EGFR c.2573T>G, p.(Leu858Arg)

Reasoning

No ClinGen CSPEC/VCEP or local gene-specific framework exists for EGFR (cspec not found; domain_tables empty), so PS3 is governed by generic ACMG/AMP 2015 (PMID 25741868): well-established functional studies showing a deleterious effect, default strength strong, with SVI functional-evidence calibration (Brnich et al. 2019, PMID 31892348) informing strength. Three independent, well-controlled functional studies of the exact variant c.2573T>G (p.Leu858Arg) consistently demonstrate an activating (gain-of-function) effect: (1) PMID:15118073 (Lynch et al., NEJM 2004) - L858R introduced into full-length EGFR by site-directed mutagenesis and expressed in Cos-7 cells showed EGF-stimulated Y1068 autophosphorylation doubled or tripled versus wild-type and sustained to 3 hours (wild-type signal down-regulated by 15 minutes), plus markedly enhanced gefitinib sensitivity (50% inhibition at ~0.015 uM, complete inhibition at ~0.2 uM, versus 0.1 uM and 2.0 uM for wild-type), with matched protein expression and ligand-dependence controls; (2) PMID:15118125 (Paez et al., Science 2004) - the H3255 lung adenocarcinoma cell line carrying endogenous L858R was ~50-fold more gefitinib-sensitive than wild-type EGFR lines (IC50 40 nM, 72 h survival assay), and 100 nM gefitinib completely inhibited EGFR autophosphorylation and downstream Akt/Erk1/2 phosphorylation in H3255 versus ~100-fold higher concentrations required in wild-type lines; (3) PMID:15329413 (Pao et al., PNAS 2004) - 293T cells transiently expressing L858R showed Y1092 autophosphorylation similar to wild-type but inhibition at ~10-fold lower gefitinib and erlotinib concentrations than wild-type (>=3 independent experiments). PMID:34526717 (Robichaux et al., Nature 2021) corroborates the phenotype clinically, defining L858R as a 'classical' EGFR mutation with markedly better TKI outcomes (TTF HR=1.8 versus atypical mutations, P<0.0001) but adds no new in vitro assay. Readout is biologically relevant: EGFR tyrosine-kinase autophosphorylation and downstream signaling is the direct biochemical function underlying the variant's established disease association (EGFR-driven oncogenesis), and OncoKB independently curates the variant as gain-of-function/Oncogenic. Calibration: three independent assay systems (Cos-7 ectopic, 293T ectopic, endogenous H3255 cell line), each with matched wild-type controls and drug dose-response, all concordant in direction - per SVI functional guidance (PMID 31892348) multiple independent, consistent, well-controlled assays support strong strength; no single-assay downgrade applies. Caveat driving human review: every assay measures oncogenic activation/TKI hypersensitivity (somatic context); no functional study is anchored to an established germline EGFR disease mechanism, and germline ClinVar submissions for this variant are conflicted (1 Pathogenic, 1 Likely pathogenic, 1 VUS, plus expert-panel drug-response). Mechanism-matching of this gain-of-function evidence to the germline phenotype being classified should be confirmed by a human reviewer before the criterion is relied on for a germline Mendelian classification.

Evidence considered

• No ClinGen CSPEC/VCEP or local gene-specific framework exists for EGFR (cspec not found); generic ACMG/AMP 2015 (Richards et al., PMID 25741868) governs PS3: well-established functional studies show a deleterious effect.
• PMID:15118073 (Lynch et al., NEJM 2004): the exact variant (substitution of G for T at nucleotide 2573; L858R) was created by site-directed mutagenesis in full-length EGFR and expressed in Cos-7 cells with wild-type controls; EGF-stimulated Y1068 autophosphorylation was doubled or tripled versus wild-type and sustained up to 3 hours while wild-type activation was down-regulated after 15 minutes; gefitinib produced 50% inhibition at ~0.015 uM and complete inhibition at ~0.2 uM versus ~0.1 uM and ~2.0 uM for wild-type; protein expression was equivalent and no ligand-independent autophosphorylation was seen.
• PMID:15118125 (Paez et al., Science 2004): H3255 lung adenocarcinoma cell line carrying endogenous L858R (confirmed in its primary tumor) was ~50x more sensitive to gefitinib than wild-type EGFR lines (IC50 40 nM, 72-hour survival assay); 100 nM gefitinib completely inhibited EGFR autophosphorylation and downstream Akt/Erk1/2 phosphorylation in H3255, with comparable inhibition in wild-type lines only at ~100x higher concentrations.
• PMID:15329413 (Pao et al., PNAS 2004): 293T cells transiently expressing L858R EGFR showed Y1092 autophosphorylation similar to wild-type but were inhibited at ~10-fold lower gefitinib and erlotinib concentrations than wild-type EGFR (>=3 independent experiments).
• PMID:34526717 (Robichaux et al., Nature 2021): defines L858R as a 'classical' EGFR mutation with markedly improved TKI outcomes (time-to-treatment-failure HR=1.8 for classical versus atypical, P<0.0001); corroborates the activating/TKI-sensitive phenotype but reports no new L858R-specific in vitro assay.
• Strength calibration: generic ACMG/AMP 2015 assigns PS3 strong by default; under SVI functional-evidence guidance (Brnich et al., Genome Medicine 2019, PMID 31892348), multiple independent, well-controlled assays with a biologically relevant readout (here EGFR kinase autophosphorylation and TKI sensitivity across Cos-7, 293T, and the endogenous H3255 line) support retaining strong rather than downgrading.
• Biological relevance: EGFR tyrosine-kinase autophosphorylation and downstream signaling is the direct biochemical function of the protein relevant to the variant's established disease association (activating EGFR-driven tumorigenesis); OncoKB independently curates L858R as gain-of-function with an Oncogenic label.
• Context caveat: the assays establish an activating/gain-of-function effect in the somatic oncogenic context; germline ClinVar submissions for this variant are conflicted (1 Pathogenic, 1 Likely pathogenic, 1 VUS, plus ClinPGx expert-panel drug response), and no germline EGFR disease mechanism is established, so germline applicability of PS3 was flagged for human review.

Still missing

• All functional assays were performed in the somatic/oncogenic context (EGF-driven kinase activation, TKI hypersensitivity); none is anchored to an established germline EGFR disease mechanism, so mechanism-matching of this gain-of-function evidence to the germline interpretation target needs human confirmation.
• The assay papers do not report formal validation statistics against large reference sets of known benign/pathogenic EGFR variants; strong strength therefore rests on independence and concordance of three well-controlled assays rather than documented per-assay validation metrics.

Sources

Reasoning

No applicable EGFR-specific VCEP/CSPEC or gene-specific framework was available, so the generic ACMG/AMP population rules were used. The variant is absent (0 alleles) from the default all-comers gnomAD v2.1 and gnomAD v4.1 records, and absence is corroborated by the gnomAD v2.1 non-cancer (exome-only figures) and gnomAD v3.1 non-cancer (genome-only) subsets and by gnomAD-Canada v1.0. An observed allele frequency of 0 is well below this laboratory's generic PM2 rarity threshold (<0.1%) and is consistent with absence in population controls for a disease-associated variant. The variant's high recurrence as a somatic cancer driver (COSMIC, n=2857) does not confound this assessment, because no germline allele is observed in any population source. PM2 is met; under the generic ACMG/AMP 2015 framework (no VCEP modification retrieved), PM2 is applied at moderate strength.

Rule

Generic ACMG/AMP PM2: apply when the variant is absent or at extremely low frequency (<0.1% per this laboratory's generic gnomAD operating threshold) in appropriate population databases; applied at moderate strength under the ACMG/AMP 2015 framework.

