c.2598_2606dup (p.(Ser868_Gly870dup)): the case bundle contains no phased genotype data, no familial co-segregation data, and no second (pathogenic) CUX1 variant identified in any proband (literature_pass.json total_pmids = 0; ClinVar submission SCV004739178 reports no co-occurring variant and no assertion criteria). Although the germline CUX1 disease context suggests dominant/heterozygous inheritance, which would satisfy the 'fully dominant' precondition of the first BP2 clause, the required trans observation with a pathogenic variant is absent, so BP2 is not met. The ClinVar Benign label (0-star, no assertion criteria) is not used as evidence for BP2 because criteria must not be inferred from another source's final classification.

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 (ACMG/AMP 2015, Richards et al., PMID 25741868)

Evidence considered

• BP2 definition per generic ACMG/AMP 2015 (Richards et al., PMID 25741868): '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'; generic fallback applies because no CUX1 CSPEC/VCEP exists (vcep_materials.json: found=false).
• No trans or cis observation with a pathogenic variant is present in any case artifact: no phased genotypes, no segregation data, no second CUX1 variant in any proband (literature_pass.json total_pmids=0; prefetch.json/evidence.json contain no family data).
• pvs1_gene_context indicates a dominant/heterozygous germline CUX1 disease context (heterozygous truncating variant segregating in proband and affected father; de novo CUX1 variant; LoF/haploinsufficiency mechanism), satisfying the fully-dominant precondition of the first BP2 clause, but the required trans observation itself is absent.
• ClinVar VCV003041850 was reviewed for co-occurrence/phase information; the single submission (SCV004739178, PreventionGenetics, 'no assertion criteria provided', 0 stars, no PMIDs) provides none. Its Benign classification is excluded as evidence per the rule against inferring criteria from another source's final classification.
• BP2 is not met because the defining observation (in trans with a pathogenic variant in a fully dominant disorder, or in cis with a pathogenic variant) is entirely absent from the evidence base.

Still missing

• No observation of this variant in trans with a pathogenic variant in an affected individual, nor in cis with a pathogenic variant
• No phased (parental or family) genotype data available for this variant
• No second CUX1 variant identified in any proband within the available evidence

Sources

Reasoning

Adjudicated under generic ACMG/AMP 2015 (Richards et al. 2015, PMID 25741868), which defines BP4 as 'multiple lines of computational evidence suggest no impact on gene or gene product'. The only applicable computational line for this in-frame duplication is the splice path: SpliceAI max delta score 0.006 predicts no splice impact (splice-impact BP4 direction). However, this is a single line of computational evidence, and the missense in-silico path does not apply (REVEL/BayesDel skipped for non-SNV; BayesDel uncalibrated for this pipeline), and no other in-silico or conservation evidence is available in the case bundle. A single low SpliceAI score is insufficient to satisfy BP4's multiple-lines requirement under the generic framework, so BP4 is not met. If additional independent computational lines were available (e.g., a second applicable splice predictor), BP4 could be reconsidered at supporting strength.

Evidence considered

• Adjudicated under generic ACMG/AMP 2015 BP4 ('multiple lines of computational evidence suggest no impact', Richards et al. 2015, PMID 25741868); vcep_materials.json confirms generic_acmg_fallback (no CSPEC for CUX1).
• SpliceAI Lookup (source_registry key 'spliceai') returned max delta score 0.006 (DS_AG 0.006, DS_AL 0.001, DS_DG 0.000, DS_DL 0.002; predicted altered splice-site distances -422/-490/+297/+497 nt), i.e., no predicted splice impact; the SpliceAI algorithm (Jaganathan et al. 2019, PMID 30661751) reports delta scores in [0,1] with values >=0.2 conventionally used to indicate splice-altering potential. This provides a single line of computational evidence in the BP4 (no-impact) direction.
• Missense in-silico predictors cannot contribute additional lines for this variant class: prefetch.json records revel skipped ('not snv or no coords') and bayesdel skipped ('not snv or no coords'); BayesDel has no verified published threshold available to this pipeline.
• Variant is a 9-bp in-frame duplication, p.(Ser868_Gly870dup), in a Ser/Gly-rich region of CUX1; protein-level neutrality is addressed by other criteria groups (in-frame insertion path), not by BP4 computational evidence.
• No other computational evidence (conservation scores, additional splice predictors) is present in the case evidence bundle; single-line evidence does not meet BP4's multiple-lines requirement.

