Article Text

Download PDFPDF

Research paper
Efficacy of mitoxantrone in neuromyelitis optica spectrum: clinical and neuroradiological study
  1. Philippe Cabre1,
  2. Stephane Olindo1,
  3. Romain Marignier2,
  4. Severine Jeannin1,
  5. Harold Merle3,
  6. Didier Smadja1,
  7. under the Aegis of the French National Observatory of Multiple Sclerosis
  1. 1Department of Neurology, CHRU Pierre Zobda-Quitman, Fort de France, Martinique, French West Indies
  2. 2Department of Neurology A, Pierre Wertheimer Hospital, Civilian Hospices of Lyon, Lyon, France
  3. 3Department of Ophthalmology, CHRU Pierre Zobda-Quitman, Fort de France, Martinique, French West Indies
  1. Correspondence to Dr Philippe Cabre, Department of Neurology, CHRU Pierre Zobda-Quitman, Fort de France, Martinique 97261, French West Indies; pcabre_fr{at}


Objective To evaluate the efficacy of mitoxantrone (MTX) on clinical and neuroradiological parameters of patients who had a relapse of neuromyelitis optica spectrum (NMOS) within the 12 previous months.

Methods MTX (12 mg/m2) combined with methylprednisolone 1 g as three monthly courses followed by three quarterly courses was administered during an observational multicentre open study including 51 consecutive patients (28 NMO, 23 limited forms of NMO) of the French Caribbean and Guyana. The main outcome measure was the reduction of the annualised relapse rate (ARR), and the secondary outcome measures were alteration of disability measured by expanded disability status scale (EDSS) score, the time to onset of the first relapse, and the progression of neuroradiological lesions at 1 year of treatment.

Results At 1 year of treatment, the ARR dropped from 1.82 to 0.37 (p<0.0001). The mean EDSS score improved by 1.3 points, going from 5.8 at baseline to 4.5 at 1 year (p<0.0001). The number of patients showing gadolinium (Gad)+ spinal cord lesions at baseline, that is, 46.9%, dropped to 10.6% (a 77.4% reduction; p=0.02). The median time to onset of the first relapse was 18 months. IgG-NMO seropositivity was a predictive factor of relapse (p=0.006). A case of acute myeloid leukaemia was observed after a mean time span of 4.8 years.

Conclusions In this observational NMO study, MTX decreased dramatically the frequency of relapses, which is directly related to progression of disability or even death in this disorder.

  • MRI

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Neuromyelitis optica (NMO) is a relapsing inflammatory demyelinating disease of the central nervous system (CNS) with episodes of optic neuritis (ON) and longitudinally extensive myelitis (LEM). In the French Caribbean, NMO stands out for its very severe prognosis, leading to a 25% mortality after 10 years of evolution.1 The progression of disability during NMO is directly related to the severity of the relapses, leading on average in the French Caribbean to blindness after two episodes of ON and paraplegia and/or quadriplegia after three episodes of myelitis.1 Preventive treatment for NMO relapses is therefore essential when NMO is diagnosed or even immediately after a first demyelinating event suggestive of NMO. However, immunomodulators such as interferon-β proved useless during NMO,2 and the efficacy of various immunosuppressive drugs relies only on small series of less than 30 cases, mainly retrospective and with short follow-up.3–9

Histopathological data,10 the recent discovery of a highly specific biomarker for NMO called immunoglobulin G (IgG)-NMO, acting against aquaporin-4 (AQP4)11 and the involvement of the B cells activating factor (BAFF)12 call for a role of type B autoimmunity in the pathogenesis of NMO. Mitoxantrone (MTX) is an anthracenedione with immunosuppressive properties that demonstrated its efficacy on very active forms of multiple sclerosis (MS),13 ,14 acting on pathways involved in the inductive and effective phases of the immune-induced demyelinisation and on B cells15 in particular.

We report the results of MTX treatment in an observational multicentre open study on a larger cohort of patients during NMO spectrum (NMOS), the efficacy of which was prospectively established based on clinical and neuroradiological parameters.



