Assessing the duration of EDSS improvement after a therapy start: A novel approach applied to the long-term extension of the PRISMS study.

Background: In a chronic and progressive disease such as multiple sclerosis (MS), the improvement on Expanded Disability Status Scale (EDSS) can be a transient event. Therefore, estimating the prevalence of disability improvement over time, accounting both for improvement incidence and duration, is of interest. The aim of this study was to show the application of a simple estimator for the proportion of patients with sustained improvement over time using data from the long-term extension of the PRISMS trial.

Methods: A total of 534 relapsing-remitting MS (RRMS) patients from the PRISMS trial were included. Patients with a baseline EDSS of 0 were excluded. Patients were randomized to placebo (n = 178), subcutaneous interferon beta-1a (sc IFN β-1a) 22 µg (n = 181) or sc IFN β-1a 44 µg (n = 175). At Year 2, patients receiving placebo were re-randomized to sc IFN β-1a 22 µg or 44 µg (delayed sc IFN β-1a) while patients receiving sc IFN β-1a 22 µg or 44 µg continued their initial regimen. Patients were followed up for over 7 years post-randomization. Disability improvement was defined as a 1-point decrease in EDSS from baseline confirmed at 6 months. Prevalence of improvement was estimated as difference of Kaplan-Meier (KM) estimators while the cumulative incidence of improvement was calculated using the standard KM curves.

Results: No significant differences in cumulative incidence of EDSS improvement at 3 years between delayed sc IFN β-1a (20.3%) and sc IFN β-1a 22 µg (20.8%; p = 0.49) or 44 µg (21.3%; p = 0.33). When taking duration of improvement into account, the proportion of patients showing an improved condition after 3 years was 10.1% with delayed sc IFN β-1a, 11.3% with sc IFN β-1a 22 µg (p = 0.17) and 15.4% with sc IFN β-1a 44 µg (p = 0.037) that was substantially maintained over the long term.

Conclusions: With the use of this new statistical methodology, it is possible to estimate the time to improvement as well as the duration of improvement, information that is better suited to describing a non-final outcome like disability improvement. In this case, early sc IFN β-1a 44 µg initiation had a greater proportion of patients with a sustained disability improvement over a long period of follow-up as compared to patients who had initially been randomized to placebo. In contrast, no significant differences on the cumulative incidence of improvement were observed.