Evidence considered

• The default all-comers gnomAD v2.1 record reports the variant as absent (0 alleles).
• The default all-comers gnomAD v4.1 record reports the variant as absent (0 alleles).
• gnomAD-Canada v1.0 reports the variant as absent.
• The gnomAD v2.1 non-cancer (exome) and gnomAD v3.1 non-cancer (genome) subsets report absence, corroborating rarity in non-cancer populations; no applicable VCEP required substituting these subsets for the default all-comers sources.
• Generic ACMG/AMP PM2 (Richards et al. 2015, PMID:25741868) supports PM2 when a variant is absent or extremely rare in appropriate population databases; this laboratory's generic gnomAD rarity threshold is <0.1%, and the observed frequency is 0.
• Under the generic ACMG/AMP 2015 framework with no VCEP-specific modification, PM2 is applied at moderate strength.

Sources

Reasoning

PM1 requires location in a mutational hot spot and/or a critical, well-established functional domain without benign variation. vcep_materials.domain_tables is empty for EGFR, so no VCEP-approved critical-domain table exists against which residue 858 could be matched, and no ClinGen CSPEC governs this case (generic ACMG/AMP 2015 fallback). Residue 858 does lie in the EGFR tyrosine kinase domain activation loop immediately adjacent to the highly conserved DFG motif (residues 855-857), a residue described as nearly invariant across protein kinases (Paez et al. 2004; Pao et al. 2004), and somatic EGFR mutations are reported to cluster near the ATP cleft of the tyrosine kinase domain (Lynch et al. 2004); CancerHotspots flags L858 as a statistically significant somatic mutational hotspot and COSMIC records 2,857 somatic occurrences of L858R. However, all of this hotspot/domain evidence derives from somatic gain-of-function cancer biology rather than from an established germline disease mechanism for EGFR: no authoritative specification in this case establishes that the EGFR kinase-domain activation loop is a critical functional domain for a germline phenotype, and no domain-level data demonstrate depletion of benign missense variation (only this exact variant is absent from gnomAD v2.1/v4.1). Under the governing generic framework, PM1 is therefore not met. Human review is recommended because the somatic domain/hotspot evidence for residue 858 is genuine and well-documented, and a reviewer may reconsider PM1 if an EGFR germline context in which this domain is disease-critical is established.

Rule

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

Evidence considered

• vcep_materials.domain_tables is empty for EGFR: no VCEP-approved critical functional domain list exists for this gene in this case, so no authoritative domain entry can be matched for residue 858.
• Pao et al. 2004 (PMID 15329413): L858R 'occurs adjacent to the highly conserved DFG motif (amino acids 855-857) in the activation loop of the kinase', and L858R accounted for 20/49 (41%) of reported TK-domain hotspot mutations in that analysis.
• Paez et al. 2004 (PMID 15118125): L858R is 'adjacent to the highly conserved DFG motif in the activation loop' and 'the mutated residues are nearly invariant in all protein kinases', with analogous residues (G463, L596) somatically mutated in B-Raf.
• Lynch et al. 2004 (PMID 15118073): the recurrent somatic EGFR mutations, including L858R (Patients 5 and 6), are interpreted as clustered near the ATP cleft of the tyrosine kinase domain.
• CancerHotspots (evidence.json) reports residue L858 as lying in a statistically significant somatic mutational hotspot; COSMIC (evidence.json) records COSV51765161 with 2,857 somatic occurrences - both are somatic-cancer resources, not germline disease-domain evidence.
• gnomAD v2.1 and v4.1 report this exact variant absent; variant-level absence is the only population evidence available and does not by itself establish domain-level 'without benign variation'.
• Generic ACMG/AMP 2015 framework (generic_acmg_combination_rules) applies; no CSPEC/VCEP and no local gene framework are available for EGFR in this case.

Still missing

• No VCEP/CSPEC-approved critical functional domain table for EGFR (vcep_materials.domain_tables empty) against which residue 858 could be anchored for PM1.
• No domain-level benign-missense-depletion (constraint) data for the EGFR kinase domain; only variant-level absence from gnomAD v2.1/v4.1 is available.
• Germline disease relevance of the somatic EGFR TK-domain hotspot evidence is unresolved (no authoritative germline gene-disease mechanism in this case).

Sources

Reasoning

Governing rule for this group: PP5 applies ONLY for an exact-variant ClinVar expert-panel classification of Pathogenic or Likely pathogenic (supporting strength); ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions must never trigger PP5. ClinVar VCV000016609 matches the exact variant (NM_005228.5(EGFR):c.2573T>G, p.Leu858Arg; input transcript NM_005228.4). The ClinVar aggregate classification is 'drug response' with review status 'reviewed by expert panel' (3 stars), but the sole expert-panel submission, ClinPGx SCV000268169 (2021-03-24, origin germline), is classified 'drug response' for condition 'gefitinib response - Efficacy' - a pharmacogenomic efficacy assertion, NOT a germline-disease Pathogenic/Likely pathogenic expert-panel classification. The Pathogenic submissions (Mendelics SCV001137371, criteria provided single submitter, origin unknown; Pfizer-University of Granada SCV003806315, somatic) and the Likely pathogenic submission (Central South University SCV004042724, somatic, no assertion criteria provided) are from ordinary laboratories, and the somatic-origin and no-criteria submissions are excluded by the rule in any event. Because no exact-variant ClinVar expert-panel Pathogenic/Likely pathogenic classification exists, PP5's trigger condition is determinably absent: not_met.

Rule

PP5 (supporting): exact-variant ClinVar expert-panel Pathogenic/Likely pathogenic classification required; ordinary laboratory, aggregate, and non-expert assertions excluded

Evidence considered

• Group governing rule: PP5 applies only for an exact-variant ClinVar expert-panel classification of Pathogenic or Likely pathogenic (supporting); ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions never trigger PP5.
• ClinVar VCV000016609 (exact variant NM_005228.5(EGFR):c.2573T>G p.Leu858Arg, 12 submissions/10 submitters) has aggregate classification 'drug response' with 3-star 'reviewed by expert panel' status.
• The only expert-panel submission is ClinPGx SCV000268169 (reviewed by expert panel, 2021-03-24, origin germline), classified 'drug response' for 'gefitinib response - Efficacy' - a pharmacogenomic drug-efficacy assertion, not a Pathogenic/Likely pathogenic germline-disease classification, so it does not satisfy PP5.
• Pathogenic/Likely pathogenic submissions are ordinary-laboratory or somatic assertions (Mendelics SCV001137371 Pathogenic origin unknown; Pfizer-University of Granada SCV003806315 Pathogenic somatic; Central South University SCV004042724 Likely pathogenic somatic, no assertion criteria), which the governing rule excludes from PP5.
• The aggregate 3-star 'drug response' ClinVar label was not used as a proxy for an expert-panel Pathogenic classification.

Sources

Reasoning

No applicable EGFR-specific VCEP/CSPEC or gene-specific framework was available, so the generic ACMG/AMP population rules were used. BA1 is a stand-alone benign criterion requiring a population allele frequency far exceeding that expected for the disorder (this laboratory's generic gnomAD operating threshold for a non-VCEP gene: >1%). The variant is absent (0 alleles) from the default all-comers gnomAD v2.1 and gnomAD v4.1 records, and absence is corroborated by the gnomAD v2.1 non-cancer (exome) and gnomAD v3.1 non-cancer (genome) subsets and by gnomAD-Canada v1.0. An observed allele frequency of 0 cannot approach the BA1 threshold, so BA1 is not met.

