This is a retrospective study with a national database that benefits from its consequent sample size but suffers from multiple methodological biases inherent to these records, which limits the extraction of reliable result considerablys.

Firstly, patients with at least two EDSS (Expanded Disability Status Scale) evaluations are included, which could lead to an underestimation of the effect in patients with few evaluations (specifically, those with only two evaluations, where it is impossible to assess if subsequent confirmation of the event is required). Secondly, the treatment group includes patients with at least one cycle of anti-CD20 antibodies, without specifying the number of cycles or the number of patients for each treatment duration. Thirdly, there are significant differences in most baseline characteristics, with younger and more aggressive patients in the treatment group, who had shorter follow-ups. This could bias this group towards a poorer response. Although they attempt to mitigate this with propensity score matching, this statistical technique has limitations (mainly that it matches patients based on variables handpicked by the researchers). Fourthly, most patients were treated with rituximab, without specifying the doses and frequency of administration (as it was off-label use). Fifthly, only variables with statistically significant results in the univariate analyses are included in the multivariate analysis, which is a statistical error because all variables that could plausibly have an effect should be included.

Therefore, although real-world studies are necessary as a complement to clinical trials, the significant limitations of this study cast doubt on its superiority over the demonstrated results of ocrelizumab in primary progressive multiple sclerosis (given that rituximab has not been formally tested in this disease).

It is certainly an interesting study, but it should be interpreted with caution. It is a retrospective study, and like all such studies there are certain methodological issues that limit its findings and conclusions.

Of the total number of patients treated with anti-CD20 therapies in the French multicentre cohort, more than half could not be included for analysis and the authors do not provide data to know whether the included group is truly representative of the patients indicated for these therapies and something similar happens with the control group. It is also important to note that very few patients were included with ocrelizumab (which is the currently approved drug for the management of patients with primary progressive multiple sclerosis - PPMS) and that, due to the indications for which this treatment can be prescribed, the characteristics of the patients included in the two groups (treated and untreated) are very different. As much as the authors used different statistical techniques to control for these differences, the underlying biological changes can often not be adjusted for and ensure a correct comparison.
As expected, many of the untreated patients were included before 2018, when ocrelizumab was approved for patients with PPMS, and therefore, in many of these cases there was no information on MRI scans, a key tool for monitoring and stratifying patients with multiple sclerosis (MS).

Furthermore, it is not specified which treatment regimen they underwent. In the case of ocrelizumab (131 patients) it is marked in the data sheet, but the dose and frequency of administration of rituximab (the largest group in this study with 295 patients) is not clear and is very centre-dependent, so the results reported in any case would refer to these specific doses. There are data suggesting that the impact of anti-CD20 therapies on disability may be related to the dose received, so this would be essential to know.

We are used to seeing published studies of effectiveness in clinical practice in cohorts of patients with relapsing-remitting MS, where the measures of response are somewhat clearer. This is one of the few papers that assesses this effectiveness in patients with progressive forms of MS, but the response measure in this case (the EDSS disability scale) is a measure that is more difficult to interpret, especially when it is collected retrospectively without the rigour required in clinical trials. Of course, these are interesting results, but I don't think they will lead us to change our usual clinical practice.