NM_000321.2:c.2236G>T (p.Glu746Ter) is a nonsense variant in exon 22 of RB1, introducing a premature termination codon predicted to trigger nonsense-mediated decay in a gene with an established loss-of-function disease mechanism for retinoblastoma.¹ PVS1 is met at very strong strength.² This variant is extremely rare in population databases, absent from gnomAD v2.1 and observed in only 1 of 1,187,988 alleles in gnomAD v4.1 (AF = 8.42e-7), satisfying PM2 at supporting strength.³ This variant has been reported in ClinVar as Pathogenic by two clinical laboratories (Variation ID 1069507, 2-star review status) and is listed in association with retinoblastoma.⁴ This variant has been observed in the somatic context in COSMIC (COSV57314725, n = 6), consistent with a pathogenic role, though somatic data alone do not independently satisfy germline ACMG criteria. No variant-specific functional studies, de novo observations, co-segregation data, or statistical case-frequency data were identified in the reviewed literature to satisfy PS3, PS2, PS4, or PP1. Under the generic ACMG/AMP 2015 combination rules (PMID:25741868), one very strong criterion (PVS1) plus one supporting criterion (PM2) does not reach the threshold for Likely Pathogenic or Pathogenic classification. The combination of PVS1 and PM2 alone results in a variant of uncertain significance (VUS) by strict scoring.⁵ Note: In clinical practice, a PVS1-level nonsense variant in RB1 with ClinVar Pathogenic consensus and extreme population rarity would typically be classified as at least Likely Pathogenic. The strict ACMG/AMP 2015 scoring framework requires additional supporting criteria to reach Likely Pathogenic (e.g., 1 Very Strong + 1 Moderate, or 1 Strong + 2 Supporting). ClinVar submissions from two clinical laboratories and somatic COSMIC observations provide orthogonal clinical support not captured by the generic scoring matrix.⁶