Clinical Predictors at Diagnosis of Low-Risk Histopathology in Unilateral Advanced Retinoblastoma.

PURPOSE: Attempted eye salvage for unilateral (cT2b/group D) retinoblastoma may risk tumor spread compared with primary enucleation. Identification of clinical features predictive of low histopathologic risk support safe trial salvage. DESIGN: Retrospective, noncomparative single-institutional observational case series. PARTICIPANTS: Children with unilateral cT2b/group D retinoblastoma managed with primary enucleation at the Hospital for Sick Children, Toronto, Canada, January 2008 through February 2018. METHODS: Data included clinical features (intraocular pressure, optic nerve obscuration, macular involvement, tumor seeding, and serous retinal detachment [RD] >1 quadrant), timing to enucleation, histopathologic features, and follow-up. MAIN OUTCOME MEASURES: Primary outcome was low-risk (LR; pT1/pT2) versus high-risk (HR; pT3/pT4) histopathologic features with clinicopathologic correlations. Secondary outcomes were positive predictive (probability that certain clinical features would predict LR histopathologic features) and negative predictive values (probability that absence of these clinical features would predict HR histopathologic features). RESULTS: Thirty-eight eyes were eligible and showed vitreous seeding and normal intraocular pressure. The median diagnosis to enucleation interval was 4 days (range, 0-14 days). Histopathologic analysis diagnosed 4 (10.5%) HR and 34 (89.5%) LR eyes. High-risk eyes demonstrated massive choroidal invasion (4/38) or trans-scleral, extraocular, and postlaminar optic nerve invasion (1/38). Clinical findings included macular involvement (31/38), complete optic nerve obscuration (27/38), and RD (28/38). The proportion of eyes with HR histopathologic features was 13% (4/31; 95% confidence interval [CI], 1%-25%) with macular involvement, 15% (4/27; 95% CI, 1%-28%) with complete optic nerve obscuration, and 14% (4/28; 95% CI, 1%-27%) with RD. The predictability of LR histopathologic features was 100% with macular sparing (7/7; 95% CI, 47%-100%), optic nerve visibility (10/10; 95% CI, 63%-100%), and less than 1 quadrant of RD (10/10; 95% CI, 63%-100%). In 1 child lacking all 3 clinical LR predictive features with HR histopathologic features (pT3a), metastases developed and the patient died; other children are alive and well (mean follow-up, 65 months). CONCLUSIONS: Presence of macular sparing, optic nerve visibility, less than 1 quadrant of RD, or a combination thereof predicted LR histopathologic features at primary enucleation, suggesting safe trial eye salvage. No clinical sign predicted HR histopathologic features.