Rule

Generic ACMG/AMP BA1: apply only when the population allele frequency is above the stand-alone benign threshold (>1% per this laboratory's generic gnomAD operating threshold); an absent variant cannot meet BA1.

Evidence considered

• The default all-comers gnomAD v2.1 record reports the variant as absent (0 alleles).
• The default all-comers gnomAD v4.1 record reports the variant as absent (0 alleles).
• gnomAD-Canada v1.0 reports the variant as absent.
• The gnomAD v2.1 non-cancer and gnomAD v3.1 non-cancer subset records also report absence; no applicable VCEP specified a non-cancer or exome-only population source, so the all-comers records remain the default.
• Generic ACMG/AMP BA1 (Richards et al. 2015, PMID:25741868) requires a stand-alone benign population frequency; this laboratory's generic gnomAD operating threshold is >1%, and the observed frequency of 0 is far below it.

Sources

Reasoning

No applicable EGFR-specific VCEP/CSPEC or gene-specific framework was available, so the generic ACMG/AMP population rules were used. BS1 requires an allele frequency greater than expected for the disorder (this laboratory's generic gnomAD operating threshold for a non-VCEP gene: >0.3%). The variant is absent (0 alleles) from the default all-comers gnomAD v2.1 and gnomAD v4.1 records, with corroborating absence in the gnomAD v2.1 non-cancer (exome) and gnomAD v3.1 non-cancer (genome) subsets and in gnomAD-Canada v1.0. With no observed population frequency, BS1 is not met.

Rule

Generic ACMG/AMP BS1: apply when the allele frequency is greater than expected for the disorder (>0.3% per this laboratory's generic gnomAD operating threshold); an absent variant cannot satisfy this benign-frequency criterion.

Evidence considered

• The default all-comers gnomAD v2.1 record reports the variant as absent (0 alleles).
• The default all-comers gnomAD v4.1 record reports the variant as absent (0 alleles).
• gnomAD-Canada v1.0 reports the variant as absent.
• The gnomAD v2.1 non-cancer and gnomAD v3.1 non-cancer subset records also report absence.
• Generic ACMG/AMP BS1 (Richards et al. 2015, PMID:25741868) requires a frequency greater than expected for the disorder; this laboratory's generic gnomAD operating threshold is >0.3%, and no frequency above it is observed because the variant is absent.

Sources

Reasoning

No applicable EGFR-specific VCEP/CSPEC or gene-specific framework was available, so the generic ACMG/AMP population rules were used. BS2 is a benign criterion requiring the variant to be observed in healthy adults (heterozygotes for a dominant disorder, homozygotes for a recessive disorder) with early-onset full penetrance expected. The variant is absent from the default all-comers gnomAD v2.1 and gnomAD v4.1 records and from gnomAD-Canada v1.0, so no healthy adult carrier or homozygote observation is documented; absence of the variant cannot affirmatively establish BS2. The criterion is therefore not met.

Rule

Generic ACMG/AMP BS2: apply when the variant is observed in healthy adults (heterozygotes or, where relevant, homozygotes) for a disorder expected to be fully penetrant at an early age; absence alone is insufficient.

Evidence considered

• The available gnomAD v2.1 and v4.1 records report the variant as absent and do not document healthy adult carriers or homozygotes.
• gnomAD-Canada v1.0 reports the variant as absent.
• The gnomAD v2.1 non-cancer and gnomAD v3.1 non-cancer subsets also report absence, providing no unaffected-carrier observations.
• Generic ACMG/AMP BS2 (Richards et al. 2015, PMID:25741868) requires positive observation of the variant in healthy individuals; absence from population databases is not equivalent to such an observation and does not establish BS2.

Still missing

• No individual-level healthy-carrier or healthy-homozygote observations with age, phenotype, and zygosity were available to positively support BS2.

Sources

Reasoning

Generic ACMG/AMP 2015 (PMID 25741868) BS3 requires well-established functional studies showing no damaging effect on protein function or splicing. The well-established functional evidence for this exact variant shows the opposite: three independent, well-controlled studies (PMID:15118073 Cos-7 expression; PMID:15118125 endogenous H3255 cell line; PMID:15329413 293T expression) consistently demonstrate an activating/damaging gain-of-function effect - increased and sustained EGF-induced EGFR autophosphorylation (2-3x wild-type, persisting to 3 h versus 15 min for wild-type) and enhanced sensitivity to EGFR TKI inhibition (~10-50x lower effective concentrations than wild-type). No published functional study demonstrates loss or absence of a damaging effect. The splice arm of BS3 is also not met: SpliceAI predicts no significant splice impact (max delta score 0.013) for this deep-exonic missense, but an in-silico splicing prediction is not a well-established functional study and cannot satisfy BS3, and it does not counteract the demonstrated damaging protein-level effect. BS3 is therefore not met.

Evidence considered

• Generic ACMG/AMP 2015 (Richards et al., PMID 25741868) BS3 requires well-established functional studies showing no damaging effect on protein function or splicing.
• The well-established functional studies of the exact variant (PMID:15118073, PMID:15118125, PMID:15329413) demonstrate a damaging activating effect (2-3x increased and sustained EGF-induced EGFR autophosphorylation; ~10-50x enhanced gefitinib/erlotinib sensitivity versus wild-type), i.e., the opposite of the no-damaging-effect finding BS3 requires; see PS3 evidence_used for assay details.
• SpliceAI (registry key spliceai) predicts no significant splice impact for this variant (max delta score 0.013); however this is an in-silico prediction rather than a well-established functional study, so it neither satisfies BS3 nor outweighs the demonstrated damaging protein-level effect.

Still missing

• No experimental (RNA/cDNA-based) splicing assay is available for this variant; only SpliceAI in-silico output (max delta 0.013) exists. This does not change the verdict, since BS3 is already not met on protein-function grounds.

Sources

Reasoning

BP2 (Richards et al. 2015, PMID 25741868) requires observing the variant in trans with a pathogenic variant for a fully dominant disease/disorder, or in cis with a pathogenic variant in any inheritance pattern (supporting, benign direction). Under the generic ACMG/AMP 2015 fallback governing this case (no EGFR CSPEC/VCEP), the BP2-defining observation is absent from all available proband-level data: EGFR c.2573T>G (p.Leu858Arg) appears as the sole EGFR alteration in tested probands - heterozygous and somatic, with matched normal tissue wild type (PMID:15118073, PMID:15118125, PMID:15329413) - and no second germline EGFR variant or phase determination is reported anywhere. The single co-occurrence of L858R with a second EGFR mutation (R776C) in one erlotinib-sensitive tumor (patient E3, PMID:15329413) was considered and rejected as BP2 evidence: it is a somatic tumor finding without resolved germline phase, R776C has no established pathogenicity, BP2 concerns germline allelic configuration, and applying it would conflict with the variant's documented activating gain-of-function effect. Because the required in-trans (fully dominant) or in-cis observation of a pathogenic variant was never made in any proband, BP2 is not met. Residual limitation: no dedicated germline phase or family testing data exist in this case, so BP2 would be reassessable if a second germline EGFR pathogenic variant with resolved phase were documented.

Rule

BP2 (Richards et al. 2015, 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. Requires a phase-known observation of a second pathogenic variant.

