ATM c.273dup, p.(Lys92GlufsTer8)

Reasoning

NM_000051.4:c.273dup is a 1-bp duplication in exon 4 of ATM (default VCEP transcript NM_000051.3/ENST00000278616.8; MANE Select NM_000051.4; all exons constitutive per ATM VCEP v1.5). It produces a frameshift, p.(Lys92GlufsTer8), with a premature termination codon at protein position 99 of the 3,056-residue wild-type protein (Mutalyzer predicted mutant protein ends at residue 99). This is a null-variant class (frameshift) explicitly covered by the ATM VCEP v1.5 PVS1 decision guide (adapted from PMID 30192042), and germline LoF is an established disease mechanism (VCEP PVS1 applicable; pvs1_gene_gate 'eligible'). NMD is predicted: the PTC at codon 99 is far upstream of the most C-terminal pathogenic residue p.Arg3047 (VCEP v1.5) and of the terminal exon-exon junction, with ~59 downstream exon-exon junctions; it is not in the last exon and not near the extreme 3' end. No PVS1 caveat applies: not a 3'-end variant; not a splice/exon-skipping variant (SpliceAI max delta 0.00; not at canonical +/-1,2 positions); no alternate-isoform rescue concern for constitutive exon 4; variant is absent from gnomAD v2.1, v4.1, and gnomAD-Canada, consistent with LoF intolerance. Full-strength PVS1 (Very Strong) is therefore assigned. Note: the pipeline scaffold (pvs1_variant_assessment.json) defaults to the generic SVI PVS1 framework; the governing spec is the ATM VCEP v1.5 PVS1 decision tree, and both reach the identical conclusion.

Rule

ATM VCEP v1.5: 'Use ATM PVS1 Decision Tree' (adapted from PMID 30192042); NMD-predicted frameshift null variant upstream of p.Arg3047 in a constitutive exon -> PVS1 (Very Strong)

Evidence considered

• cspec
• vcep_atm_pvs1_1_5
• pvs1_gene_context
• pvs1_variant_assessment
• pvs1_generic_framework
• spliceai
• gnomad_v2
• gnomad_v4
• gnomad_canada

Sources

Reasoning

ATM HBOP VCEP v1.5 PM2 rule: frequency <=0.001% in the gnomAD v4 dataset, applied as PM2_Supporting (not moderate); the VCEP states that even n=1 in a single subpopulation is sufficiently rare for PM2_supporting. The variant NM_000051.4:c.273dup is absent from gnomAD v4.1 exome (AF=0, n=0), which satisfies the <=0.001% threshold; absence is corroborated by gnomAD v2.1 (exome) and gnomAD-Canada v1.0 (genomes). PM2 is therefore met at supporting strength, and per the VCEP instruction ('Use as PM2_Supporting (not moderate)') must not be upgraded to moderate. The VCEP's original-ACMG caveat that indel population data may be poorly called by NGS is noted but does not block the call because the threshold is frequency-based and the variant is absent across both exome and genome datasets.

Rule

Frequency <=0.001% in gnomAD v4 dataset; use as PM2_Supporting (not moderate); n=1 in a single subpopulation is sufficiently rare

Evidence considered

• gnomAD v4.1 (exome): variant absent (search_status=absent, AF=0, n=0)
• gnomAD v2.1 (exome): variant absent
• gnomAD-Canada v1.0 (HostSeq genomes): variant absent
• ATM HBOP VCEP v1.5 PM2 rule: Frequency <=0.001% in gnomAD v4 dataset; n=1 in a single subpopulation is sufficiently rare and PM2_supporting applies; use as PM2_Supporting (not moderate)

