Disease modifying treatments for relapsing remitting multiple sclerosis, including rituximab. A health technology assessment.

OBJECTIVE:

For disease-modifying treatments for RRMS
To assess effectiveness, based on annual relapse rate, disability progression and new lesions detected by magnetic resonance imaging (MRI).
To assess safety, based on risk of mortality, risk of serious adverse events, rate of treatment withdrawal due to adverse events, and risk of specific rare serious adverse events.
To describe legal implications of off-label use of rituximab.

KEY FINDINGS AND CONCLUSIONS:

We have systematically collected and reviewed the evidence for clinical effectiveness and general safety issues for disease modifying treatments for relapsing remitting multiple sclerosis, synthesised evidence from randomised controlled trials and non-randomised registry-based studies using net- work meta-regression, and carefully interpreted the findings. We included rituximab in our analysis as it is used off-label for the treatment of patients with RRMS, even though it does not hold marketing authorisation for RRMS.
We included 35 randomised controlled trials and 11 non-randomised registry-based studies, with a total of almost 30 000 patients. We compared estimates of our predefined outcomes from meta-analy- sis of randomised controlled trials, of non-randomised registry-based studies, the network meta-re- gression, and other network meta-analytical models, and judged that the estimates are mutually con- sistent in most cases, and that where there is inconsistency, it could be explained.
Based on the available evidence and the meta-analysis used: alemtuzumab is most likely to be the best treatment with respect to annual relapse rate; ocrelizumab and alemtuzumab are equally likely to be the best treatments with respect to risk of disability progression. Further, we estimate that rituximab is likely to have the lowest risk of serious adverse events and treatment withdrawal due to adverse events. However, the evidence for rituximab is from one small randomised trial of short dura- tion and one non-randomised study, making this finding uncertain.
Treatment rankings are based on available evidence and model assumptions, and in many cases confi- dence intervals for the highest-ranked treatments overlap, so rankings should not be interpreted as definitive.
There were very few deaths in the included studies (30 deaths out of a total of 22 060 patients). Alt- hough we performed a full network meta-analysis, we judged that the number of events was too small to support useful conclusions regarding mortality risk.
We compiled information of rare, and potentially life-threatening effects of disease modifying treat- ments from the included studies, but we have not searched other sources or databases that may be more suitable for such information. The risk of specific serious adverse events was not estimated due to the limited data available, but data were retrieved from all included studies. The events were gen- erally uncommon in the included studies, which reported no statistically significant differences in rates of serious adverse events.
The effect estimates of annualised relapse rate and sustained disability progression were used in a health economic analyses that is reported in a separate publication