Out of 216 patients randomized in the FAST trial, 163 patients (76%) completed all 12 months of follow-up. 26 patients were considered treatment successes for the first 6 months, continued on the same antimetabolite, and then failed at 12 months. These patients were not included in this study, leaving 137 patients for evaluation. Of these, 115 patients were considered treatment successes at 12 months - 88 patients who were kept on the same antimetabolite throughout the study and 27 patients who were treatment failures at 6 months on one antimetabolite, but then controlled at 12 months on the other antimetabolite. Of these 27 patients, only three had retinal vasculitis defined by vascular leakage posterior to the equator on FA. The 22 remaining patients failed both MTX and MMF (failed one antimetabolite at 6 months, then the other antimetabolite at 12 months).

Of these 137 patients, 67% (92) were female and the mean age was 40 years. Seventy-six patients (56%) had FA images obtained at the baseline visits. Age, sex, bilateral involvement, uveitis duration, and country/study sites were not associated with failing both antimetabolites in the univariate analysis. Conversely, failing both antimetabolites was associated with a classification of anterior/intermediate uveitis, the presence of CME, and the presence of retinal vasculitis in the univariate analysis (Table 1).

The final model included age, sex, anatomic location, CME, and retinal vasculitis as covariates. Within the final model – after adjusting for age, sex, anatomic location, and the presence of CME – the presence of retinal vasculitis was the only characteristic that remained statistically significantly associated with failing both antimetabolites (adjusted odds ratio [OR], 8.6 [95% CI, 1.6 to 47]; P = 0.014) (Table 2). In further sensitivity analysis, this result remained statistically significant in simulations of different prevalence of retinal vasculitis in those without FA images, even in a scenario where patients who failed both antimetabolites were 20% less likely to have retinal vasculitis and treatment successes were 20% more likely to have retinal vasculitis compared to the respective probabilities in those who did have FA images in this study (OR, 6.8 [95% CI, 2.4 to 19]).