Sources

Reasoning

Under ATM VCEP v1.5, PM5 is applicable but is specified as a truncation-cutoff rule anchored to the most C-terminal known pathogenic residue p.Arg3047*, and the panel explicitly directs 'Use as PM5_Supporting (not moderate)'. The governing rule states PM5_Supporting applies to 'frameshifting or truncating variants with premature termination codons upstream of p.Arg3047' (and to splice variants with an NMD-prone PVS1_VS(RNA) event); the panel forbids use for start-loss and for missense variants. c.273dup is a frameshifting duplication producing p.Lys92GlufsTer8 with a premature termination codon at residue ~99 (position_last_predicted 99), well upstream of p.Arg3047, and is neither a start-loss nor a missense variant. PM5_Supporting is therefore met on a rule-based determination directly from the VCEP ruleset; pm5_candidates.json independently confirms the governing mode is 'truncation_cutoff_gene_specific' with zero classic same-residue missense comparators (classic same-residue PM5 search not eligible). Note: this is the VCEP-repurposed truncation PM5, not classic same-residue missense PM5, and it is distinct from (not double-counting) the PVS1 code assigned by the null-variant group; the VCEP framework explicitly places PM5_Supporting in the supporting bucket (e.g., Rule19 Condition2) combinable with PVS1 codes.

Rule

ATM VCEP v1.5 PM5: 'Apply to frameshifting or truncating variants with premature termination codons upstream of p.Arg3047. Apply to splice variants as with premature termination codons upstream of p.Arg3047 where PVS1_VS(RNA) is applied based on high quality observed splicing impact and must be NMD prone.' with instructionsToUse 'Use as PM5_Supporting (not moderate). Do not use for start-loss variants. Do not use for missense changes.'

Evidence considered

• ATM VCEP v1.5 CSPEC PM5 instructionsToUse ('Use as PM5_Supporting (not moderate)')
• ATM VCEP v1.5 PM5 rule (frameshifting/truncating variants with PTC upstream of p.Arg3047)
• Mutalyzer protein consequence p.Lys92GlufsTer8, PTC at codon 99 (< Arg3047)
• pm5_candidates.json (mode truncation_cutoff_gene_specific, 0 same-residue candidates)

Sources

Reasoning

Governing framework is the ClinGen HBOP ATM VCEP v1.5 points-based PM3/BP2 table (vcep_atm_pm3_bp2_1_5), which applies regardless of variant class; no generic ACMG fallback is needed. PM3 points accrue per unrelated A-T proband: phenotype 'confident' = 4.0 (confirmed in trans) / 2.0 (phase unknown) / 1.0 (second variant unidentified or VUS) / 2.0 (homozygous, max 2 individuals); 'consistent' phenotype = half those values; PM3 Supporting = 1.0 point, Moderate = 2.0, Strong = 4.0, Very Strong = 8.0. The co-occurring P/LP variant must itself be classified under VCEP specifications. For NM_000051.4:c.273dup (p.Lys92GlufsTer8, frameshift), zero A-T proband observations of the variant were identified in any searched source: (1) ClinVar search for ATM c.273dup returned no result (zero submissions - the variant is completely absent from ClinVar, so no A-T proband observation with any phase status exists there); (2) gnomAD v2.1, v4.1, and gnomAD-Canada all report the variant as absent (no alleles, no homozygotes); (3) COSMIC has no record; (4) none of the 3 retrieved full-text publications (PMID:27413114, PMID:30348496, PMID:30553448) mentions the variant. PM3 points total = 0.0, below the 1.0-point Supporting threshold, so PM3 is not met.

Rule

ATM VCEP v1.5 PM3/BP2 points table: PM3 Supporting = 1.0, Moderate = 2.0, Strong = 4.0, Very Strong = 8.0 points per unrelated A-T proband (phenotype confident: 4.0 confirmed-in-trans / 2.0 phase-unknown / 1.0 second variant unidentified or VUS / 2.0 homozygous [max 2 individuals]; consistent phenotype: half). Co-occurring P/LP variant must be classified under VCEP specifications. Observed A-T probands = 0 -> 0.0 points -> below the 1.0-point Supporting threshold -> PM3 not met.

