Introduction

Neuromyelitis optica spectrum disorder (NMOSD) includes relapsing inflammatory demyelinating disorders affecting the CNS and, more specifically, the spinal cord and the optic nerve.1 Three different subtypes have been described depending on the serostatus of patients: positive antiaquaporin-4 antibody (NMOSD-AQP4+), positive antimyelin oligodendrocyte glycoprotein antibody (NMOSD-MOG+), and double-seronegative (NMOSD-DN).2,3 Recently, because of a broad clinical spectrum than “classical” NMOSD4 and different pathologic mechanisms,5,6 patients with anti-MOG antibodies were reclassified as MOG antibody–associated disease (MOGAD)7 with recent international diagnostic criteria.8 The French Caribbean is a geographical area primarily affected by these diseases,9 with a last estimated prevalence of 10/100,000 habitants in Martinique.10,11 Moreover, the Caribbean population with African ancestry could experiment with a worse prognosis than others.12 Genetic and environmental factors are suspected to explain the increased disease frequency in this area.11,13,14

Mitoxantrone (MiTX), an anthracenedione antineoplastic agent, was the first long-term immunosuppressant (IS) used in the late 1990s and 2000s for NMOSD in French Caribbean. The use of MiTX in NMOSD was inspired by its application in patients with highly active relapsing-remitting multiple sclerosis (MS)15 and because of its action on B cells. Encouraging results on disease activity and safety were observed in observational studies16,17 and a recent meta-analysis.18 The latest NMO treatment updates and reviews recommend MiTX as a second-line treatment worldwide, while azathioprine, mycophenolate mofetil, and rituximab are recommended as first line.6,12,19,-,22 In addition, due to difficulties accessing other IS, such as rituximab, until the 2010s, MiTX was often used in first-line therapy in AQP4+ and seronegative patients, especially in French West Indies.

More recently, new long-term IS have emerged, targeting several immune mechanisms involved in NMO and, more specifically, in NMOSD-AQP4+ patients (eculizumab,23 satralizumab,24,25 tocilizumab,26 and inebilizumab27). These new therapies are the subject of a recent international consensus for their first-line uses in AQP4-IgG+ NMOSD.28 However, the economic cost of MiTX is drastically lower than new molecules and could be the first choice in many areas largely affected by this disease but displaying a fragile health care system.

In this study, we reported the evaluation of efficacy, safety, and predictive factors of relapse under MiTX in a large cohort of patients with NMOSD in the French Caribbean.

MethodsPatients and Study Design

All adult patients diagnosed between January 1993 and December 2020 from 4 different sites in French West Indies (Martinique, Guadeloupe, Saint-Martin, and Saint-Barthelemy) and treated with the MiTX regimen (at least the first 3 infusions in the year, see further) were included in the study according to the last international consensus diagnostic criteria for NMOSD.4 The diagnostic criteria were applied retrospectively for patients diagnosed between 1993 and 2015 after reviewing cases and prospectively after 2015. Note that patients were included regardless of severity or economic status and benefited from 100% health insurance, giving them free access to all medical care and medications if indicated. From 2004 and the AQP4 antibody test availability, newly diagnosed patients were currently tested during their first inflammatory attack. Patients with NMOSD diagnosed before 2004 were retested during their follow-up or after a relapse unless they died first. From 2016 and the possibility of screening for MOG antibody in serum using cell-based assay (CBA), patients with a first attack were tested for the MOG antibody during hospital care. Most seronegative patients before 2016 were retested when the test became available, during a new relapse, or during a follow-up consultation. Some of these patients were finally diagnosed with MOGAD. In addition, we have next defined in our study 3 subgroups according to serostatus: NMOSD-AQP4+, MOGAD, and NMOSD-DN.

In our study, patients were assessed at least twice a year by a referring neurologist and if a relapse occurred. All serum samples were analyzed at the University Hospital Center of Lyon (France) using CBA to detect AQP4 and MOG antibodies.

Exclusion criteria were previous treatment with other anthracyclines or anthracenedione and rituximab treatment within 6 months before MiTX, current pregnancy or breastfeeding, active or previous heart disease, abnormal echocardiography with or without decreased left ventricular ejection fraction, current use of cardiotoxic drugs, and history of radiotherapy in the mediastinal or pericardial region. Patients who had not completed at least 3 infusions in the year (36 mg/m2) were excluded. Moreover, patients who did not meet the Wingerchuk et al. diagnostic criteria4 were excluded from the study. These patients, diagnosed in the 1990s, had highly probable NMOSD and received MiTX. However, they could not meet the diagnostic criteria because they could not be tested for AQP4-IgG before their deaths (unavailable until 2004).

