Stopping Disease-Modifying Treatments in Multiple Sclerosis: A Systematic Review and Meta-Analysis of Real-World Studies

Background The question of whether multiple sclerosis requires life-long disease-modifying treatments (DMTs) remains unanswered. Some studies suggest that older patients with stable disease may safely discontinue their DMTs, yet comprehensive evidence-based data are scarce and real-world studies have provided mixed results. Objective The aim of this study was to assess the rate of disease reactivation and associated risk factors after discontinuation of DMTs in patients with multiple sclerosis. Methods We searched scientific databases (PubMed/MEDLINE, Scopus and Google Scholar) to identify real-world studies published until 31 July, 2023 that reported the number of patients who experienced relapses and/or disability accrual (outcomes of interest) following a therapy discontinuation longer than 12 months. Magnetic resonance activity and treatment re-start after DMT discontinuation were also considered as additional outcomes. We excluded studies where therapy discontinuation was explicitly related to an unintended or planned pregnancy or preceded a treatment switch. We ran random-effects meta-analyses, subgroup analyses and meta-regression models to provide pooled estimates of post-discontinuation relapse and disability events, and to identify their potential moderators (predictors). Results After an independent screening, 22 articles met the eligibility criteria, yielding a pooled sample size of 2942 patients followed for 1–7 years after discontinuation (11,689 patient-years). The pooled rates for relapse and disability events were 6.7 and 5.8 per 100 patient-years, respectively. However, available data did not allow us to disentangle isolated disability accrual from relapse-associated worsening. Studies including older patients (β = −0.65, p = 0.006), patients with a longer exposure to DMTs (β = −2.22, p = 0.001) and patients with a longer period of disease stability (β = −2.74, p = 0.002) showed a lower risk of relapse events. According to meta-regression equations, the risk of relapse events after DMT discontinuation became negligible (arbitrarily set at < 1% per year) at approximately 60 years of age, and after either 10 years of DMT exposure, or 8 years of disease stability. Additional analyses showed pooled rates for magnetic resonance imaging activity and re-start events of 16.7 and 17.5 per 100 patient-years, respectively. Conclusions Based on our quantitative synthesis of real-world data, in the absence of definitive answers from clinical trials, DMT discontinuation appears feasible with a high degree of certainty in selected patients. While our findings are robust regarding relapse events, future efforts are warranted to determine if DMT discontinuation is associated with isolated disability accrual.