Evidence considered

• vcep_atm_pm3_bp2_1_5
• cspec
• clinvar

Still missing

• No A-T proband observation of ATM c.273dup (p.Lys92GlufsTer8) exists in ClinVar (0 submissions), gnomAD (v2.1/v4.1/Canada all absent), COSMIC, or the retrieved literature; the variant appears to be novel/unreported.
• No ATM/A-T disease-specific variant database (e.g., LOVD) was queried in this case's evidence bundle; proband-level trans observations could in principle exist outside the queried sources.
• No phase or parental testing data exist for any observation of this variant, so trans configuration cannot be confirmed for any current or future proband report.

Sources

Reasoning

Variant is NM_000051.4:c.273dup (p.Lys92GlufsTer8), a coding frameshift duplication in exon 4 - not a missense, silent, in-frame, or intronic variant. ATM HBOP VCEP v1.5 PP3 has two sub-paths, neither is met: (1) Missense sub-path (REVEL > 0.7333) is not applicable - this is not a missense SNV; the pipeline REVEL and BayesDel lookups were skipped ('not snv or no coords') and the HCI prior/aGVGD resource does not support ATM (gene_not_supported). (2) Splicing sub-path (SpliceAI >= 0.2 for silent, missense/in-frame, and intronic variants outside donor/acceptor +/-1,2) is not met on the merits: SpliceAI max delta score = 0.00 (pangolin_SG 0.0, pangolin_SL -0.09), far below the 0.2 threshold, and a frameshift duplication is outside the variant classes the splicing sub-path enumerates. Additionally, the VCEP explicitly states PP3 for splice predictions may not be applied in addition to PVS1 or PVS1_Variable(RNA) codes; this frameshift null variant is expected to carry PVS1 (see pvs1_variant_assessment), which independently bars stacking PP3. No VCEP supporting spreadsheet pre-assigns PP3 for this variant: Suppl_TableS1_PMID 40580951 covers SNVs only (29,375 rows) and contains no c.273dup row.

Rule

ATM HBOP VCEP v1.5 PP3 (Pathogenic Supporting): Missense REVEL > 0.7333; Splicing: SpliceAI >= 0.2 for silent, missense/in-frame and intronic variants outside donor/acceptor +/-1,2. PP3 for splice predictions may not be applied in addition to PVS1 or PVS1_Variable(RNA).

Evidence considered

• spliceai: SpliceAI max delta score 0.00 (< 0.2 PP3 splicing threshold)
• cspec: ATM VCEP v1.5 PP3 rule (REVEL > 0.7333 missense; SpliceAI >= 0.2 splicing) and non-stacking note with PVS1
• prefetch/evidence: REVEL and BayesDel skipped for non-SNV; HCI prior not supported for ATM

Still missing

• No REVEL/BayesDel score exists for this variant because in silico missense predictors are not applicable to a non-SNV frameshift duplication; no alternative computational predictor for frameshift variants is part of the VCEP framework.

Sources

Reasoning

Governing framework is the ClinGen HBOP ATM VCEP v1.5 CSPEC (precedence rank 1), whose BA1 rule requires grpmax filtering allele frequency >0.5% in the gnomAD v4 dataset. The variant NM_000051.4:c.273dup (chr11-108229264-A-AG, GRCh38) is absent from gnomAD v4.1 exome (search_status=absent, AF=0), so the >0.5% threshold is not approached and BA1 is not met. Absence is corroborated by gnomAD v2.1 (exome) and gnomAD-Canada v1.0 (genomes). No population-frequency evidence supports a stand-alone benign call.