Still missing

• Only one applicable computational predictor (SpliceAI) exists for this non-SNV in-frame duplication; REVEL/BayesDel are not applicable to this variant class and no other in-silico evidence (conservation, additional splice predictors) is present. Additional independent computational lines would be required before BP4 could be considered met.

Sources

Reasoning

BP6 is restricted by group policy to an exact-variant ClinVar expert-panel classification of Benign or Likely benign (3-star expert panel assertion). The only ClinVar entry for this variant (VCV003041850) is a single ordinary clinical-laboratory submission (PreventionGenetics) labeled Benign with review status 'no assertion criteria provided' (0 stars); it is not an expert-panel assertion. Per group policy, ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions must never trigger BP6. Therefore BP6 is not met despite the laboratory Benign label.

Rule

Policy-restricted BP6: exact-variant ClinVar expert-panel (3-star) Benign/Likely benign classification only; derived from ACMG/AMP 2015 BP6 (PMID 25741868)

Evidence considered

• ClinVar VCV003041850 audit: single submission SCV004739178 (PreventionGenetics), Benign, 'no assertion criteria provided', 0 review stars, is_expert_panel=false, usable_for_classification=false
• Group policy: BP6 applies only for an exact-variant ClinVar expert-panel Benign/LB classification; the 0-star ordinary laboratory label is deliberately not used for BP6
• The ClinVar 'Benign' aggregate label was considered and rejected as a BP6 trigger because it derives from a non-expert assertion without assertion criteria

Still missing

• No ClinVar expert-panel (3-star) Benign/Likely benign classification exists for this variant; only a 0-star single-laboratory label is available

Sources

Reasoning

No proband-parent trio testing, parental confirmation (maternity and paternity), or family-history data for this variant exists anywhere in the case bundle. No ClinGen CSPEC/VCEP for CUX1 was available (vcep_materials: 'no cspec url available'; framework_mode = generic_acmg), so the generic ACMG/AMP 2015 PS2 rule applies: de novo occurrence with both maternity and paternity confirmed in a patient with the disease and no family history. The case provides no proband genotype for c.2598_2606dup, so PS2 cannot be evaluated, let alone met.

Rule

Generic ACMG/AMP 2015 (Richards et al., PMID 25741868): PS2 = de novo (both maternity and paternity confirmed) in a patient with the disease and no family history - strong evidence. Applied as generic fallback because no CUX1 CSPEC/VCEP exists.

Evidence considered

• Governing framework: generic ACMG/AMP 2015 (Richards et al., PMID 25741868) fallback, because no ClinGen CSPEC/VCEP exists for CUX1 (vcep_materials.json note 'no cspec url available'; case_summary framework_mode = generic_acmg; source_registry key generic_acmg_combination_rules). PS2 under this framework requires a de novo variant with both maternity and paternity confirmed in a patient with the disease and no family history.
• The case bundle contains no proband genotype, no parental testing results, and no family history data for NM_001202543.1:c.2598_2606dup; literature_pass.json reports total_pmids = 0 and fulltext_map.json is empty, so no source in the bundle reports this exact variant in any proband.
• PMID 41662332 was read in full (output/pvs1_articles/PMID_41662332.md): it reports a de novo CUX1 variant in a monozygotic twin pair with PANS, but the variant is c.G1808A (p.R603Q) (and a second family with c.T1822C, p.Y608H) - not c.2598_2606dup - so it provides no PS2 evidence for this variant.
• ClinVar (key clinvar) contains one Benign submission (PreventionGenetics, SCV004739178, review status 'no assertion criteria provided', 0 stars) with no validated PMIDs and no criterion-level leads; a lab classification without assertion criteria is not de novo evidence and contributes nothing to PS2.