Evidence considered

• No ClinGen CSPEC/VCEP or local gene-specific criteria specification applies to EGFR (vcep_materials.json: cspec found=false; final_classification_framework found=false, precedence rank 3); the generic ACMG/AMP 2015 fallback governs (generic_acmg_combination_rules, source basis Richards et al. 2015, PMID 25741868).
• Richards et al. 2015 (PMID 25741868) defines BP2 as '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' (supporting, benign direction); the criterion turns entirely on a phase-known observation of a second pathogenic variant.
• PMID:15118073 (Lynch et al., NEJM 2004): whole EGFR coding sequence was analyzed in gefitinib responders and non-responders; L858R in Patients 5 and 6 was the only EGFR alteration reported (heterozygous, somatic), providing no in-trans or in-cis observation of a second pathogenic variant.
• PMID:15118125 (Paez et al., Science 2004): L858R was a heterozygous somatic missense mutation absent from matched normal lung tissue ('not present in the DNA from normal lung tissue from the same patients'); no second EGFR variant or phase data were reported.
• PMID:15329413 (Pao et al., PNAS 2004): the only two-mutation EGFR tumor in the case evidence is erlotinib-sensitive patient E3, carrying L858R plus a second somatic mutation (R776C, nucleotide 2326 C>T, exon 20); matched normal tissue showed only wild-type sequence. This somatic co-occurrence was considered and rejected as BP2 evidence because germline phase is unresolved, R776C is not established as pathogenic, and BP2 targets germline allelic configuration - not somatic tumor co-mutation patterns.
• ClinVar germline submissions (Ambry germline VUS; ClinPGx germline drug-response for gefitinib efficacy) record no second EGFR variant or phase information and were not used to infer this criterion from any final classification.
• The BP2-defining observation (variant in trans with a pathogenic variant in a fully dominant disorder, or in cis with a pathogenic variant) is absent from every available proband observation; BP2 is not met.

Still missing

• No germline phase/family testing data are available; the absence of an in-trans (fully dominant) or in-cis pathogenic-variant observation rests on tumor-tissue sequencing and ClinVar germline records, not on dedicated phase analysis.

Sources

Reasoning

Governing rule for this group: BP6 applies ONLY for an exact-variant ClinVar expert-panel classification of Benign or Likely benign (supporting strength); ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions must never trigger BP6. For ClinVar VCV000016609 (exact variant NM_005228.5(EGFR):c.2573T>G, p.Leu858Arg), NO expert-panel Benign or Likely benign classification exists, and in fact no Benign or Likely benign submission of any kind is present among the 12 submissions (significances recorded: Pathogenic x2 [Mendelics, Pfizer-University of Granada], drug response x3 [OMIM x2, LMM, plus ClinPGx expert panel], Likely pathogenic x1 [Central South University], Tier I - Strong x3 [CIViC x2, Wagner Lab], Oncogenic x1 [Rigshospitalet], Uncertain significance x1 [Ambry]). The sole expert-panel submission (ClinPGx SCV000268169) is 'drug response' (gefitinib efficacy), which is not a Benign/Likely benign classification. Because the required exact-variant expert-panel Benign/Likely benign classification is determinably absent, BP6 is not met.

Rule

BP6 (supporting): exact-variant ClinVar expert-panel Benign/Likely benign classification required; ordinary laboratory, aggregate, and non-expert assertions excluded

Evidence considered

• Group governing rule: BP6 applies only for an exact-variant ClinVar expert-panel classification of Benign or Likely benign (supporting); ordinary laboratory submissions, aggregate ClinVar labels, and non-expert assertions never trigger BP6.
• ClinVar VCV000016609 (exact variant, 12 submissions/10 submitters) contains no Benign or Likely benign submission of any type; submission significances are Pathogenic, Likely pathogenic, Uncertain significance, drug response, Oncogenic, and Tier I - Strong.
• The only expert-panel submission, ClinPGx SCV000268169, is classified 'drug response' for 'gefitinib response - Efficacy', which is not a Benign/Likely benign classification and does not satisfy BP6.
• The aggregate ClinVar 'drug response' label was not used as a benign assertion for BP6.

Sources

Reasoning

PS1 (Richards et al. 2015, PMID 25741868) requires that the same amino-acid change as the variant under evaluation - here p.Leu858Arg - has previously been established as pathogenic through a DIFFERENT nucleotide change. The variant under evaluation is itself the canonical EGFR c.2573T>G (CTG>CGG) change, and every source consulted documents only this identical nucleotide change: all ClinVar submissions for variation 16609 are exact matches for c.2573T>G (including the ClinPGx expert-panel drug-response assertion and the single-submitter Pathogenic/Likely pathogenic/VUS submissions), and the full-text papers describing L858R (Lynch et al. 2004; Paez et al. 2004; Pao et al. 2004) report only the nucleotide-2573 T>G substitution. The PM5 candidate-collection artifact likewise recorded zero same-residue comparator variants. Because no different-nucleotide change producing p.Leu858Arg with an established pathogenic classification was identified in any consulted source, PS1 cannot be assessed; identical-nucleotide submissions do not satisfy PS1's different-nucleotide requirement.

Rule

PS1: Same amino acid change as a previously established pathogenic variant regardless of nucleotide change (Richards et al. 2015, PMID 25741868; generic ACMG/AMP fallback)

Evidence considered

• ClinVar variation 16609 contains only submissions for the identical nucleotide change NM_005228.5(EGFR):c.2573T>G (p.Leu858Arg); none documents a different nucleotide change producing p.Leu858Arg.
• The PM5 candidate-collection artifact (pm5_candidates) recorded 0 same-residue candidates (total_candidates_seen 0; mode 'unknown') and returned no comparator variants usable for same-residue assessment.
• Fetched full texts (Lynch et al. 2004, PMID 15118073; Paez et al. 2004, PMID 15118125; Pao et al. 2004, PMID 15329413) describe only the CTG>CGG change at nucleotide 2573 for L858R; no alternate-nucleotide p.Leu858Arg variant is reported in any of them.
• Generic ACMG/AMP 2015 framework (generic_acmg_combination_rules, Richards et al. PMID 25741868) applies because no CSPEC/VCEP governs this case; PS1's definition requires the same amino-acid change arising from a different nucleotide change.

Still missing

• No curated enumeration of alternate nucleotide changes at EGFR codon 858 (e.g., a gene-wide ClinVar EGFR query) with established pathogenic classifications; all ClinVar submissions found are the identical c.2573T>G change, which cannot serve as a PS1 comparator.

Sources

Reasoning

PS2 (generic ACMG/AMP 2015, PMID 25741868) requires a de novo germline occurrence of the variant in a proband with the disease, with maternity and paternity confirmed. No proband-level data exist to evaluate this: the case evidence contains no proband germline genotype, no parental testing, and no maternity/paternity confirmation for any carrier of EGFR c.2573T>G (p.Leu858Arg). The ClinVar record (variation 16609) has 12 submissions, none asserting de novo evidence; the only germline-origin submission (Ambry SCV005992183, VUS) cites only an in silico (PP3) assertion. The published occurrences of this variant in the reviewed full-text literature are heterozygous somatic tumor mutations with wild-type matched normal tissue (PMID:15118073, Patients 5 and 6; PMID:15118125; PMID:15329413), demonstrating that reported cases arose somatically in tumors rather than as germline de novo events; somatic tumor data cannot satisfy PS2. Evidence is insufficient, so PS2 is not assessed.

Rule

Generic ACMG/AMP 2015 (Richards et al., Genet Med 2015;17:405-424, PMID 25741868), applied because no ClinGen CSPEC/VCEP specification exists for EGFR: PS2 - de novo occurrence (maternity and paternity confirmed) in a patient with the disease and no family history (strong pathogenic).