Rule

Grpmax Filtering AF >0.5% in gnomAD v4 dataset

Evidence considered

• gnomAD v4.1 (exome): variant absent (search_status=absent, AF=0)
• gnomAD v2.1 (exome): variant absent
• gnomAD-Canada v1.0 (HostSeq genomes): variant absent
• ATM HBOP VCEP v1.5 BA1 rule: Grpmax Filtering AF >0.5% in gnomAD v4 dataset

Sources

Reasoning

ATM HBOP VCEP v1.5 BS1 requires grpmax filtering allele frequency >0.05% in the gnomAD v4 dataset. The variant is absent from gnomAD v4.1 exome (AF=0), which is far below the >0.05% threshold, so BS1 is not met. This is consistent with absence in gnomAD v2.1 and gnomAD-Canada v1.0; the variant does not occur at a frequency greater than expected for ATM-related disease in any queried population dataset.

Rule

Grpmax Filtering AF >0.05% in gnomAD v4 dataset

Evidence considered

• gnomAD v4.1 (exome): variant absent (search_status=absent, AF=0)
• gnomAD v2.1 (exome): variant absent
• gnomAD-Canada v1.0 (HostSeq genomes): variant absent
• ATM HBOP VCEP v1.5 BS1 rule: Grpmax Filtering AF >0.05% in gnomAD v4 dataset

Sources

Reasoning

Under the ClinGen HBOP ATM VCEP v1.5 points-based PM3/BP2 table (vcep_atm_pm3_bp2_1_5), BP2 is applied to unaffected (non-A-T) individuals carrying the variant with a pathogenic/likely-pathogenic ATM variant in trans (or homozygous), where the individual is >=18 years old with no evidence of A-T (cspec instructionsToUse). Points: -4.0 (confirmed in trans), -2.0 (phase unknown), -2.0 lab / -1.0 database (homozygous, max -2.0); BP2 Supporting = -1.0 point, Moderate = -2.0, Strong <= -4.0. No such observations exist for c.273dup: the variant is absent from ClinVar (zero submissions, so no unaffected-carrier co-occurrence with a VCEP-classified P/LP ATM variant in trans is reported anywhere), gnomAD v2.1/v4.1/Canada show the variant as absent (no homozygous carriers, no carriers at all), COSMIC has no record, and none of the 3 retrieved full-text publications reports this variant in any individual. No evidence of the variant in cis with a pathogenic variant (generic ACMG BP2 axis) exists either. BP2 points total = 0.0 (no qualifying observations), so BP2 is not met.

Rule

ATM VCEP v1.5 PM3/BP2 points table: BP2 Supporting = -1.0, Moderate = -2.0, Strong <= -4.0 points per unaffected non-A-T individual (>=18 years, no evidence of A-T) carrying the variant with a P/LP variant in trans (confirmed trans -4.0; phase unknown -2.0; homozygous -2.0 lab / -1.0 database, max -2.0). Qualifying unaffected individuals = 0 -> 0.0 points -> above (i.e., less negative than) the -1.0 Supporting threshold -> BP2 not met.

Evidence considered

• vcep_atm_pm3_bp2_1_5
• cspec
• clinvar

Still missing

• No unaffected-carrier co-occurrence data (variant plus a VCEP-classified P/LP ATM variant in trans, age >=18 without A-T) are available for this variant; it is absent from ClinVar with zero submissions.
• No homozygous carriers observed in any population database queried (gnomAD v2.1/v4.1/Canada all absent), so the homozygous BP2 pathway cannot be assessed.
• No ATM/A-T disease-specific variant database (e.g., LOVD) was queried in this case's evidence bundle.

Sources

Reasoning

OVERRIDE CHANGES THE OUTCOME FROM THE VCEP'S LITERAL TEXT: Under the ATM HBOP VCEP v1.5 BP4 literal rule, the splicing sub-path (SpliceAI <= 0.1) is numerically satisfied (max delta score = 0.00) and the VCEP's BP4 instructions state splice analysis should be considered for all variant types including frameshift. However, the lab's BP4 one-mechanism override (vcep_materials.json gene_notes) requires evaluating exactly one applicable mechanism per variant: the missense sub-path (REVEL <= 0.249) applies to missense variants only and is not applicable here (non-SNV, REVEL unavailable/skipped); the splicing sub-path is reserved for non-missense variants for which splicing is the primary plausible mechanism (synonymous, intronic outside canonical +/-1,2). A frameshift duplication producing p.Lys92GlufsTer8 has protein truncation as its primary mechanism - a clean SpliceAI score (0.00) only rules out splice impact and does not clear the variant's own frameshift protein-level effect, exactly the double-counting the override forbids. BP4 therefore is not_met. No VCEP supporting spreadsheet pre-assigns BP4: Suppl_TableS1_PMID 40580951 (SNV-only, 29,375 rows) has no c.273dup row, and its computational predictors are secondary corroboration only.