Still missing

• No proband-parent trio testing data in the case bundle; requires demonstration of a de novo occurrence of c.2598_2606dup with both maternity and paternity confirmed
• No phenotype-consistency data linking the proband's phenotype to a CUX1-related condition is present in the bundle
• No family-history information (unaffected parents) is available

Sources

Reasoning

No well-established functional assay evidence for NM_001202543.1:c.2598_2606dup (p.(Ser868_Gly870dup)) in CUX1 was identified in any source in this case. The literature pass returned 0 PMIDs (literature_pass.json: 'No PMIDs found across evidence sources'), fulltext_map.json is empty, and no publications directory exists for this case. OncoKB explicitly reports no variant-specific reviewed functional evidence for this variant (classification 'Unknown Oncogenic Effect', gene-level curated context only). The single ClinVar submission (SCV004739178, PreventionGenetics, Benign) carries review status 'no assertion criteria provided' and no PMIDs, so it provides no functional data and cannot be used to infer any criterion. SpliceAI max delta 0.006 is an in-silico splice prediction, not a functional study. Generic ACMG/AMP 2015 PS3 requires well-established functional studies demonstrating a deleterious effect; per the ClinGen SVI functional evidence framework (Brnich et al., Genome Med 2019, PMID:31980526) this requires a validated assay with appropriate positive/negative controls and is strength-calibrated (single well-validated assay = moderate; strong requires more). No such evidence exists, so PS3 cannot be applied and is returned as not_assessed.

Rule

PS3: Well-established functional studies show a deleterious effect (ACMG/AMP 2015, Richards et al., PMID:25741868; generic_acmg_combination_rules). Strength calibrated per ClinGen SVI functional evidence framework (PMID:31980526).

Evidence considered

• OncoKB page for CUX1 S868_G870dup states: 'OncoKB did not identify variant-specific reviewed functional evidence for this variant; gene-level curated context is available for reviewer follow-up' (evidence.json, oncokb result; classification 'Unknown Oncogenic Effect') - documents absence of variant-specific functional evidence.
• ClinVar variation 3041850 (VCV003041850) contains a single submission (SCV004739178, PreventionGenetics, Benign, 'no assertion criteria provided', evaluated 2019-10-28) with no PMIDs and no functional assay described; per submission audit it is not usable for classification or criterion leads, and per adjudication rules a final classification from another source is never used to infer PS3.
• literature_pass.json reports total_pmids = 0 and note 'No PMIDs found across evidence sources'; fulltext_map.json is empty {} - no publication containing functional data for this variant was retrieved.
• SpliceAI max delta score 0.006 (evidence.json spliceai result) is an in-silico splice prediction; it does not qualify as a well-established functional study and is handled by the in-silico/splicing group, not PS3/BS3.
• PS3 strength calibration follows the ClinGen SVI functional evidence framework (Brnich et al., Genome Med 2019, PMID:31980526): well-established functional studies with validated controls; no assay evidence exists here, so no strength is assigned.

Still missing

• No functional studies testing this variant were located: literature pass retrieved 0 PMIDs for the case and the only CUX1-related papers present (gene-context set used for PVS1: PMIDs 32990483, 40836298, 41662332, 25645650, 38443608) address CUX1 loss-of-function disease context (myeloid malignancy haploinsufficiency, germline GDD/ID) at gene level and do not assay c.2598_2606dup.
• A validated functional assay demonstrating a deleterious effect of p.(Ser868_Gly870dup) (e.g., transactivation/reporter, DNA-binding, proliferation or differentiation assay with controls) would be required to apply PS3; none is available.

Sources

Reasoning

PS4 requires evidence that the variant's prevalence is significantly increased in affected individuals versus controls (case-control studies or an equivalent case series). No such evidence exists for CUX1 c.2598_2606dup: literature_pass.json returned 0 PMIDs, no published case report or case series mentions this exact variant in affected individuals, and the only variant-level observation outside population databases is a single somatic COSMIC entry (COSV52917296, n=1), which is somatic cancer data, not germline case-control enrichment, and is far below any PS4 threshold. Population frequency data (gnomAD v4.1 total AF 0.1046%, 12 homozygotes; grpmax FAF 0.0191) provides context suggesting the variant is common in the general population, but that is population evidence handled by other criteria, not case-control data. Criterion not assessable with the available evidence.

