Higher observed 12-month relapse-free survival with satralizumab than with rituximab in aquaporin-4 immunoglobulin G-positive neuromyelitis optica spectrum disorder: a multicenter retrospective cohort study

Background Direct real-world comparisons of satralizumab and rituximab in aquaporin-4 immunoglobulin G-positive neuromyelitis optica spectrum disorder remain uncommon. Methods We conducted a multicenter retrospective cohort study across eight hospitals in Fujian Province, China. Patients aged 12 years or older contributed treatment episodes beginning at the first verifiable administration of satralizumab or rituximab. Patients who changed therapy could contribute one episode to each treatment group, while uncertainty estimates accounted for patient-level clustering. The primary outcome was time to the first clinician-documented NMOSD relapse. Kaplan–Meier methods estimated 12-month relapse-free survival and its absolute between-treatment difference. Restricted mean relapse-free time, recurrent relapses, treatment-period safety events, and exploratory Expanded Disability Status Scale outcomes were also examined. Sensitivity analyses considered treatment switching, episode selection, calendar era, center composition, age restriction, and propensity-score overlap weighting. Results The cohort included 103 patients and 115 treatment episodes (50 satralizumab and 65 rituximab). First relapse within 12 months occurred in 1 satralizumab episode and 10 rituximab episodes. Twelve-month relapse-free survival was 98.0% (95% CI, 94.2–100.0%) with satralizumab and 84.6% (95% CI, 76.2–93.8%) with rituximab, an absolute difference of 13.44 percentage points (patient-clustered bootstrap 95% CI, 4.20–23.59). The direction was consistent across the principal unweighted sensitivity analyses. Corrected overlap weighting yielded a 5.28-percentage-point difference (95% CI, −11.69 to 21.21); limited common support, an effective rituximab sample size of 9.38, and appreciable patient influence made this an exploratory estimate. Infection rates were 18.2 and 14.7 events per 100 treatment-episode years, respectively (rate ratio, 1.24; 95% CI, 0.58–2.37), and one serious adverse event occurred in each group. Exploratory EDSS analyses did not identify a clear between-group difference. Conclusions Satralizumab episodes had higher observed 12-month relapse-free survival than rituximab episodes in this cohort. The finding is a clinically relevant real-world comparison signal, while the treatment-specific contribution to the difference remains uncertain because treatment eras were separated, common support was limited, and satralizumab events were sparse.