Treatment Regimen

MiTX regimen consisted of 1 monthly infusion of 12 mg/m2 (M1-M2-M3) and then 3 quarterly courses with the same dosage (M6-M9-M12). Therefore, the therapeutic target was 72 mg/m2 (6 infusions in the year). The complete protocol included an infusion of 0.9% of sodium chloride (100 mL) and medication with 8 mg of ondansetron (PO) 30 minutes before the infusion of MiTX diluted in 100 mL of 5% glucose.

Outcomes

The primary outcome to evaluate MiTX efficacy was the first physician-determined relapse between the initiation of MiTX (T0) and the 96th week of follow-up (T96). The occurrence of relapse was defined by an increase of 0.5 points in patients with an Expanded Disability Status Scale (EDSS) ≥ 6, an increase of 1 point in patients with an EDSS <6, or a rise of ≥2 points in functional system score. Specifically, a referring ophthalmologist specialized in NMOSD confirmed attacks on the optic nerve. In addition, the visual functional system score had to increase by ≥ 2 points. MRI began in the French West Indies in the early 2000s and was performed when patients relapsed as a diagnostic complement. However, diagnostic confirmation of an inflammatory attack was primarily clinical. In our study, symptoms lasting less than 24 hours (n = 1), an acute painful episode without modification of EDSS (n = 3), functional somatic disorders (n = 1), and Uhthoff phenomenon (n = 1) were not considered relapses.

Secondary criteria were evaluating (1) the time to first relapse at 96-week follow-up, (2) the EDSS and annualized relapse rate (ARR) at 96-week follow-up (EDSS-96 and ARR-96), and (3) the occurrence of severe adverse effects. We considered 3 groups of increasing severity: minor (EDSS [0–3.5]), moderate (EDSS [4–5.5]), and severe (EDSS [6–10]). The ARR is the number of relapses with onset occurring during a specific period, adjusted to 1 year.

We considered the complete success of MiTX whether the patient was relapse-free at 96 weeks follow-up and incomplete achievement whether the patient had a relapse while having a decrease in ARR-96. Conversely, we retained partial failure whether the patient had a relapse without modification in ARR-96 and complete failure when a relapse occurred while having an increase in ARR-96.

Concomitant Immunosuppressant Medications

Gradually from 2015, and especially after the publication of a study by Montcuquet et al.29 on the effectiveness of mycophenolate mofetil as first-line therapy in patients with NMOSD regardless of their serologic status, our patients benefited from concomitant immunosuppressive treatment after the end of 6 infusions of MiTX. Therefore, 9 of our patients (6 NMOSD-AQP4+ and 3 MOGAD) received an IS relay by mycophenolate-mofetil (from 500 mg to 2 g per day) between the sixth infusion (M12) of MiTX and the 96th week of follow-up. Before 2015, implementing concomitant therapy immediately after the completion of MiTX infusions was uncommon in the French West Indies. Moreover, due to relapses, 1 patient had a rituximab infusion between the M12 of MiTX and the 96th week of follow-up, while 20 patients had rituximab after 96 weeks.

Statistical Analysis

Quantitative variables were summarized as median with interquartile range (IQR), and categorical data were expressed as numbers and frequencies. The χ2 test (categorical data) and the Kruskal-Wallis test (quantitative data) were used for comparison of patients according to the serostatus (NMOSD-AQP4+, NMOSD-DN, and MOGAD).

We performed an event-free survival analysis, considering the occurrence of relapse in the first 96 weeks after initiation of MiTX. We generated Kaplan-Meier plots of relapse-free survival for subgroups of patients according to their serostatus and used the log-rank test to compare these curves.

Patients with and without relapses during the first 96 weeks of follow-up were compared using the χ2 or Fisher exact test for categorical variables and the Wilcoxon-Mann-Whitney test to compare quantitative variables. The Cox proportional hazards regression model was used to estimate the crudes and adjusted hazard ratios (HRs) and their 95% confidence intervals.