Evidence considered

• ClinVar variation 16609: 12 submissions under conditions including somatic NSCLC/TKI-response and 'Hereditary cancer-predisposing syndrome'; no submission asserts de novo or parental-testing evidence, and the only germline-origin submission (SCV005992183, Ambry, VUS, 2025-03-10) lists only an in silico (PP3) assertion.
• PMID:15118073 (Lynch et al., NEJM 2004): L858R tumors in gefitinib responders (Patients 5 and 6) were heterozygous, and matched normal tissue available for Patients 1, 4, 5, and 6 showed only wild-type sequence, indicating the mutations arose somatically during tumor formation - not as germline de novo events.
• PMID:15118125 (Paez et al., Science 2004): the first three EGFR-mutant tumors (all L858R; S0361, S0388, S0389) had mutations absent from normal lung tissue DNA, and all 16 EGFR mutations in the 119-tumor cohort were heterozygous in tumor DNA with wild-type paired normal lung tissue, confirming somatic origin.
• PMID:15329413 (Pao et al., PNAS 2004): L858R tumors in gefitinib/erlotinib-responsive and never-smoker/smoker cohorts had only wild-type sequence in corresponding normal tissue where available (patients E1-E3, E5; specimens 5, 65, 134), indicating somatic origin.
• Keyword review of all five fetched full-text publications found no family, pedigree, germline-inheritance, or segregation content for this variant.

Still missing

• Parental genotypes (maternity and paternity confirmation) for any proband carrying EGFR c.2573T>G
• A germline (not tumor) testing result documenting a de novo EGFR c.2573T>G event in a proband
• Proband phenotype-consistency documentation required for de novo assessment

Sources

Reasoning

Governing framework: no ClinGen CSPEC/VCEP or gene-specific specification exists for EGFR germline interpretation in this case (cspec found=false; vcep_materials reports 'no cspec url available'; framework_mode=generic_acmg), so generic ACMG/AMP 2015 (Richards et al., PMID 25741868) PS4 applies: 'Prevalence of the variant in affected individuals is significantly increased compared with the prevalence in controls.' The generic fallback rules for this case define PS4 qualitatively and specify NO numeric odds-ratio threshold, so there is no governing VCEP threshold whose comparison operator (>=, <=, >, <) could be quoted or applied; the verdict rests on study-design eligibility, not threshold arithmetic. Evidence reviewed for case-control enrichment of the exact variant (EGFR c.2573T>G, p.Leu858Arg): (1) PMID 15118073 (Lynch et al., NEJM 2004) found EGFR kinase-domain mutations in 8/9 gefitinib-responsive NSCLC patients vs 0/7 non-responders (P<0.001), with L858R in 2 responders (Patients 5, 6); (2) PMID 15118125 (Paez et al., Science 2004) found kinase-domain mutations in 5/5 gefitinib-responsive US patients vs 0/4 progressors (P=0.0027), with L858R in one gefitinib-sensitive adenocarcinoma; (3) PMID 15329413 (Pao et al., PNAS 2004) found TK-domain mutations in 7/10 gefitinib-sensitive vs 0/8 refractory tumors (P=0.004) and 5/7 erlotinib-sensitive vs 0/10 refractory (P=0.003), and L858R in 5/15 (33%) never-smoker vs 4/81 (5%) smoker NSCLCs (P=0.0001 for mutation frequency); (4) PMID 15897572 (Chou et al., 2005) reported L858R in 9/33 EGFR-mutated gefitinib-treated patients, with EGFR mutation presence (not the single variant) predicting disease control and longer PFS/OS. COSMIC documents the variant in 2,857 somatic cancer samples (COSV51765161) and Cancer Hotspots lists L858 as a statistically significant hotspot. ALL of these comparisons are somatic tumor prevalence or pharmacogenomic (TKI-response) enrichment series; none is a germline case-control study of the variant in affected vs unaffected individuals for a hereditary phenotype, no odds ratio versus an unaffected control group can be computed, and the variant is absent from gnomAD v2.1/v4.1/gnomAD-Canada population controls (control-side count only, no affected-side cohort). Additionally, the case record contains no proband phenotype, so the affected population a PS4 comparison would reference is undefined. The ClinVar Pathogenic/Likely pathogenic submissions and OncoKB 'Oncogenic' label are not case-control evidence and were not used to infer PS4. Evidence is therefore insufficient to apply PS4 under the generic germline ACMG definition: status not_assessed, with the gaps below.

Rule

ACMG/AMP 2015 PS4 (generic fallback, PMID 25741868): prevalence of the variant in affected individuals significantly increased vs controls; no VCEP/CSPEC numeric threshold in force for this case

Evidence considered

• Generic ACMG/AMP 2015 PS4 (Richards et al., PMID 25741868): 'Prevalence of the variant in affected individuals is significantly increased compared with the prevalence in controls'; the generic fallback rules for this case specify no numeric PS4 threshold and no VCEP/CSPEC source governs EGFR, so no comparison operator could be quoted from a governing VCEP text for this criterion.
• PMID 15118073 (Lynch et al., NEJM 2004): EGFR kinase-domain mutations in 8/9 gefitinib-responsive NSCLC patients vs 0/7 non-responders (P<0.001); L858R specifically in two responders (Patients 5 and 6); somatic tumor sequencing with matched normal tissue - drug-response enrichment, not germline case-control prevalence.
• PMID 15118125 (Paez et al., Science 2004): EGFR kinase-domain mutations in all 5/5 gefitinib-responsive US patients vs 0/4 progressors (P=0.0027); one gefitinib-sensitive lung adenocarcinoma carried L858R (also in cell line H3255); somatic tumor cohort design.
• PMID 15329413 (Pao et al., PNAS 2004): TK-domain mutations in 7/10 gefitinib-sensitive vs 0/8 refractory tumors (P=0.004) and 5/7 erlotinib-sensitive vs 0/10 refractory (P=0.003); in untreated resected NSCLCs, L858R in 5/15 never-smoker adenocarcinomas vs 4/81 former/current-smoker tumors (P=0.0001); somatic tumor epidemiology without unaffected germline controls.
• PMID 15897572 (Chou et al., 2005): L858R in 9/33 EGFR-mutated gefitinib-treated Taiwanese advanced NSCLC patients; EGFR mutation presence (not the exact variant) predicted disease control and tumor response; flagged as a PS4 candidate by the literature pass but contains no case-control comparison for the exact variant.
• COSMIC reports the variant in 2,857 somatic cancer samples (COSV51765161) and Cancer Hotspots reports L858 as a statistically significant hotspot; these document somatic recurrence only and were not converted into germline case-control evidence for PS4.
• The variant is absent from gnomAD v2.1, gnomAD v4.1, and gnomAD-Canada v1.0 population controls; this establishes the control-side allele count (0) but, without an affected-side case-control cohort, cannot demonstrate PS4 enrichment.
• ClinVar VCV000016609 aggregate classification is 'drug response' with expert-panel (ClinPGx) pharmacogenomic review; per the group rule, no criterion may be inferred from another source's final classification, so the Pathogenic/Likely pathogenic laboratory submissions and OncoKB 'Oncogenic' label were not used as PS4 case-control evidence.

Still missing

• No case-control or cohort-prevalence study comparing EGFR c.2573T>G (p.Leu858Arg) frequency in affected germline individuals versus unaffected controls exists for any defined phenotype; available enrichment data (PMID 15118073, 15118125, 15329413, 15897572) are somatic tumor and pharmacogenomic TKI-response series with no unaffected control arm.
• No proband phenotype or clinical indication is recorded in the case, leaving the affected population that a PS4 comparison would reference undefined.
• No CSPEC/VCEP governs EGFR, and the generic ACMG/AMP 2015 fallback defines PS4 qualitatively with no numeric odds-ratio threshold, so no threshold-based met/not_met determination is possible even in principle.

