Abstract

Abstract PURPOSE/OBJECTIVE Progressive SCNSL after failure of CNS-directed systemic therapy is a challenging clinical scenario. CNS-RT can effectively salvage, but its optimal role is poorly defined. METHODS We analyzed diffuse large B-cell patients who received brain RT for progressive SCNSL following 1+ line of therapy for active SCNSL (not prophylaxis). Overall survival (OS) was calculated by Kaplan-Meier and RT response was evaluated using IPCG or RANO criteria for parenchymal or leptomeningeal disease, respectively. Risk of intracranial failure assumed death as a competing risk. Pre-RT SCNSL patterns were classified as parenchymal-only versus diffuse (periventricular or leptomeningeal). Local failure was progression only of pre-RT lesion(s). RESULTS We identified 55 patients with median follow up of 3.5m. 85% had neurologic symptoms, 45% had extracranial disease. Patients received a median of 2 prior SCNSL-directed therapies (IQR 1-3); 91% received methotrexate-based regimens with median 21d between last methotrexate and RT. Median CNS-RT dose was 30 Gy in 10 fractions delivered to whole brain (n=44, 80%), partial brain (n=9, 16%), or craniospinal axis (n=2, 4%). Median OS was 3.5m. Overall response for 38 evaluable was 79% (24% complete), with median of 60d to best response. 58% were successfully bridged with CNS-RT to subsequent therapies and 12m intracranial failure risk was 37% for those bridged, and 64% for those not bridged (p=0.05). 19 had parenchymal-only disease (10 unifocal, 9 multifocal) with median largest lesion diameter 4.1cm (range 1.7–7.4). 19 had diffuse pattern (12 periventricular, 7 leptomeningeal). 7 with parenchymal-only disease failed intracranially, 5 of which were local. 8 with diffuse disease failed intracranially, only 1 of which was local (p = 0.035). CONCLUSION SCNSL patients referred for RT had symptomatic, treatment-refractory disease. While CNS-RT response rates were high, duration was often limited, suggesting RT is best used for bridging/palliation. Patients with parenchymal-only disease were enriched for local failure, suggesting potential suitability of involved site CNS-RT, though further validation is needed.