The percentage of variation of ARR during the 96-week follow-up was calculated as follows: 100 × (ARR_T0 - ARR_T96)/ARR_T0. The 3 serostatus subgroups were compared using Kruskal-Wallis and Dunn post hoc tests for pairwise comparisons. The Wilcoxon test for paired series was used to compare the evolution of the EDSS between T0 and T96 globally and according to each antibody.

For all analyses, the significance threshold was set at 5% (p < 0.05). Analyses were performed using R 4.2.1 software and RStudio 2022.07.1 (R Core Team (2022). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL R-project.org/).

Standard Protocol Approvals, Registrations, and Patient Consents

From 1993 to 2010, patients with NMOSD were prospectively reported in the European Database for MS (EDMUS) created at the Hospices Civils de Lyon in 1976, with a European dimension since 1990, and which listed cases of MS and related diseases. From 2010, the period of creation of the French national cohort "Neuro-optico-myélite aiguë de Devic et des syndromes neurologiques apparentés, NOMADMUS," which is associated with the OFSEP (Observatoire Français de la Sclérose En Plaques), new patients with NMOSD were prospectively registered, and older patients’ data were transferred from EDMUS to NOMADMUS. All patients gave written informed consent, and the national ethical authority approved data collection (Commission Nationale de l’Informatique et des Libertés; CNIL; registration number 914066v2).

Data Availability

Data supporting the findings of this study are available from the corresponding author upon reasonable request.

ResultsGeneral Characteristics

A total of 86 patients were included (Figure 1), and their characteristics are summarized in Table 1. Patients were almost all the Afro-Caribbean descent (n = 84, 97.7%), with a women predominancy (n = 74, 86%). The median age at onset of the disease was 39 years (IQR 26.2–51.0). Regarding the serostatus, 59 patients (68.6%) were AQP4+, 7 (8.1%) MOG+, and 20 (23.3%) DN.

Figure 1 Patient Selection Flowchart

AQP4+ = positive antiaquaporin-4 antibody; DN = double-seronegative; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease; NMOSD = neuromyelitis optica spectrum disorder. *Patients diagnosed in the 1990s had highly probable NMOSD and received MiTX. However, they could not meet the Wingerchuk diagnostic criteria because they could not be tested for AQP4-IgG before their deaths (unavailable until 2004).

Table 1

Demographic and General Characteristics of Patients

Before the initiation of MiTX, the most recent attack registered (P0) was transverse myelitis in 52 cases (60.5%), optic neuritis (ON) in 35 patients (40.7%), an area postrema syndrome in 3 cases (3.5%), and a symptomatic cerebral syndrome in 1 case (1.2%). The same attack could affect multiple neurologic locations (9 patients, 10.5%).

Nineteen patients (22.1%) were previously treated with an immunosuppressive drug before MiTX (Table 1). Before MiTX, this group of patients experienced more relapses than others (mean 5.53, ±2.59, vs 1.94 ± 1.09, p < 0.001), with a higher pretreatment ARR (mean 1.16 ± 0.80, vs 0.76 ± 0.43, p = 0.024) (eTable 1, links.lww.com/NXI/A929).

Evaluation of Mitoxantrone Efficacy and Safety

In the global cohort, the median time from disease onset to the initiation of MiTX was 9 months (IQR 2.0–52.0) (Table 2). Nearly 45% of patients experienced at least 3 relapses before initiating MiTX. We more frequently started the MiTX after the first attack in MOGAD patients vs others (85.7% vs 33.9% in NMOSD-AQP4+ and 25% in NMOSD-DN patients, p = 0.015).

Table 2

Efficacy and Safety of Mitoxantrone

During the 96-week follow-up, the first physician-adjudicated relapse occurred in 25 patients (29.1%). Among them, the median time to relapse was 31 weeks (IQR 18.0–58.0) (Table 2, Figure 2). Overall, 36 patients (41.9%) relapsed during the entire follow-up. Before the sixth (M12) infusion of MiTX, 16 patients “fast” relapsed (eTable 2, links.lww.com/NXI/A930). These patients were older than those who relapsed after M12 (9/16 vs 3/20 ≥ 40 years old, respectively, p = 0.009). A first physician-adjudicated relapse was more frequent in NMOSD-AQP4+ patients (23/59 (39%), vs 1/20 (5%) in DN and 1/7 (14.3%) in MOG+ patients, p = 0.007) (Table 2, Figure 2). Regardless of serologic status, of the 31 patients who received MiTX from the first attack, 27 (87%) did not relapse at the 96th week of follow-up vs 34/55 (61.8%) (p = 0.013) patients who did not receive the MiTX from their first attack (HR 0.29, 95% CI [0.10–0.84], p = 0.023) (Table 3).