Sources

Reasoning

PM5 (Richards et al. 2015) applies to a novel missense change at an amino acid residue where a different missense change has already been determined pathogenic. p.Leu858Arg is the missense change under evaluation, and no source consulted provides a different missense change at residue 858 with an established pathogenic classification: ClinVar variation 16609 contains only submissions for the identical c.2573T>G change; the PM5 candidate-collection artifact recorded zero same-residue candidates (harvest skipped, mode 'unknown', with an explicit instruction that PM5 must not be applied from that artifact alone); and the reviewed full-text papers describe alternate amino-acid changes only at other residues (L861Q at codon 861 in Lynch et al. 2004; acquired-resistance substitutions L718V, V765L, C797S at other codons in Robichaux et al. 2021), which are not same-residue PM5 comparators. The required comparator evidence is therefore absent, so PM5 is not assessed; the gap is the lack of curated same-residue different-missense data (e.g., a gene-wide enumeration of codon-858 EGFR missense variants and their classifications).

Rule

PM5: Novel missense change at an amino acid residue where a different missense change determined to be pathogenic has been seen before (Richards et al. 2015, PMID 25741868; generic ACMG/AMP fallback)

Evidence considered

• ClinVar variation 16609: all submissions are for the identical c.2573T>G (p.Leu858Arg) change; no different missense change at residue 858 with a pathogenic classification is represented.
• The PM5 candidate-collection artifact (pm5_candidates) recorded 0 same-residue candidates and 0 expert-panel LP/P candidates; its recommendation level is 'not_applicable' and it states PM5 must not be applied from that artifact alone.
• Lynch et al. 2004 (PMID 15118073): at exon 21 only L858R (Patients 5 and 6) and L861Q (Patient 7) are described; L861Q is at codon 861, a different residue, so it is not a PM5 comparator for codon 858.
• Paez et al. 2004 (PMID 15118125) and Pao et al. 2004 (PMID 15329413): describe only the CTG>CGG L858R change at codon 858; no alternate missense at residue 858 is reported.
• Generic ACMG/AMP 2015 framework (generic_acmg_combination_rules) applies; PM5 requires a different missense change at the same residue with an established pathogenic determination, which no consulted source provides.

Still missing

• No same-residue (codon 858) different-missense variant with an established pathogenic classification identified; the pm5_candidates harvest returned 0 candidates (mode 'unknown') and ClinVar variation 16609 contains only the identical c.2573T>G change.

Sources

Reasoning

PM6 (generic ACMG/AMP 2015, PMID 25741868) requires an assumed de novo germline occurrence of the variant in a proband, without confirmation of paternity and maternity. No proband germline observation, trio result, or family-history documentation that would support an assumed de novo event is present in the case evidence, and no ClinVar submission asserts PM6-type evidence. The reviewed full-text literature instead documents EGFR p.Leu858Arg as a somatic tumor mutation: matched normal tissue from affected patients showed only wild-type sequence (PMID:15118073, PMID:15118125, PMID:15329413), which is the opposite of a germline event, even an unconfirmed one. Evidence is insufficient, so PM6 is not assessed.

Rule

Generic ACMG/AMP 2015 (Richards et al., Genet Med 2015;17:405-424, PMID 25741868): PM6 - assumed de novo occurrence, without confirmation of paternity and maternity (moderate pathogenic).

Evidence considered

• ClinVar variation 16609: none of the 12 submissions reports an assumed de novo occurrence or parental genotypes for this variant; the single germline-origin submission is a VUS with an in silico assertion only.
• PMID:15118073 (Lynch et al., NEJM 2004): L858R in gefitinib-responsive patients arose somatically (matched normal tissue wild-type), i.e., not a germline event eligible for assumed-de-novo reasoning.
• PMID:15118125 (Paez et al., Science 2004): L858R mutations were absent from paired normal lung tissue DNA, confirming somatic origin in the tumor.
• PMID:15329413 (Pao et al., PNAS 2004): L858R tumors showed only wild-type sequence in corresponding normal tissue, confirming somatic origin.
• Keyword review of all five fetched full-text publications found no germline-inheritance or parental-testing content for this variant.

Still missing

• A proband germline test result consistent with de novo occurrence in the absence of parental confirmation
• Family-history documentation (e.g., unaffected parents) supporting an assumed de novo event
• Any trio or single-parent genotype data for a proband carrying this variant

Sources

Reasoning

PP1 (generic ACMG/AMP 2015, PMID 25741868) requires co-segregation of the variant with disease in multiple affected family members in a gene definitively known to cause the disease. The case evidence contains no pedigree or family testing of any kind, and a keyword review of all five fetched full-text publications found no family, pedigree, germline-inheritance, or segregation content. The clinical context of EGFR p.Leu858Arg in the available sources is predominantly somatic NSCLC (drug-response) disease: the landmark L858R papers describe sporadic tumors with wild-type matched normal tissue (PMID:15118073, PMID:15118125, PMID:15329413) and report no affected-relative testing. No germline disease phenotype or familial aggregation attributable to this variant is documented in any source, and ClinVar submissions report no segregation data. Evidence is insufficient, so PP1 is not assessed.

Rule

Generic ACMG/AMP 2015 (Richards et al., Genet Med 2015;17:405-424, PMID 25741868): PP1 - co-segregation with disease in multiple affected family members in a gene definitively known to cause the disease (supporting pathogenic).

Evidence considered

• ClinVar variation 16609: no submission reports co-segregation or family segregation testing for EGFR c.2573T>G; conditions are predominantly somatic drug-response phenotypes.
• PMID:15118073 (Lynch et al., NEJM 2004): describes 9 gefitinib-responsive NSCLC patients with somatic EGFR mutations (L858R in Patients 5 and 6); no family or segregation analysis is reported.
• PMID:15118125 (Paez et al., Science 2004): EGFR mutations in NSCLC tumor cohorts confirmed as somatic via paired normal tissue; no familial segregation data are reported.
• PMID:15329413 (Pao et al., PNAS 2004): L858R reported in gefitinib/erlotinib-responsive and smoking-status cohorts of unrelated NSCLC patients; no family or segregation data are reported.
• Keyword review of all five fetched full-text publications returned zero matches for family, pedigree, germline-inheritance, or segregation content.

Still missing

• Family segregation study with genotypes of affected and unaffected relatives in a germline disease context
• Documentation of a germline disease phenotype or familial aggregation attributable to this variant
• Any pedigree or family-history data for a proband carrying this variant

Sources

Reasoning

PP2 (Richards et al. 2015) requires (a) a gene with a low rate of benign missense variation and (b) a gene in which missense variants are a common mechanism of disease. No gene-level benign-missense constraint metric for EGFR (e.g., a gnomAD missense Z-score or observed/expected ratio) is available in the case evidence, so prong (a) cannot be evaluated. For prong (b), no ClinGen CSPEC, VCEP, or other authoritative source in this case establishes a germline disease for EGFR in which missense variants are a common mechanism; the only germline-mechanism signal available is a literature-only generic-fallback PVS1 gene-context note pointing to a loss-of-function germline mechanism, whose supporting citations are largely not EGFR-specific and which would in any case argue against a missense-common mechanism. The variant's well-documented disease associations are somatic gain-of-function (OncoKB 'Oncogenic', gain-of-function; COSMIC 2,857 somatic occurrences), which do not establish germline PP2. Evidence is insufficient to satisfy either prong, so PP2 is not assessed.