Rule

ATM HBOP VCEP v1.5 BP4 (Benign Supporting): Missense REVEL <= 0.249; Splicing: SpliceAI <= 0.1 - as modified by the lab BP4 one-mechanism override (splice sub-path not applicable when splicing is not the variant's primary plausible mechanism, e.g. missense and frameshift variants).

Evidence considered

• spliceai: SpliceAI max delta score 0.00 (<= 0.1 BP4 splicing threshold, satisfied numerically but overridden)
• cspec: ATM VCEP v1.5 BP4 rule (REVEL <= 0.249 missense; SpliceAI <= 0.1 splicing)
• vcep_materials gene_notes: lab BP4 one-mechanism override restricting splicing sub-path to variants where splicing is the primary plausible mechanism

Still missing

• None evident for this variant class; note that if the lab ever chose to apply the VCEP's literal all-variant-type splice sub-path, BP4 would be met at supporting - this is a policy decision captured in the override, not an evidence gap.

Sources

Reasoning

No variant-specific functional assay evidence exists in the case bundle for NM_000051.4:c.273dup (p.Lys92GlufsTer8). The governing ClinGen HBOP ATM VCEP v1.5 framework (CSPEC, precedence rank 1) defines PS3 as: never use as Strong; PS3_Moderate when a variant fails to rescue BOTH an ATM-specific feature (e.g., phosphorylation of ATM-specific targets) AND radiosensitivity; PS3_Supporting when it fails to rescue an ATM-specific feature only (radiosensitivity-only is explicitly excluded). PS3 may only be applied from VCEP-approved assays - Mitui 2009 (PMID 18634022), Barone 2009 (PMID 19431188), Scott 2002 (PMID 11805335), all stable transfection of ATM cDNA into ATM-null LCLs with IR-induced substrate phosphorylation (S1981, SMC1, NBS1, CHK2, p53) and/or radiosensitivity readouts - and the calibration table (clingen_hbop_atm_supplementary_tables_1_and_2_v1.xlsx) explicitly states it is NOT a per-variant results table: the variant under assessment must be confirmed to actually appear in one of the three papers. None of those three papers are present in this case's publication set, and no report of c.273dup in any of them was found. The Sun et al. 2025 saturation screen (PMID 40580951, Suppl Table S1) was checked directly: it covers only SNVs (single-nucleotide Ref/Alt, ~29,374 rows) and c.273dup is a duplication, so no row exists for this variant and no DeepATM/Combined_score readout applies. OncoKB's 'Likely Loss-of-function' annotation is a curated variant-class label in a somatic-cancer context, not well-established functional assay evidence, and the three OncoKB-sourced papers (PMID 27413114, 30348496, 30553448) do not mention this variant in their full texts. PS3 is therefore not_assessed rather than met: a damaging functional effect is mechanistically expected for a frameshift, but the criterion requires variant-specific validated assay data that is absent here.

Rule

ATM VCEP v1.5 PS3: Do not use as strong. PS3_Moderate = variant fails to rescue both an ATM-specific feature (e.g., phosphorylation of ATM-specific targets) AND radiosensitivity. PS3_Supporting = variant fails to rescue an ATM-specific feature only; do not use for radiosensitivity-only (not specific to ATM deficiency). Applicable only from VCEP-approved assays (Mitui 2009/PMID 18634022, Barone 2009/PMID 19431188, Scott 2002/PMID 11805335) with the variant confirmed tested in the paper.