Figure 2 First Physician-Adjudicated Relapse in the Overall Population

AQP4+ = positive antiaquaporin-4 antibody; DN = double-seronegative; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease; NMOSD = neuromyelitis optica spectrum disorder.

Table 3

Relapse Predictors at the 96-Week Follow-up

The mean ARR was significantly reduced from 0.85 (±0.55) (the year before MiTX) to 0.32 (±0.63) at 96-week follow-up (p < 0.001), corresponding to a decrease of 61.3% (±77.1%) (Table 2). The mean ARR-96 is even lower for the NMOSD-DN subgroup vs NMOSD-AQP4+ (mean 0.07 (±0.31) vs 0.41 (±0.69), respectively, p = 0.02) (Figure 3A), and the percentage of decrease in ARR is also more remarkable for the NMOSD-DN than for NMOSD-AQP4+ (93.1% (±30.9) vs 47% (±87.8), respectively, p = 0.012) (Figure 3B).

Figure 3 Evolution of Mean ARR From Initiation of Mitoxantrone to 96-Week Follow-up

The mean ARR-96 between the three subgroups (A). The percentage of decrease in ARR at the 96th week in the three subgroups (B).AQP4+ = positive antiaquaporin-4 antibody; ARR = annualized relapse rate; ARR-96 annualized relapse rate at the 96-week follow-up; DN = double-seronegative; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease; NMOSD = neuromyelitis optica spectrum disorder.

However, 61 patients (70.9%) were relapse-free (complete success), 6 (7%) had a relapse while having a decrease in ARR (incomplete achievement), 8 (9.3%) had a relapse without modification in ARR (partial failure), and 11 (12.8%) had a relapse while having an increase in ARR (complete failure) (Table 2).

There was no significant difference in the efficacy of MiTX at week 96 between patients with concomitant treatment (9 with mycophenolate mofetil and 1 with rituximab) and the others (eTable 3, links.lww.com/NXI/A931).

During the entire follow-up, we registered 3 patients (3.5%) with severe secondary adverse events of MiTX: 2 severe but resolutive neutropenia and 1 acute myeloid leukemia (AML), resulting in the patient's death. Moreover, 5 additional patients died, 3 from disability complications and 2 after a dramatic relapse.

Characteristics of Relapsed Patients and Predictive Factors at 96-Week Follow-up

Relapse predictors at 96-week follow-up in univariate analysis are summarized in Table 3. We identified in a multivariate analysis that an AQP4-IgG seropositivity (HR 12.3, 95% CI 1.64–91.6, p = 0.015), a delay ≥24 months from the first attack and MiTX (HR 2.76, 95% CI 1.23–6.17, p = 0.014), and a pretreatment ARR ≥1 (HR 2.38, 95% CI 1.05–5.39, p = 0.037) were predictive factors of new relapse under MiTX (Table 3). Relapsed patients had at least 1 of these 3 predictive factors (eTable 4, links.lww.com/NXI/A932, Figure 4). In our analysis, we showed that the accumulation of these predictors of relapse was significantly correlated with an increased risk of relapse (p < 0.001) (eTable 4, Figure 4).

Figure 4 Relapsed Patients Based on Accumulated Relapse Predictors at 96-Week Follow-up

IRP = identified relapse predictors. *Identified relapse predictors in multivariate analysis: delay between the first attack and MiTX ≥24 months, AQP4-IgG seropositivity, and an ARR the year before MiTX ≥1.

Change in EDSS at 96-Week Follow-up

We recorded a decrease in the mean EDSS from T0 (4.9 ± 2.4) to T96 (4.2 ± 2.6) in overall population (p < 0.001), mainly among the NMOSD-AQP4+ subgroup (from 5.1 ± 2.3 to 4.3 ± 2.6, p = 0.004) (Figure 5). At T96, there were more patients in the minor EDSS group than at T0 (48% vs 42%) (Table 2). Of interest, even in NMOSD-AQP4+ patients with at least 1 identified relapse predictor, there is no increase in mean EDSS-96 (from 5.4 ± 2.1 to 5.0 ± 2.3, p = 0.235) so that we observe a significant decrease in those without relapse predictors (from 4.5 ± 2.5 to 3.1 ± 2.6, p < 0.001) (eTable 5, links.lww.com/NXI/A933).