Rule

PP2: 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 (Richards et al. 2015, PMID 25741868; generic ACMG/AMP fallback)

Evidence considered

• No gene-level benign-missense constraint metric for EGFR (e.g., gnomAD missense Z-score or observed/expected ratio) is present in the case evidence; prong (a) of PP2 cannot be evaluated.
• The PVS1 gene-context note (pvs1_gene_context) is a literature-only, generic-fallback determination asserting germline loss-of-function mechanism eligibility; its supporting citations are largely not EGFR-specific (papers on ERBB3, NOTCH3, COL4A3/4, BRCA2, PTPN11), so it does not establish a missense-common germline mechanism for EGFR.
• The variant's established disease associations are somatic gain-of-function: OncoKB classifies L858R 'Oncogenic' with biological-effect 'Gain-of-function', and COSMIC records 2,857 somatic occurrences (evidence.json); somatic oncogenic driver status does not establish germline PP2.
• No ClinGen CSPEC or VCEP exists for EGFR in this case (framework_mode 'generic_acmg'), so no gene-specific PP2 specification is available.
• The variant is a missense substitution (pvs1_variant_assessment consequence class 'missense'), satisfying PP2's variant-type prerequisite only; the gene-level prongs remain unsupported.

Still missing

• No gnomAD or equivalent gene-level benign-missense constraint metric (missense Z-score, observed/expected ratio) for EGFR in the case evidence.
• No authoritative (CSPEC/VCEP or equivalent) germline gene-disease mechanism evidence establishing that missense variants are a common mechanism of EGFR disease; the only germline signal is a literature-only generic-fallback note.

Sources

Reasoning

PP3 is in scope because NM_005228.4:c.2573T>G is a missense variant (NP_005219.2:p.(Leu858Arg)) and no EGFR ClinGen CSPEC/VCEP or gene-specific PP3/BP4 lookup exists, so the generic ACMG/AMP 2015 definition applies (multiple lines of computational evidence support a deleterious effect on the gene or gene product). The calibrated missense-computational evidence required for PP3 is absent: REVEL lookup returned no score for this variant (chr7:55191822T>G), BayesDel lookup returned no score (and BayesDel has no threshold from a named publication available, so it could not be treated as calibrated evidence even if a score existed), and aGVGD data are not present. The only computational prediction available, SpliceAI max delta 0.013 (DS_AG 0.0, DS_AL 0.0, DS_DG 0.013, DS_DL 0.01), addresses splice impact and is reserved for splice-region/intronic variants; c.2573 lies deep within exon 21 (exon spans c.2470-2625, placing the variant ~103 nt from the exon 21 acceptor and ~52 nt from the exon 21 donor), so this exonic missense is not evaluated on the SpliceAI path, and a low splice-delta could not in any case support PP3. With no calibrated missense predictor result available, there is no evidence base to conclude PP3 met or not met; evidence is insufficient, so PP3 is not assessed.

Evidence considered

• NM_005228.4:c.2573T>G is annotated as the missense variant NP_005219.2:p.(Leu858Arg) in exon 21, so PP3 is in scope through the calibrated missense-computational path.
• No EGFR ClinGen CSPEC/VCEP or gene-specific PP3/BP4 supporting lookup is available; the generic ACMG/AMP 2015 framework (Richards et al., PMID 25741868) applies, under which PP3 requires multiple lines of computational evidence supporting a deleterious effect on the gene or gene product.
• REVEL lookup returned no score for this variant (chr7:55191822T>G), so no REVEL-based evidence is available.
• BayesDel lookup returned no score for this variant, and no BayesDel threshold from a named publication is available, so BayesDel cannot serve as calibrated evidence for PP3 or BP4.
• SpliceAI reports max delta 0.013 (DS_AG 0.0, DS_AL 0.0, DS_DG 0.013, DS_DL 0.01) for this exonic missense; c.2573 sits ~103 nt from the exon 21 acceptor boundary and ~52 nt from the exon 21 donor boundary, so the variant is not a splice-region/intronic change and the SpliceAI path scoped to such variants is not applied.

Still missing

• No calibrated missense predictor score is available for c.2573T>G: REVEL lookup returned no record for chr7:55191822T>G, BayesDel lookup returned no record, and aGVGD is not available; obtain a REVEL score (published ClinGen SVI thresholds, Pejaver et al. 2022, PMID 36413997) or an equivalent calibrated predictor with a named published threshold to resolve PP3.

Reasoning

Generic ACMG/AMP 2015 PP4 (Richards et al., PMID 25741868): 'Phenotype or family history highly specific for a gene with a single genetic etiology'; PP4 may also carry the supporting-strength role for case-control odds ratios of approximately 2-5 per the ACMG/AMP 2015 text, but that usage likewise requires case-control data. No CSPEC/VCEP modifies PP4 for EGFR. The case record contains no proband phenotype, clinical presentation, indication, or family history in any of its evidence artifacts, so no phenotype-to-gene specificity statement can be made. The variant's curated associations are somatic and pharmacogenomic (ClinVar conditions including 'Adenocarcinoma of lung, response to tyrosine kinase inhibitor in, somatic'; OncoKB 'Oncogenic', gain-of-function), and these do not constitute a germline phenotype or family history for PP4 under the generic germline framework. No case-control odds ratio for the exact variant exists (see PS4), so the supporting-strength OR usage of PP4 is also unavailable. Evidence insufficient: status not_assessed.

Rule

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

Evidence considered

• Generic ACMG/AMP 2015 PP4 (Richards et al., PMID 25741868): 'Phenotype or family history highly specific for a gene with a single genetic etiology'; no EGFR CSPEC/VCEP modifies this criterion in the case.
• The case record contains no proband phenotype or family-history information (no phenotype/indication fields in the review manifest or case summary), so no phenotype-to-gene specificity determination is possible.
• The exact variant is a well-established somatic NSCLC oncogenic driver whose curated clinical associations are drug-response phenotypes (ClinVar 'drug response' conditions; OncoKB 'Oncogenic' gain-of-function), which are not a germline phenotype or family history and cannot anchor PP4.
• No case-control odds ratio for EGFR c.2573T>G is available, so the supporting-strength case-control usage of PP4 cannot be invoked.

Still missing

• No proband phenotype, clinical presentation, indication, or family history is present anywhere in the case record, so phenotype-gene specificity (a phenotype highly specific for EGFR with a single genetic etiology) cannot be assessed.
• No case-control odds ratio exists for this exact variant, so PP4's supporting-strength role for case-control data (ACMG/AMP 2015) cannot be applied either.

Sources

Reasoning

BS4 (generic ACMG/AMP 2015, PMID 25741868) requires demonstrated non-segregation with disease, e.g., the variant present in an unaffected family member or absent in an affected family member of a fully penetrant disorder. No such family genotype data exist in this case: there are no pedigrees, no affected/unaffected relative genotypes, and no ClinVar or literature reports of familial testing for EGFR c.2573T>G. Contextual observations were explicitly set aside because they cannot substitute for family testing: absence from gnomAD v2.1/v4.1/Canada is a population-frequency observation (expected for a recurrent somatic oncogenic variant, and BS4 concerns affected/unaffected relatives, not population absence), and somatic origin in tumors does not constitute non-segregation within a family. Evidence is insufficient, so BS4 is not assessed.

Rule

Generic ACMG/AMP 2015 (Richards et al., Genet Med 2015;17:405-424, PMID 25741868): BS4 - non-segregation of the variant with disease, e.g., variant observed in an unaffected family member or absent in an affected family member where the phenotype is fully penetrant (strong benign).