Evidence considered

• cspec (PS3 strength rules: no strong; Moderate = fails to rescue both ATM-specific feature and radiosensitivity; Supporting = fails to rescue ATM-specific feature only)
• vcep_clingen_hbop_atm_supplementary_tables_1_and_2_v1 (approved assay calibration; no per-variant result for c.273dup; three approved papers not in bundle)
• vcep_suppl_tables1_pmid_40580951 (SNV-only table verified; c.273dup absent)
• oncokb (Likely Loss-of-function annotation - curated label, not validated functional assay evidence; rejected for PS3)
• PMID:27413114, PMID:30348496, PMID:30553448 (full-text checked; variant not mentioned; rejected for PS3)

Still missing

• No evidence that c.273dup was tested in any VCEP-approved ATM functional assay (Mitui 2009/PMID 18634022, Barone 2009/PMID 19431188, Scott 2002/PMID 11805335); the calibration table does not report per-variant results and these papers are not in the case bundle.
• Sun et al. 2025 (PMID 40580951) Suppl Table S1 is SNV-only; c.273dup is a duplication and has no row, so no functional readout (DeepATM_predicted/Combined_score/Classification) exists for this variant.
• The three OncoKB-sourced publications (PMID 27413114, 30348496, 30553448) contain general ATM biology only and never mention c.273dup/p.Lys92GlufsTer8; they cannot ground PS3.
• If functional data is desired, obtain a variant-specific assay result (e.g., from the three approved papers or a VCEP-recognized equivalent) before assigning PS3 at any strength.

Sources

Reasoning

ATM HBOP VCEP v1.5 (CSPEC doc 639508985) makes PS4 applicable only for case-control studies meeting: p-value <= 0.05 AND (OR, HR, or RR >= 2 OR lower 95% CI >= 1.5). No case-control or large cohort association study reporting the exact variant c.273dup (p.Lys92GlufsTer8) exists in the evidence bundle. ClinVar has no record for this variant (variation_ids: [], submissions: []), gnomAD v2.1/v4.1/gnomAD-Canada report it as absent, and COSMIC has no somatic entry. The three retrieved full-text papers (PMID:27413114, PMID:30348496, PMID:30553448), sourced via OncoKB and triaged as candidate PS4/PP5/PP3 sources, are functional/review papers; full-text search confirms none mentions c.273dup or p.Lys92GlufsTer8, and none contains case-control or case-observation data for this variant. PS4 therefore cannot be met or refuted on available evidence and is left not_assessed (evidence gap, not a demonstrated lack of association).

Rule

Case-control studies; p-value <= 0.05 AND (Odds ratio, hazard ratio, or relative risk >= 2 OR lower 95% CI >= 1.5). (ATM HBOP VCEP v1.5, default strength Pathogenic Strong)

Evidence considered

• clinvar
• cspec
• PMID:27413114
• PMID:30348496
• PMID:30553448
• gnomad_v2
• gnomad_v4
• gnomad_canada
• cosmic

Still missing

• No variant-specific case-control or cohort association study (OR/HR/RR with p<=0.05 per VCEP PS4 rule) reporting c.273dup; such a study is required before PS4 can be applied.

Sources

Reasoning

The ATM VCEP v1.5 defines PP1 for autosomal-recessive A-T (segregation in 1 affected relative = Supporting, 2 = Moderate, >=3 = Strong), but no segregation evidence for NM_000051.4:c.273dup is available in this case: there is no proband phenotype, no affected relatives tested, no family/pedigree data, the variant is absent from ClinVar, and none of the retrieved publications report segregation for this variant. PP1 cannot be applied without variant-specific segregation data.

Rule

ATM VCEP v1.5 CSPEC PP1 (AR condition): Segregation in 1 affected relative = PP1 (Supporting); 2 affected relatives = PP1_Moderate; >=3 affected relatives = PP1_Strong.

