Divergent Trends in Relapse Activity and Relapse-Independent Progression Over Two Decades in Multiple Sclerosis

INTRODUCTION: Multiple sclerosis (MS) management has increasingly favored earlier high-efficacy disease-modifying therapy (DMT). We examined diagnosis-era changes in relapse activity and disability accrual alongside evolving DMT use. METHODS: We retrospectively analyzed two Italian cohorts of adults with MS onset in 2001-2024 across four diagnosis eras. Annualized relapse rate (ARR) was modeled by Poisson regression; confirmed disability accrual (CDA), relapse-associated worsening (RAW), and progression independent of relapse activity (PIRA) by Andersen-Gill Cox models. CDA was classified as RAW when relapse onset occurred from 90 days before through 30 days after the event, otherwise as PIRA. Treatment was time-varying: untreated, low-/moderate-efficacy therapy (LET/MET), or high-efficacy therapy (HET). Inverse probability-weighted models included residual adjustment. Sensitivity analyses examined a strict common 2-year horizon, MRI restriction/adjustment, relapsing-onset MS, first-event analyses, 90-day DMT lag, early-HET exclusion, pre-treatment versus never-treated person-time, visit frequency, and center stratification. RESULTS: Among 1096 patients (228, 214, 292, and 362 across eras 1-4), first-line HET use reached 45.9% in era 4. Each later diagnosis year was associated with lower ARR (RR 0.98, 95% CI 0.96-0.99; p < 0.001), with steeper calendar-time declines during LET/MET and HET than untreated person-time (joint interaction p < 0.001). Later diagnosis year was also associated with lower RAW (HR 0.94, 95% CI 0.92-0.97; p < 0.001) and modestly lower CDA over full follow-up (HR 0.98, 95% CI 0.96-1.00; p = 0.026), although CDA attenuated in weighted and common-horizon analyses. PIRA was not associated with diagnosis year (HR 1.01, 95% CI 0.99-1.04; p = 0.262). Lower ARR and RAW persisted across sensitivities, whereas the strict 2-year analysis suggested higher early PIRA in more recent eras. CONCLUSION: Later diagnosis-era cohorts had lower relapse activity and relapse-associated disability accrual, whereas overall disability accrual was less consistent and PIRA did not decline. Improvements in relapse-related disease activity were not accompanied by a comparable reduction in relapse-independent disability accrual.