Rule

ACMG/AMP 2015 PS4 (PMID 25741868): prevalence of the variant in affected individuals is significantly increased compared with controls; requires case-control study or equivalent case evidence

Evidence considered

• literature_pass.json reports 0 PMIDs; no case-control or case-series evidence for c.2598_2606dup was found in the literature pass
• COSMIC lists a single somatic observation (COSV52917296, somatic count 1); a single somatic tumor finding does not establish germline case-control enrichment and is insufficient for PS4
• gnomAD v4.1 total AF 0.1046% (1688/1613378 alleles, 12 homozygotes), AFR AF 1.99%, grpmax FAF 0.0191 and gnomAD v2.1 total AF 0.1739% were consulted as population context; they are population frequency data, not affected-versus-control enrichment data, and were not used to support PS4
• Generic ACMG/AMP 2015 (PMID 25741868) PS4 rule applied because no CSPEC/VCEP exists for CUX1

Still missing

• No case-control study comparing variant frequency in CUX1-related disorder cases versus matched controls
• No case series or case reports of this exact variant (c.2598_2606dup / p.(Ser868_Gly870dup)) in affected individuals
• literature_pass returned 0 PMIDs; no eligible literature was retrieved for this variant

Sources

Reasoning

PM6 requires an assumed de novo occurrence (maternity and paternity NOT confirmed) of the variant in a proband. The case bundle contains no proband-parent testing data whatsoever for c.2598_2606dup, so an assumed de novo event cannot even be hypothesized from the evidence. The only de novo CUX1 variants in the reviewed literature (PMID 41662332: p.R603Q, p.Y608H; PMID 40836298: c.2637G>A p.Trp879*) are different variants and cannot be used.

Rule

Generic ACMG/AMP 2015 (Richards et al., PMID 25741868): PM6 = assumed de novo, but without confirmation of maternity and paternity - moderate evidence. Applied as generic fallback because no CUX1 CSPEC/VCEP exists.

Evidence considered

• Generic ACMG/AMP 2015 (Richards et al., PMID 25741868) PM6 = assumed de novo, without confirmation of maternity and paternity - moderate evidence; governing fallback since no CUX1 CSPEC/VCEP exists (vcep_materials.json: 'no cspec url available').
• No proband or parental genotype data for NM_001202543.1:c.2598_2606dup appears in any case artifact (case_summary, prefetch, evidence, literature_pass total_pmids = 0, fulltext_map empty); there is therefore no basis for an assumed-de novo assertion.
• PMID 41662332 (full text read) reports a de novo CUX1 variant (c.G1808A, p.R603Q) in a twin pair - a different variant; PMID 40836298 (full text read) reports CUX1 c.2637G>A (p.Trp879*) segregating in a father/daughter pair - also a different variant. Neither supports PM6 for c.2598_2606dup.
• ClinVar submission (SCV004739178, Benign, no assertion criteria provided, 0 validated PMIDs) carries no de novo or family testing information usable for PM6.

Still missing

• No parental testing data for the proband of this case; cannot establish an assumed de novo occurrence (even without confirmation of maternity/paternity)
• No proband phenotype/segregation context linking this exact variant to disease is available in the bundle

Sources

Reasoning

PP1 requires co-segregation of the variant with disease in multiple affected family members in a gene definitively known to cause the disease. No pedigree, affected-family-member genotyping, or segregation data for c.2598_2606dup exists in the case bundle (literature_pass: 0 PMIDs; fulltext_map empty; ClinVar submission carries no segregation information). PP1 cannot be evaluated.

Rule

Generic ACMG/AMP 2015 (Richards et al., PMID 25741868): PP1 = co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease - supporting evidence. Applied as generic fallback because no CUX1 CSPEC/VCEP exists.

