Résumé
Introduction: In relapsing-remitting multiple sclerosis (RRMS), the early identification of suboptimal responders would prevent irreversible disability progression.Objectives: We aimed to develop and validate a dynamic scoring system to help in the early decision of switch from first to second-line therapy.Methods: From a French cohort of 12,823 patients with RRMS and by using time-dependent propensity scores, we constructed a 1:1 emulated clinical trial (ECT) to compare patients who switched to a second-line treatment because of inefficacy versus comparable patients remaining on first-line treatment. Then, we proposed a frailty Cox model to predict the time to relapse according to the two groups. The validation of the scoring system was performed by two additional ECTs from independent patients.Results: From the learning ECT, the switch benefit was higher for younger patients at the disease onset, with a low level of EDSS at first-line therapy initiation, or in case of at least one observed relapse or Gd-enhancing T1 lesion. It allowed to predict the individual hazard ratio (iHR) of relapse in case of switch versus waiting and to identify two sub-populations: the patients with iHR⩽0.69 with a significant benefit of the switch and the others. From the validation ECT of 348 independent patients with iHR⩽0.69, we reported the 5-year relapse-free survival at 0.14 (95%CI from 0.09 to 0.22) in the waiting group and 0.40 (95%CI from 0.32 to 0.51) in the switched group. From the validation ECT of 518 independent patients iHR>0.69, these values equalled 0.37 (95%CI from 0.30 to 0.46) and 0.44 (95%CI from 0.37 to 0.52), respectively.Conclusions: Based on an original methodology for causal prediction, we proposed a dynamic scoring system to support the early decision to switch to second-line therapy. We estimated that at least one-third of patients could benefit from an earlier switch to prevent relapses.