Evidence considered

• ClinVar variation 16609: no submission reports testing of unaffected or affected relatives for this variant.
• gnomAD v2.1, v4.1, and gnomAD-Canada v1.0: variant absent from all three population datasets; this is a population-frequency observation and was not treated as BS4 non-segregation evidence.
• PMID:15118073, PMID:15118125, PMID:15329413: L858R is documented as a somatic tumor mutation with wild-type matched normal tissue in the reported patients; somatic acquisition in tumor is not family non-segregation and was not used as BS4 evidence.
• Keyword review of all five fetched full-text publications found no family, pedigree, or segregation content that could support or refute BS4.

Still missing

• Genotypes of affected and unaffected family members needed to test non-segregation with disease
• Pedigree and family-history data for any proband carrying this variant
• A germline disease phenotype definition against which segregation status could be evaluated

Sources

Reasoning

BP1 (Richards et al. 2015) applies to a missense variant in a gene for which primarily truncating variants are known to cause disease. The variant is a missense substitution (consequence class 'missense'), satisfying BP1's variant-type prerequisite, but the gene-mechanism prerequisite is not established in this case: no CSPEC, VCEP, or authoritative curation defines an EGFR germline disease caused primarily by truncating variants. The PVS1 gene-context note (literature-only, generic fallback) asserts germline loss-of-function mechanism eligibility, but its supporting citations are largely not EGFR-specific (papers on ERBB3, NOTCH3, COL4A3/4, BRCA2, and PTPN11), it was not designed to gate BP1, and a loss-of-function mechanism per se does not establish that truncating variants are the primary cause of an EGFR germline disease to the exclusion of missense effects. Evidence is insufficient to establish BP1's gene-mechanism prerequisite, so BP1 is not assessed.

Rule

BP1: Missense variant in a gene for which primarily truncating variants are known to cause disease (Richards et al. 2015, PMID 25741868; generic ACMG/AMP fallback)

Evidence considered

• The variant is a missense substitution (pvs1_variant_assessment consequence class 'missense'), so BP1's variant-type prerequisite is met, but that is not sufficient for BP1.
• The PVS1 gene-context note (pvs1_gene_context) is a literature-only generic-fallback determination of germline LoF-mechanism eligibility; its supporting citations are largely not EGFR-specific and it does not establish that truncating variants primarily cause an EGFR germline disease.
• No ClinGen CSPEC or VCEP exists for EGFR in this case (framework_mode 'generic_acmg'), so no gene-specific BP1 specification is available.
• Generic ACMG/AMP 2015 framework (generic_acmg_combination_rules) applies; BP1 additionally requires the gene-disease mechanism to be one in which truncating variants predominate, which no consulted source establishes for EGFR germline disease.

Still missing

• No authoritative (CSPEC/VCEP or equivalent) evidence establishing that EGFR germline disease is primarily caused by truncating variants; the only LoF-mechanism signal is a literature-only generic-fallback note whose supporting citations are largely not EGFR-specific.
• No gene-level truncating-vs-missense constraint data (e.g., gnomAD pLI/observed-expected for LoF) are available in the case evidence to evaluate the mechanism prerequisite.

Sources

Reasoning

BP4 is in scope because NM_005228.4:c.2573T>G is a missense variant (NP_005219.2:p.(Leu858Arg)) and no EGFR ClinGen CSPEC/VCEP or gene-specific PP3/BP4 lookup exists, so the generic ACMG/AMP 2015 definition applies (multiple lines of computational evidence suggest no impact on the gene or gene product). The calibrated missense-computational evidence required for BP4 is absent: REVEL lookup returned no score for this variant (chr7:55191822T>G), BayesDel lookup returned no score (and BayesDel has no threshold from a named publication available, so it could not be treated as calibrated evidence even if a score existed), and aGVGD data are not present. The only computational prediction available, SpliceAI max delta 0.013 (DS_AG 0.0, DS_AL 0.0, DS_DG 0.013, DS_DL 0.01), is reserved for splice-region/intronic variants; c.2573 lies deep within exon 21 (exon spans c.2470-2625, placing the variant ~103 nt from the exon 21 acceptor and ~52 nt from the exon 21 donor), so this exonic missense is not evaluated on the SpliceAI path, and even if considered it would constitute only a single line of evidence, not the multiple lines BP4 requires. With no calibrated missense predictor result available, there is no evidence base to conclude BP4 met or not met; evidence is insufficient, so BP4 is not assessed.

Evidence considered

• NM_005228.4:c.2573T>G is annotated as the missense variant NP_005219.2:p.(Leu858Arg) in exon 21, so BP4 is in scope through the calibrated missense-computational path.
• No EGFR ClinGen CSPEC/VCEP or gene-specific PP3/BP4 supporting lookup is available; the generic ACMG/AMP 2015 framework (Richards et al., PMID 25741868) applies, under which BP4 requires multiple lines of computational evidence suggesting no impact on the gene or gene product.
• REVEL lookup returned no score for this variant (chr7:55191822T>G), so no REVEL-based evidence is available.
• BayesDel lookup returned no score for this variant, and no BayesDel threshold from a named publication is available, so BayesDel cannot serve as calibrated evidence for PP3 or BP4.
• SpliceAI reports max delta 0.013 (DS_AG 0.0, DS_AL 0.0, DS_DG 0.013, DS_DL 0.01) for this exonic missense; c.2573 sits ~103 nt from the exon 21 acceptor boundary and ~52 nt from the exon 21 donor boundary, so the variant is not a splice-region/intronic change, the SpliceAI path scoped to such variants is not applied, and a single splice-prediction line would not satisfy BP4's multiple-lines requirement in any case.

Still missing

• No calibrated missense predictor score is available for c.2573T>G: REVEL lookup returned no record for chr7:55191822T>G, BayesDel lookup returned no record, and aGVGD is not available; obtain a REVEL score (published ClinGen SVI thresholds, Pejaver et al. 2022, PMID 36413997) or an equivalent calibrated predictor with a named published threshold to resolve BP4.

Reasoning

Generic ACMG/AMP 2015 BP5 (Richards et al., PMID 25741868): 'Variant found in a case with an alternate molecular basis for disease' - a benign indicator applied when the patient's phenotype is explained by a different genetic etiology (variant in another gene/locus, or a clearly explanatory non-genetic cause) found in the same case. BP5 is a proband-level criterion: it requires knowledge that an individual carrying the variant was evaluated for, and found to have, an alternate molecular explanation. The case record contains no proband-level clinical or molecular data whatsoever: no phenotype, no reported alternative variant or copy-number finding, and no testing details. The variant's germline ClinVar context (one ClinPGx 'drug response' submission and one Ambry VUS) does not report any alternate-molecular-basis finding. No VCEP/CSPEC governs EGFR and the generic fallback defines BP5 qualitatively with no numeric threshold, so no threshold comparison applies. Evidence insufficient to determine presence or absence of an alternate molecular basis: status not_assessed.

Rule

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

Evidence considered

• Generic ACMG/AMP 2015 BP5 (Richards et al., PMID 25741868): 'Variant found in a case with an alternate molecular basis for disease'; a proband-level benign indicator requiring evidence that the individual's phenotype is explained by a different genetic etiology.
• The case record contains no proband phenotype or alternate-cause testing results, and the ClinVar record for the exact variant (VCV000016609) reports no alternate-molecular-basis finding in any submission.
• No CSPEC/VCEP governs EGFR; the generic fallback defines BP5 qualitatively with no numeric threshold, so no threshold-based comparison applies to this criterion in this case.

Still missing

• No proband-level clinical or molecular data exist in the case (no phenotype, no reported alternate variant, no alternate-cause evaluation), so it cannot be determined whether an alternate molecular basis of disease was found in a variant carrier.

Sources