Evidence considered

• Generic ACMG/AMP 2015 (Richards et al., PMID 25741868) PP1 = co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease - supporting evidence; governing fallback since no CUX1 CSPEC/VCEP exists (vcep_materials.json: 'no cspec url available').
• No segregation/pedigree data for NM_001202543.1:c.2598_2606dup is present in the case bundle; literature_pass.json total_pmids = 0 and fulltext_map.json = {}, so no publication in scope reports familial segregation of this exact variant.
• The ClinVar record (variation 3041850) has a single submission (SCV004739178, Benign, no assertion criteria provided, 0 stars, no PMIDs) with no family/segregation detail; it cannot support PP1.
• PMID 40836298 describes segregation of a DIFFERENT CUX1 variant (NM_001202543.2:c.2637G>A, p.Trp879*) in a proband and her father; because the variant differs from the case variant, it is not PP1 evidence here.

Still missing

• No affected-family-member genotyping or pedigree segregation data for this exact variant
• No evidence of multiple affected individuals carrying c.2598_2606dup in any family in the bundle

Sources

Reasoning

PP4 requires a phenotype or family history highly specific for a disease with a single genetic etiology in a patient carrying the variant. The case bundle contains no proband clinical phenotype or family-history data, and no publication was found that reports this exact variant in a patient with a CUX1-related phenotype (e.g., the GDD/ID phenotype described for CUX1 loss-of-function variants). The germline CUX1 context papers in the bundle (e.g., PMID:40836298, PMID:32990483, PMID:41662332) describe other CUX1 variants or gene-level context, not this variant, so they cannot supply variant-specific phenotype evidence. Criterion not assessable with the available evidence.

Rule

ACMG/AMP 2015 PP4 (PMID 25741868): phenotype or family history highly specific for a disease with a single genetic etiology

Evidence considered

• Case bundle contains no proband clinical phenotype or family-history information; PP4 cannot be evaluated
• Gene-context publications screened (PMID:40836298, PMID:32990483, PMID:41662332, PMID:25645650, PMID:38443608) discuss other CUX1 variants or gene-level disease mechanism; none report the index variant with a specific phenotype
• ClinVar VCV003041850 submission (PreventionGenetics, condition label 'CUX1-related condition') provides no phenotype-level detail usable for PP4

Still missing

• No proband phenotype or family-history data provided in the case bundle
• No published case of this exact variant with a phenotype highly specific for CUX1-related disorder
• No ClinVar submission with usable clinical phenotype detail (only submission is a laboratory label without assertion criteria)

Sources

Reasoning

No well-established functional study showing no damaging effect on protein function or splicing was identified for this variant, so BS3 cannot be applied (generic ACMG/AMP 2015). The evidence that exists is not functional in nature: gnomAD population frequency (AF ~0.10% overall, ~2% in African/African American, 12 homozygotes in gnomAD v4.1) is population evidence handled by the population group (BS1/BA1), not a functional assay; the ClinVar Benign submission lacks assertion criteria and PMIDs and is a laboratory classification, not a functional result; OncoKB explicitly states no variant-specific reviewed functional evidence; SpliceAI max delta 0.006 is in-silico only and does not substitute for an RNA/splicing functional assay (RT-PCR, minigene). Per the ClinGen SVI functional evidence framework (Brnich et al., PMID:31980526), BS3 requires validated assay evidence with appropriate controls. Absence of such evidence means BS3 is returned as not_assessed rather than met.

Rule

BS3: Well-established functional studies show no damaging effect on protein function or splicing (ACMG/AMP 2015, Richards et al., PMID:25741868; generic_acmg_combination_rules). Strength calibrated per ClinGen SVI functional evidence framework (PMID:31980526).

