Local and systemic toxicity of intravitreal melphalan for vitreous seeding in retinoblastoma: a preclinical and clinical study.

PURPOSE: Intravitreal melphalan is emerging as an effective treatment for refractory vitreous seeds in retinoblastoma, but there is limited understanding regarding its toxicity. This study evaluates the retinal and systemic toxicity of intravitreal melphalan in retinoblastoma patients, with preclinical validation in a rabbit model. DESIGN: Clinical and preclinical, prospective, cohort study. PARTICIPANTS: In the clinical study, 16 patient eyes received 107 intravitreal injections of 30 µg melphalan given weekly, a median of 6.5 times (range, 5-8). In the animal study, 12 New Zealand/Dutch Belt pigmented rabbits were given 3 weekly injections of 15 µg of intravitreal melphalan or vehicle to the right eye. METHODS: Electroretinogram (ERG) responses were recorded in both humans and rabbits. For the clinical study, ERG responses were recorded at baseline, immediately before each injection, and at each follow-up visit; 82 of these studies were deemed evaluable. Median follow-up time was 5.2 months (range, 1-11). Complete blood counts (CBCs) were obtained on the day of injection at 46 patient visits. In the animal study, ERG responses were obtained along with fluorescein angiography, CBCs, and melphalan plasma concentration. After humane killing, the histopathology of the eyes was evaluated. MAIN OUTCOME MEASURES: For the clinical study, we measured peak-to-peak ERG amplitudes in response to 30-Hz photopic flicker stimulation with comparisons between ERG studies before and after intravitreal melphalan. For the animal study, we collected ERG parameters before and after intravitreal melphalan injections with histopathologic findings. RESULTS: By linear regression analysis, over the course of weekly intravitreal injections in retinoblastoma patients, for every additional injection, the ERG amplitude decreased by approximately 5.8 µV. The ERG remained stable once the treatment course was completed. In retinoblastoma patients, there were no grade 3 or 4 hematologic events. One week after the second injection in rabbits, the a- and b-wave amplitude declined significantly in the melphalan treated eyes compared with vehicle-treated eyes (P<0.05). Histopathology revealed severely atrophic retina. CONCLUSIONS: Weekly injections of 30 µg of melphalan can result in a decreased ERG response, which is indicative of retinal toxicity. These findings are confirmed at an equivalent dose in rabbit eyes by ERG measurements and by histopathologic evidence of severe retinal damage. Systemic toxicity with intravitreal melphalan at these doses in humans or rabbits was not detected.