Evidence considered

• ATM VCEP v1.5 CSPEC PP1 AR-condition rule: segregation in 1/2/>=3 affected relatives = Supporting/Moderate/Strong.
• No segregation data present in the case bundle; variant absent from ClinVar; 3 retrieved publications are functional-theme papers with no variant or family data.

Still missing

• No family segregation studies or affected-relative testing reported for c.273dup.
• No proband phenotype or inheritance context (A-T vs cancer phenotype) documented in the case.
• Variant absent from ClinVar; no literature case reports describing co-segregation.

Sources

Reasoning

No variant-specific functional assay evidence exists in the case bundle for NM_000051.4:c.273dup (p.Lys92GlufsTer8). The governing ClinGen HBOP ATM VCEP v1.5 framework (CSPEC, precedence rank 1) defines BS3 as: BS3_Moderate when a variant rescues BOTH an ATM-specific feature (e.g., phosphorylation of ATM-specific targets) AND radiosensitivity; BS3_Supporting when it rescues EITHER an ATM-specific feature OR radiosensitivity. BS3 requires well-established functional studies showing no damaging effect, from the same VCEP-approved assay set as PS3 (Mitui 2009/PMID 18634022, Barone 2009/PMID 19431188, Scott 2002/PMID 11805335). The approved-assay calibration table (clingen_hbop_atm_supplementary_tables_1_and_2_v1.xlsx) is explicitly not a per-variant results table - the variant must be confirmed to appear in one of the cited papers, and none of those papers are in this case's bundle with any report of c.273dup. The Sun et al. 2025 screen (PMID 40580951, Suppl Table S1) was checked directly and is SNV-only; c.273dup (a duplication) has no row, so no 'Functional' classification or DeepATM readout can be applied. OncoKB's curated 'Likely Loss-of-function' label is not functional assay evidence and the three OncoKB-sourced papers do not mention this variant. BS3 is therefore not_assessed: there is no assay evidence that this frameshift variant preserves ATM function, and no benign functional readout is available.

Rule

ATM VCEP v1.5 BS3: BS3_Moderate = variant rescues both an ATM-specific feature (e.g., phosphorylation of ATM-specific targets) AND radiosensitivity. BS3_Supporting = variant rescues EITHER an ATM-specific feature OR radiosensitivity. Requires well-established functional studies from VCEP-approved assays (Mitui 2009/PMID 18634022, Barone 2009/PMID 19431188, Scott 2002/PMID 11805335) with the variant confirmed tested in the paper.

Evidence considered

• cspec (BS3 strength rules: Moderate = rescues both ATM-specific feature and radiosensitivity; Supporting = rescues either)
• vcep_clingen_hbop_atm_supplementary_tables_1_and_2_v1 (approved assay calibration; no per-variant result for c.273dup)
• vcep_suppl_tables1_pmid_40580951 (SNV-only table verified; c.273dup absent)
• oncokb (Likely Loss-of-function annotation - no benign functional evidence; rejected for BS3)
• PMID:27413114, PMID:30348496, PMID:30553448 (full-text checked; variant not mentioned; rejected for BS3)

Still missing

• No VCEP-approved assay result demonstrating rescue of an ATM-specific feature and/or radiosensitivity for c.273dup (papers Mitui 2009/PMID 18634022, Barone 2009/PMID 19431188, Scott 2002/PMID 11805335 not in bundle; no per-variant result in the calibration table).
• Sun et al. 2025 (PMID 40580951) Suppl Table S1 does not cover duplications; c.273dup is absent, so no benign functional classification can be drawn from it.
• The OncoKB-sourced publications (PMID 27413114, 30348496, 30553448) do not mention the variant and cannot ground BS3.
• If benign functional evidence is sought, a variant-specific assay showing rescue of ATM-specific targets and/or radiosensitivity would be required before BS3 (Supporting or Moderate) could be applied.

Sources