Figure 5 Evolution of Mean EDSS From the Initiation of Mitoxantrone to the 96-Week Follow-up

AQP4+ = positive antiaquaporin-4 antibody; DN = double-seronegative; EDSS = Expanded Disability Status Scale; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease; NMOSD = neuromyelitis optica spectrum disorder.

Discussion

In our observational cohort study, 71% of patients with NMOSD were relapse-free at 96-week follow-up. This percentage of relapse-free patients was increased to 87% at 96-week follow-up if patients were treated with MiTX from the first attack, regardless of their serologic status. In the relapsed group at 96-week follow-up (n = 25), the median time to relapse was 31 weeks (IQR 18.0–58.0). During this same period, relapsed patients were mainly NMOSD-AQP4+ (23/25), while 95% of NMOSD-DN and 86% of MOGAD patients were relapse-free. At the 96-week follow-up, MiTX allowed a significantly decreased ARR (from 0.85 to 0.32) and EDSS (from 4.95 to 4.24). Moreover, multivariate analysis identified the AQP4-IgG seropositivity, the delay between the first attack and MiTX ≥24 months, and a pretreatment ARR ≥1 as relapse predictors. Of interest, even more at risk of relapses because of their serologic status, NMOSD-AQP4+ patients do not worsen functionally or even improve if they do not accumulate other relapse predictors. During the entire follow-up, the safety profile seems acceptable compared with the disease's poor prognosis, specifically in our patients of African descent who present a more severe clinical phenotype and attacks,30,31 20% of which are resistant to plasma exchange32 or even fulminant leading to death.33 In the 1990s, when high-efficiency IS did not exist, NMOSD had a fatal course in most patients within 5 years in the French West Indies.34

The efficacy of MiTX on the primary endpoint is close to satralizumab in 2 recent double-blind, randomized, placebo-controlled studies, where 78%24 and 72%25 of patients were relapse-free at 96-week follow-up. The efficacy at 96 weeks even seems to be higher (87% of relapse-free patients) if patients are treated with MiTX from the first attack, regardless of their serologic status. According to the serostatus, the proportion of relapse-free patients in the AQP4-IgG seropositive subgroup seems to be higher in satralizumab studies (92% in Yamamura et al.24 and 77% in Traboulsee et al.25) than in our study (61%). In addition, in these studies, there was insufficient evidence of reduced risk of relapse in the AQP4-IgG seronegative subgroup with satralizumab. Conversely, we showed an encouraging signal in our 20 NMOSD-DN patients. In the lack of a prospective randomized trial in this specific condition, our result is interesting given the recent findings in a large cohort of double seronegative longitudinal extensive transverse myelitis, which suggests an early IS because patients seem to show severe outcomes and a high risk of relapse.35 In patients with MOGAD, only 1 presented a relapse under MiTX during the follow-up. These results should be mitigated because of the small number of patients and because patients demonstrated a lower risk of relapses and disability in MOGAD vs NMOSD-AQP4+.36,-,38 In addition, MiTX was started after the first attack in most of our patients with MOGAD (n = 6/7), and the absence of relapse could only be explained by a monophasic course, which is not uncommon.38,39 Last, a double-blind, randomized, placebo-controlled study using eculizumab exclusively in NMOSD-AQP4+ showed a dramatic reduction of relapse risk with 96.4% of relapse-free patients at 96 weeks of follow-up.23 Moreover, the randomized, double-blind, placebo-controlled RIN-1 study also showed a complete absence of relapse at 72 weeks of follow-up in 19 NMOSD-AQP4+ patients treated with rituximab.40 Overall, MiTX seems to show an apparent efficacy in most of our patients, while the choice could turn to eculizumab, rituximab, or satralizumab for NMOSD-AQP4+ patients with an accumulation of relapse predictors.

