Trilateral retinoblastoma: a systematic review and meta-analysis

Background About 5% of children with retinoblastoma from germline mutation of the RB1 gene are at risk of developing trilateral retinoblastoma—intraocular retinoblastoma combined with a histologically similar brain tumour, most commonly in the pineal gland. We aimed to provide a systematic overview of published data for trilateral retinoblastoma, and to analyse how survival has changed. Methods We searched Medline and Embase for scientific literature published between Jan 1, 1966, and April 14, 2014, that assessed trilateral retinoblastoma cases. We undertook a meta-analysis of survival with the Kaplan-Meier method and Cox proportional hazards regression, stratified on the basis of the original study, to account for between-study heterogeneity. Findings We included 90 studies, with 174 patients with trilateral retinoblastoma. 5-year survival after pineal trilateral retinoblastoma increased from 6% (95% CI 2–15) in patients diagnosed before 1995, to 44% (26–61; p<0·0001) in those diagnosed from 1995 onwards. Before 1995, no patients with non-pineal trilateral retinoblastoma survived, but from 1995 onwards, 5-year survival was 57% (30–77; p=0·035). Hazard ratios (HR) adjusted for the presence of leptomeningeal metastases and trilateral retinoblastoma location, suggested that both conventional (HR 0·059, 95% CI 0·016–0·226; p<0·0001) and high-dose chemotherapy with stem-cell rescue (0·013, 0·002–0·064; p<0·0001) most strongly contributed to this improvement. Absence of leptomeningeal metastases (HR 2·13, 95% CI 0·98–4·60; p=0·055) were associated with improved survival. Non-pineal trilateral retinoblastomas were larger than pineal tumours (median 30 mm [range 6–100] vs 22 mm [7–60]; p=0·012), but both had similar outcomes since 1995. Interpretation Our results suggest that improvements in overall survival are attributable to improved chemotherapy regimens and early detection of pineal trilateral retinoblastoma. As such, successful treatment of trilateral retinoblastoma should include screening at least at the time of retinoblastoma diagnosis and chemotherapy, which would preferably be a high-dose regimen with autologous stem-cell rescue. Funding None.