Rituximab for multiple sclerosis linked to a tenfold higher risk of inflammatory bowel disease
A Swedish study of 13,748 MS therapy episodes finds rituximab users developed IBD far more often than users of other modern MS drugs, mostly Crohn disease.
Summary
Rituximab is the most widely used multiple sclerosis (MS) drug in Sweden, and it works by depleting B cells. Swedish researchers tracked more than 13,000 treatment episodes in adults with relapsing MS and compared rates of new inflammatory bowel disease (IBD) across drugs and against the general population. People taking rituximab developed IBD at about 110 per 100,000 person-years. That compares with 13 for natalizumab, dimethyl fumarate and fingolimod combined, and 41 in the general population. Most rituximab-linked cases were Crohn disease. The absolute risk was still small: roughly 1 extra case for every 131 people treated for 5 years. The findings suggest doctors and patients should weigh bowel risk when choosing an MS therapy and take new gut symptoms seriously.
Detailed Summary
Anti-CD20 therapies, which deplete B cells, are among the most effective treatments for relapsing-remitting multiple sclerosis (RRMS). Rituximab, used off-label, is now the most common MS drug in Sweden. Postmarketing reports and case series have hinted that anti-CD20 drugs, including ocrelizumab, ofatumumab and rituximab, may trigger new-onset inflammatory bowel disease (IBD). Large population-level data were missing, so clinicians could not put a number on the risk when choosing among therapies.
Swedish researchers linked the national MS register with demographic and health registers. They identified 13,748 therapy episodes in 9,251 adults aged 18 to 55 who started treatment between 2011 and 2023: 6,481 rituximab, 2,976 natalizumab, 2,767 dimethyl fumarate and 1,524 fingolimod. They also built 61,065 matched general-population episodes. Everyone had at least 10 years of lookback with no prior IBD suspicion. IBD required two records of Crohn disease, ulcerative colitis or indeterminate colitis codes. Follow-up continued through therapy plus six months.
There were 164 IBD events: 40 with rituximab, 4 with the three comparator drugs combined, and 120 in the general population. Crude incidence was 110 per 100,000 person-years for rituximab (95% CI, 76-147), 13 for comparators (3-30) and 41 for the general population (34-48). After adjusting for demographics and MS characteristics with inverse probability weighting, the hazard ratio for comparator drugs versus rituximab was 0.10 (95% CI, 0.04-0.22). The general population hazard ratio was 0.33 (0.23-0.52) after demographic adjustment.
The type of disease also differed. Crohn disease made up 62% of rituximab-associated IBD, against 29% in the general population. Median time to diagnosis was 2.9 years on rituximab. About 30% of rituximab patients with IBD stopped therapy just before or within two years of diagnosis. Rituximab episodes that developed IBD began at younger ages and in people with shorter MS duration, though the authors say the data were too sparse for firm risk-factor conclusions.
In absolute terms the risk remains modest. Cumulative IBD incidence stayed below 2% in every group at 10 years. The fully adjusted number needed to harm was 131 at 5 years (95% CI, 93-228), meaning one extra IBD case per 131 people treated with rituximab instead of a comparator. Sensitivity analyses supported this: 158 using an ever-treated approach, 131 with a stricter IBD definition, and 92 when restricting to years of greater overlap between therapy groups.
The authors conclude that IBD risk belongs in treatment decisions, with vigilance for gastrointestinal symptoms. Caveats are important. Only 4 events occurred among comparators, which forced grouping and penalized estimation. Other anti-CD20 drugs were not studied, although the authors think a class effect is likely. Missing MS-specific data ranged from 35% to 61%, residual confounding is possible, chart and pathology data were absent, and the cohort was entirely Swedish. The comparator drugs also showed lower IBD rates than the general population, which may reflect chance.
Key Findings
- Among 13,748 RRMS therapy episodes (9,251 adults), 40 IBD cases occurred with rituximab versus only 4 with natalizumab, dimethyl fumarate and fingolimod combined
- Crude IBD incidence was 110 per 100,000 person-years with rituximab, 13 with comparator DMTs and 41 in the general population (61,065 matched episodes)
- Fully adjusted hazard ratio for comparator DMTs versus rituximab was 0.10 (95% CI, 0.04-0.22); the general population versus rituximab was 0.33 (0.23-0.52) after demographic adjustment
- Crohn disease accounted for 62% of rituximab-associated IBD versus 29% in the general population
- Number needed to harm for rituximab versus comparator DMTs was 131 at 5 years (95% CI, 93-228); sensitivity analyses gave 158, 131 and 92
- Cumulative IBD incidence remained under 2% in all groups at 10 years, so absolute risk is low
- Median time to IBD was 2.9 years on rituximab, and about 30% of affected rituximab patients stopped treatment around diagnosis
Methodology
This was a population-based cohort study using linked Swedish national registers, including the Swedish MS register, with STROBE reporting. It included 13,748 therapy episodes (rituximab, natalizumab, dimethyl fumarate, fingolimod) from 2011-2023 in adults aged 18-55 with RRMS, plus 61,065 matched general-population episodes. IBD was defined by a validated rule of at least two ICD-10 K50, K51 or K52.3 records. Follow-up ended at first IBD diagnosis, therapy stop plus 6 months, death, emigration or data cutoff. Analyses used penalized Cox models, inverse probability of treatment weighting, multiple imputation for missing MS characteristics, cluster bootstrap CIs, Kaplan-Meier curves and number-needed-to-harm estimates.
Study Limitations
Authors note that other anti-CD20 drugs (ocrelizumab, ofatumumab) could not be studied, comparator groups had very few IBD events (4), and the comparators may not be risk-neutral. Treatment patterns changed over time, 35-61% of MS characteristics were missing, residual confounding is possible, and clinical chart and pathology data were unavailable. Findings come from a Swedish population only. Conflicts of interest were not included in the supplied text.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: