OBJECTIVES : To assess relationships among demographic traits, relapse profile or annual relapse rate (ARR) with treatment changes in patients with NMOSD in the CIRCLES study 2013–2020.

METHODS : The IRB-approved longitudinal, observational study, CIRCLES, was conducted at 15 North American medical centers. Patients having ≥60 days of follow-up and on-study maintenance treatment were evaluated. Mean ARR was estimated using Poisson models, while likelihood of treatment change was estimated using Cox proportional hazards models for gender and racial/ethnic cohorts and relapse features: self-reported pre-study ARR; on-study relapse(s) and phenotypes. Relapses were incorporated as time-varying covariates to estimate relationship to treatment change.

RESULTS : Of 542 patients studied, 158 (29.2%) experienced ≥1 relapse with 131 patients (24.2%) changing treatment. In univariate analyses, factors associated with significantly increased likelihood of treatment change were self-reported pre-study ARR >0.75 (vs. <0.25; HR=2.98; p<0.001) and on-study relapse (HR=2.53; p<0.001). For phenotypes, relapses including optic neuritis (ON) were significantly associated with treatment change, whether with potential brain involvement (HR=3.53; p<0.001) or transverse myelitis (HR=3.6; p=0.001); multivariate analyses were concordant.

Among 542 patients, mean on-study ARR differed per gender/race/ethnicity: females: Hispanic (0.28); White (0.23); Black (0.18); Asian (0.06); males: Black (0.48); Asian (0.33); Hispanic (0.22); White (0.15). Gender did not correlate with ARR but stratifying by gender/race/ethnicity revealed significant differences: females: Asian lower than Hispanic (p=0.009) or White (p=0.018) but not Black (p=0.065); males: Black higher than White (p=0.045) but not Hispanic (p=0.342) or Asian (p=0.563). Race and gender were not associated with likelihood of treatment change.

CONCLUSIONS : Change in treatment regimen correlated with a higher pre-study ARR, on-study relapse events, and ON relapse phenotypes but not demographic differences in ARR. Specific patient cohorts may be likely to change treatment based on NMOSD disease activity or other factors. These findings suggest patient relapse profiles may predict treatment change in NMOSD.