Rituximab for myasthenia gravis.

Rationale: Myasthenia gravis (MG) is an autoimmune disease which causes muscle weakness due to disruption in neuromuscular transmission. Rituximab is a medication increasingly used in the treatment of MG, but its potential benefits, and optimal use in terms of patient subgroup and dosing, are unclear. It is important to bring together high-quality evidence to determine how rituximab would be best used in treatment algorithms for MG.

Objectives: To assess: - the effects of rituximab (including biosimilar variants) for the treatment of MG in adults; and - the benefits and harms of rituximab in different patient subgroups, and treatment strategies, in order to aid treatment choice for individuals, and inform policymakers about those most likely to benefit.

Search methods: We searched CENTRAL, MEDLINE, Embase, and two trials registries (Clinicaltrials.gov and the World Health Organization trials registry) up to November 2024.

Eligibility criteria: We included randomised controlled trials (RCTs) or quasi-RCTs in adults (aged 16 years and over) with MG (all subtypes and severity), comparing rituximab (any dosing regimen) with placebo, no treatment, or alterative therapy. We excluded cluster-RCTs and nonrandomised studies, studies in those previously treated with rituximab, and studies analysing juvenile MG (under 16 years of age).

Outcomes: Critical outcomes were improvement in symptom severity or functional ability (as measured by Quantitative MG (QMG) and MG-Activities of Daily Living (ADL) scores), reduction in the burden of alternative treatment (steroid-sparing effect and relapse requiring rescue therapy), and serious adverse events (SAEs) in the long term (beyond nine months). Important outcomes included MG Composite (MGC) score, quality of life, hospital admissions and antibody titre, at short-term (two months or less), medium-term (two to nine months), and long-term (beyond nine months) time points. Achievement of a clinically significant improvement in QMG, MG-ADL and MGC scores was analysed dichotomously. We examined safety by looking at adverse events (AEs), treatment-related AEs, and AEs leading to discontinuation of treatment. We also analysed critical outcome measures at the short- and medium-term time points.

Risk of bias: We used the Cochrane risk of bias 1 tool RoB 1 to assess potential bias.

Synthesis methods: We synthesised results for each outcome using meta-analysis where possible, with random-effects models to calculate mean difference (MD) or risk ratios (RRs) and 95% confidence intervals (CI). Where this was not possible due to the nature of the data, we synthesised results by summarising effect estimates. Data were analysed on an intention-to-treat basis. We used GRADE to assess the certainty of evidence for each outcome. Sensitivity analysis examined whether a fixed-effect model or the use of odds ratios would alter conclusions.

Included studies: We included two RCTs with a total of 99 participants. One study was conducted in Europe and one in North America, and both were published in 2022. The study populations and treatment strategies differed; one administered rituximab at low doses in new or early-onset generalised MG, and the other at high repeated doses as add-on therapy.

Synthesis of results: The evidence has limitations. Beyond nine months, the evidence is very uncertain on the effects of rituximab on symptom severity as assessed with QMG score (MD 1.62 lower (favouring rituximab), 95% CI 3.53 lower to 0.29 higher; 2 studies, 94 participants; very low-certainty evidence), and functional ability as assessed by MG-ADL (MD 1.16 lower (favouring rituximab), 95% CI 2.48 lower to 0.16 higher; 2 studies, 95 participants; very low-certainty evidence). The evidence suggests that rituximab results in little to no difference in its steroid-sparing effect beyond nine months (RR 1.00, 95% CI 0.92 to 1.09; 2 studies, 94 participants; low-certainty evidence), but probably results in a large reduction in relapse requiring rescue therapy (220 out of 1000 people with rituximab, compared with 490 out of 1000 people with placebo, favouring rituximab, RR 0.45, 95% CI 0.26 to 0.78; 2 studies, 98 participants; moderate-certainty evidence). Rituximab may reduce SAEs, but the evidence is very uncertain (RR 0.81 (favouring rituximab), 95% CI 0.47 to 1.41; 2 studies, 99 participants; very low-certainty evidence). The main limitation of this review is that only two studies were included, which used rituximab in different ways (low dose at onset of generalisation compared with high dose as add-on therapy). The studies mainly assessed acetylcholine-receptor antibody MG. There were differences in co-administered steroid dosing between studies. We considered the studies to be at low risk of bias, apart from possible bias from differences in characteristics between treatment arms. There was serious to extremely serious imprecision in the certainty of evidence when assessing several outcomes due to wide confidence intervals, and serious indirectness in all outcomes as not all forms of MG were studied.

Authors' conclusions: Rituximab's effects on symptom severity and functional ability in the long-term are uncertain. The evidence suggests that rituximab results in little to no difference in its steroid-sparing effect; however, it probably results in a large reduction in relapse requiring rescue therapy over nine months, based on results from two studies. The available data about the effects of rituximab on SAEs are of very low certainty, and so we are not able to draw conclusions. There are inadequate data to determine optimal dosing regimen or patient characteristics. Further studies examining rituximab, and other B cell-depleting therapies, in different MG patient subgroups are warranted.

Funding: This Cochrane review had no dedicated funding.

Registration: Protocol (2023) available via DOI: 10.1002/14651858.CD014574.