Evidence considered

• OncoKB CUX1 S868_G870dup: 'OncoKB did not identify variant-specific reviewed functional evidence for this variant' - absence of any functional result, benign or damaging, for this exact variant.
• ClinVar variation 3041850 single submission classified Benign (PreventionGenetics, 'no assertion criteria provided', no PMIDs) was consulted but supplies no functional assay data; per adjudication rules a source's final classification is never used to infer BS3.
• gnomAD v2.1/v4.1 and gnomAD-Canada frequencies (incl. 5 and 12 homozygotes respectively) document high population frequency in AFR, but population frequency is not a functional study and is adjudicated by the population group (BA1/BS1), not BS3.
• SpliceAI max delta 0.006 is an in-silico splice prediction and is not a well-established functional study; it cannot support BS3 (handled by the in-silico/splicing group).
• BS3 strength calibration follows the ClinGen SVI functional evidence framework (Brnich et al., Genome Med 2019, PMID:31980526); no functional assay evidence exists here, so no strength is assigned.

Still missing

• No validated functional assay demonstrating normal (benign) function of p.(Ser868_Gly870dup) (or normal splicing of the c.2598_2606dup allele) was located in any case source; BS3 would require such an assay with appropriate positive/negative controls.
• No RNA-level assay (e.g., RT-PCR of patient/transfected RNA or minigene splicing assay) for this in-frame duplication is available; SpliceAI (max delta 0.006) is an in-silico predictor and cannot satisfy BS3.

Sources

Reasoning

BS4 requires lack of segregation of the variant in affected family members (e.g., an affected family member who does not carry the variant). No familial segregation testing data exists in the case bundle for c.2598_2606dup. Population frequency (gnomAD AF ~0.1% overall, ~2% in African/African American) and the single ClinVar Benign submission without assertion criteria are not segregation evidence and cannot substitute for BS4.

Rule

Generic ACMG/AMP 2015 (Richards et al., PMID 25741868): BS4 = lack of segregation in affected family members - strong benign evidence. Applied as generic fallback because no CUX1 CSPEC/VCEP exists.

Evidence considered

• Generic ACMG/AMP 2015 (Richards et al., PMID 25741868) BS4 = lack of segregation in affected family members - strong benign evidence; governing fallback since no CUX1 CSPEC/VCEP exists (vcep_materials.json: 'no cspec url available').
• No affected-family-member genotype data exists for NM_001202543.1:c.2598_2606dup in the case bundle (literature_pass total_pmids = 0; fulltext_map empty; no pedigree artifact), so non-segregation cannot be demonstrated.
• gnomAD v2.1/v4.1 frequencies (keys gnomad_v2, gnomad_v4) and gnomAD-Canada (key gnomad_canada) document general-population allele counts (AF 0.17391% v2.1, 0.10463% v4.1 overall; up to ~1.99% in AFR) - these are population-frequency evidence relevant to BA1/BS1/BS2-type criteria in other groups, not BS4 familial segregation evidence.
• ClinVar Benign classification (1 clinical laboratory, 'no assertion criteria provided', 0 stars, 0 usable submissions per clinvar_submission_audit) is a database-level classification, not evidence of lack of segregation within a family; it cannot be used for BS4.

Still missing

• No familial segregation testing (genotyping of affected family members) for this variant is available in the case bundle
• BS4 requires demonstrating an affected family member lacking the variant; no pedigree data is present

Sources

Reasoning

BP5 requires that the variant was found in a case with an alternate molecular basis of disease (i.e., a documented alternative cause explains the phenotype). The case bundle contains no proband-level clinical or molecular data, no identified alternative pathogenic variant or CNV for any case carrying this variant, and no published case report of this variant. Criterion not assessable with the available evidence.

Rule

ACMG/AMP 2015 BP5 (PMID 25741868): variant found in a case with an alternate molecular basis for disease

Evidence considered

• No proband or family data is present in the case bundle; no alternative molecular cause (second pathogenic variant, pathogenic CNV, etc.) is documented for any case with this variant
• literature_pass returned 0 PMIDs; no case-level report of this variant with an alternative diagnosis was found

Still missing

• No proband-level clinical/genetic data in the case bundle to evaluate an alternate molecular basis of disease
• No published case of this variant with a documented alternative molecular diagnosis

Sources