Objective To compare response to rituximab (RTX) between adult patients positive for myelin oligodendrocyte glycoprotein (MOG) and aquaporin-4 (AQP4) antibodies. Methods We prospectively studied adult patients with MOG or AQP4 antibodies who received RTX under an individualized dosing schedule adapted to the biological effect of RTX monitored by memory B-cell measurement. Memory B cells were counted monthly and when relapse occurred. The biological effect of RTX was considered significant with <0.05% memory B cells in peripheral blood lymphocytes. Results In 16 patients with MOG antibodies and 29 with AQP4 antibodies, mean follow-up was 19 (range = 9-38) and 38 (13-79) months. Under RTX, 10 relapses occurred in 6 of 16 (37.5%) patients with MOG antibodies, and 13 occurred in 7 of 29 (24%) with AQP4 antibodies. The median time of relapse after the most recent infusion was 2.6 (0.6-5.8) and 7 (0.8-13) months, respectively (p < 0.001). Memory B cells had reemerged in 2 of 10 (20%) relapses in patients with MOG antibodies and 12 of 13 (92.5%) with AQP4 antibodies (p < 0.001). Interpretation In AQP4 antibody-associated disorder, relapse mostly occurs when the biological effect of RTX decreases, which argues for treatment efficacy. In MOG antibody-associated disorder, the efficacy of RTX is not constant, because one-third of patients showed relapse despite an effective biological effect of RTX. In this subpopulation, memory B-cell depletion was unable to prevent relapse, which was probably caused by different immunological mechanisms. These findings should be used to improve treatment strategies for MOG antibody-associated disorder. ANN NEUROL 2019Read less <