Starting MiTX quickly in the course of the disease or even starting from the first inflammatory attack seems beneficial for the prognosis of patients in our analysis. To explain these results, we have proposed at least 2 hypotheses: first, patients without high-efficacy treatment before MiTX may expose to more relapses. These attacks could cause long-term injury to the blood-brain barrier (BBB), making it more porous to antibodies and exposing already affected areas to more frequent attacks despite treatment such as the use of MiTX. This hypothesis is supported by the results of a study41 in which cord attacks recurred in the same or contiguous areas in 67% of cases. Second, on the biological level, patients treated late with MiTX may have constituted a more extensive clonal stock of B lymphocytes and plasma cells, exposing the patients to earlier relapses despite the 6 infusions of MiTX. Moreover, with many pre-MiTX relapses, microglia and astrocytes could be overactivated, self-sustaining inflammatory microglia-astrocytes exchanges and hyperreactivity,42 further altering the BBB and leading to an increased risk of relapse. In addition, MiTX may play a specific role in these inflammatory conditions by suppressing astrocyte activation.43

Moreover, we assume to specify that the choice of MiTX can also be encouraged by a meager cost, especially in low-income countries where it would be illusory to see the new molecules given to all patients soon. While administration of new molecules (eculizumab and inebilizumab IV and satralizumab subcutaneously) generally cost several hundred thousand US dollars (USD) per patient yearly throughout their lifetime, approximately 30,000 USD for rituximab, MiTX only costs approximately 1500 USD for 6 infusions. Note that these amounts only represent the direct cost of medications but do not consider the hospital care costs for their administrations, which further increase the bill.44 The accessibility of novel therapies in low-income countries remains poorly known but could be limited. A recent study conducted in Argentina reports significant difficulty in accessing these new molecules, with less than 3% of patients using one of them.45 In addition, the report is not better for older IS, such as mycophenolate-mofetil or rituximab, available in less than 25% of cases in these areas.46 The biggest challenge was the economic cost of treatment, with less than 10% of the population in low-income countries able to afford 1 year of NMOSD medication.46

Regarding secondary endpoints, we observed a striking decrease in ARR (from 0.85 to 0.32, a reduction of 61.3%). These results are similar to the approximately 75% decrease in ARR found in 2 previous studies on MiTX and NMO.16,17 Regarding the evaluation of disability, we showed a decreased EDSS in the global cohort, mainly in our NMOSD-AQP4+ subgroup. Of interest, NMOSD-AQP4+ patients with additionally identified relapse predictors do not worsen functionally despite more relapse. It suggests that relapses are less severe under MiTX in the NMOSD-AQP4+ subgroup and provides valuable information, especially in geographical areas where other recent molecules have very minimal accessibility.

During the entire follow-up (almost 30 years), we observed 3 (3.5%) severe secondary adverse events (including 1 AML). The safety profile does not seem to be inferior to other molecules recently studied in NMOSD (inebilizumab,27 rituximab,40 tocilizumab,26 satralizumab,24,25 and eculizumab23). Moreover, other authors studying MiTX in NMO reported similar results.16,17 In NMOSD, we assume that the benefit of MiTX significantly outweighs the risk of severe secondary adverse events.

Our study has several limitations. First, this is a nonrandomized and noncontrolled study. Second, a few patients (9) had an additional treatment by mycophenolate-mofetil and 1 by rituximab during the 96-week follow-up. It is also the case in studies on satralizumab24 and eculizumab,23 where patients could continue their baseline IS (oral glucocorticoids, azathioprine, or mycophenolate-mofetil, except rituximab) during the follow-up. However, we assume these treatments in a few patients do not constitute a significant bias in interpreting our results. Third, in most cases, our patients have African ancestry, and the generalization of the results to other populations remains unclear. Fourth, we may underestimate the number of patients with MOGAD in our study because MOG antibody titers usually decline with time and could become seronegative with or without immunotherapy.8,47,48 Therefore, some patients diagnosed before 2016 may have become seronegative and considered NMOSD-DN at inclusion in the study. Our study also has strengths. This is the largest prospective cohort of patients with NMOSD treated by MiTX and followed up in a tertiary center over almost 30 years. Moreover, each patient met the latest international diagnostic 2015 criteria4 with updated serologic assays in a French reference center (the University Hospital Center of Lyon, France). Our study was not limited to NMOSD-AQP4+ or seronegative patients, and even if the number of cases is small, this study reports for the first time data of the MiTX in NMOSD-DN and MOGAD patients.

Overall, MiTX could be an effective, safe, and low-cost therapeutic proposal in most NMOSD Afro-descendant patients in geographical areas where access to care is difficult, even in NMOSD-AQP4+ patients, especially in the situation of low pretreatment ARR and a rapid therapy initiation time. Further randomized studies are needed to confirm our results.