Comparative Effectiveness in NMOSD: Relapse Risk, Safety Signals, and Why Treatment Choice Matters

Neuromyelitis optica spectrum disorder (NMOSD) is a relapse-driven disease in which long-term disability is closely tied to inflammatory attack prevention. As more FDA-approved therapies enter the field, clinicians must weigh not only efficacy data from clinical trials but also real-world safety, tolerability, and durability of response when selecting long-term treatment strategies.

In this 5-part panel series, Shamik Bhattacharyya, MD, Anne M. Finucane Endowed Chair in Neurology and Associate Professor of Neurology at Harvard Medical School, and Philippe-Antoine Bilodeau, MD, of the Division of Neuroimmunology at Mass General Brigham and Instructor in Neurology at Harvard Medical School, analyze their recently published comparative effectiveness study in Neurology. Drawing on a large institutional cohort, they explore how FDA-approved NMOSD therapies perform relative to rituximab and traditional immunosuppressants, while carefully addressing both relapse outcomes and serious infectious risk.

In this episode, the discussion turns to the study’s core findings. The panel reviews relapse-free survival, annualized relapse rates, and composite safety endpoints, emphasizing the absence of relapses on newer therapies, the measurable cumulative risk observed with rituximab, and the high failure rates associated with mycophenolate mofetil and azathioprine. They also contextualize these findings within real-world patient experience and long-term treatment durability.

Philippe-Antoine Bilodeau, MD:

We had a fairly large cohort of 176 patients who met our study inclusion criteria. The majority were aquaporin-4 positive—about 86%—and the basic demographics of our cohort were in line with previously published NMOSD studies. Most patients were women, and the median age at first attack was 42.

As we mentioned earlier, we had two main goals. One was comparative effectiveness: comparing time to first relapse across the different therapies. We were able to adjust for several factors to allow for causal inference, including number of prior attacks and disability at treatment start. Disability at baseline was not significantly different between groups.

What we found was a substantial failure rate, particularly on mycophenolate mofetil and azathioprine. Over half of patients on those therapies had relapsed by year 10. Perhaps more surprisingly, we also found a nontrivial relapse risk on rituximab. If you extend out to 10 years, about 25% of patients on rituximab experienced relapse.

Some previous studies suggested that most relapse risk with rituximab is concentrated early, and we did observe that the curves drop in the first two to three years. However, we also saw relapses occurring several years into treatment—five, six, even seven years later—which was somewhat surprising.

Importantly, we did not observe any relapses on the newer therapies. We had 18 treatment periods on complement inhibitors, 11 on inebilizumab, and 17 on satralizumab. Across all of those, there were no relapses. When we limited the analysis to aquaporin-4–positive patients—the group for whom these therapies are approved—we found similar results and were able to statistically demonstrate a significant benefit of inebilizumab, satralizumab, and complement inhibitors compared with rituximab.

We also examined annualized relapse rates and found very low rates on the newer therapies.

For safety, we defined events as either treatment-limiting adverse events—meaning adverse events that led to discontinuation—or serious infections, typically those requiring hospitalization. We observed a meaningful risk of treatment discontinuation or serious infection on rituximab, whereas this risk was less pronounced on the newer therapies, particularly complement inhibitors, which had very few safety events during our study period.

We then analyzed a composite endpoint combining efficacy, safety, and tolerability—time to first relapse, serious infectious adverse event, or treatment-limiting adverse event. Using that endpoint, we found significant differences between therapies. For rituximab, over half of patients by year 10 had experienced either relapse, serious infection, or treatment discontinuation. Again, the newer therapies, especially complement inhibitors, performed better on this composite measure.

Shamik Bhattacharyya, MD:

To add to what Dr. Bilodeau described, one of the main takeaways is that these treatments are not all the same. Clinical trials often focus primarily on efficacy, but real-world data allow us to evaluate infections and long-term treatment tolerability. In this study, we were able to combine those elements to better reflect the real patient experience.

Another interesting observation was the timing of events. While relapse events on rituximab were clustered in the first one to two years, many of the safety events—particularly serious infections—were clustered after three to four years of therapy. If you combine safety, tolerability, and efficacy, there is an early relapse risk, but as patients remain on the drug longer, safety considerations become increasingly important.

We felt we understood this clinically, but this study allowed us to quantify how significant those risks are.

There are limitations to our data. Approximately 142 patients were treated with rituximab, whereas fewer than 20 patients were treated with each of the newer agents, including complement inhibitors, inebilizumab, and satralizumab. While we had enough data to observe strong signals, we were not able to fully quantify the magnitude of benefit. As more patients are treated and additional data accumulate, the differences we observed may become even more pronounced.