Fifty-one consecutive patients of NMOS from three different sites (Martinique, Guadeloupe and French Guyana) were included between 1 January 1999 and 31 March 2010, presenting with either NMO defined following the 2006 criteria (n=28),16 or a limited form of NMO (n=23). Because seronegative limited forms of NMO might include other diseases, we decided to exclude patients with monophasic IgG-NMO-LEM (n=2) as well as IgG-NMO-patients with a single episode of ON suggestive of NMO (n=5). Within limited forms of NMO, the LEM included 14 recurrent LEM and two monophasic LEM seropositive for IgG-NMO. Included bilateral and synchronous ON (n=5) all presented a visual acuity <20/200° in both eyes at the peak of the first episode and were seropositive for IgG-NMO. Two patients affected by recurrent ON with severe sequelae (visual acuity <20/200° at least 6 months after last bout) and seropositive for IgG-NMO were also included. Each patient had to have presented at least one episode of ON and/or LEM in the year prior to the inclusion and an expanded disability status scale (EDSS) score ≤7.0. All patients were informed about the side-effects of MTX and gave informed consent. The study was approved by the committee of protection of individuals of the Caribbean Guyana region.


Treatment consisted in one monthly injection of 12 mg/m2 of MTX combined with 1 g of methylprednisolone (MP) for 3 months and then three quarterly courses with the same dosage. The treatment protocol combining MTX and 1 g MP (six courses) was adapted from the one used in very active MS.14 Granulocyte colony stimulating factor was not used as prophylactic or curative treatment of neutropenia in any patient. The therapeutic target of 72 mg/m2 was met in 41 patients. Five patients shown to be non-compliant, because of adverse events in three cases, received a cumulative dose of MTX between 36 and 60 mg/m2. Four patients received 36 mg/m2 of MTX due to a lack of control of relapses in three cases and a persistent infection in one case. One patient received 48 mg/m2 of MTX due to the development during treatment of a systemic lupus erythematosus, leading to switching from the main treatment to rituximab. Two patients who had experienced severe relapses during treatment were transitioned to cyclophosphamide after completion of the 1 year MTX protocol. The mean cumulative dose of MTX given during the study was of 107 mg (±28 mg).

Clinical evaluation

The patients were clinically evaluated every 6 months during the year prior to their inclusion, at the moment of each MTX injection (M1–M2–M3–M6–M9–M12), and then on an annual basis. For at least 1 year, the relapses, the EDSS scores pre- and post-MTX treatment, as well as tolerance to the treatment, were collected prospectively and input into the European Database for Multiple Sclerosis. The number of patients that completed 1, 2, 3, 4, 5, 6, 7 and 8 years of follow-up were, respectively, 51, 41, 36, 30, 25, 18, 15 and 12. The mean follow-up was 4.8 years (±3.1) which corresponded to the duration of observation after MTX treatment. Among the 51 patients, two were lost to follow-up at, respectively, 2 and 3 years. The left ventricular ejection fraction was evaluated by myocardial scintigraphy before treatment and then annually for 5 years.

Radiological and biological evaluation

MRI of the spinal cord was performed by means of the same scanning pulse sequencing protocol, and 1.0 T MRI platform. The MRI of the spine included cervical and thoracic sagittal and axial 3-mm fast spin-echo T2-weighted images, and pre-contrast and post-contrast sagittal T1-weighted images using gadolinium (Gad) (0.1 mmol/kg intravenously). A spinal cord MRI was performed at baseline and 1 year after MTX treatment onset. The neuroradiological lesions were blindly evaluated by an experienced neuroradiologist on the T2 sequence according to the following classification: major lesions (A1=≥3 vertebral segments with oedema, A2=≥3 vertebral segments without oedema, A3=cavitary lesions, A4=atrophic lesions); and minor or no lesions (B1=multifocal ≤3 vertebral segments, B2=monofocal ≤3 vertebral segments, C=absence of lesions).17 Gad intake by the lesions was noted. Overall, 49 MRI were available at baseline and 47 one year later. We did not assess brain and optic nerve lesions during follow-up in the present study. After informed consent, we could make use of 46 serums at inclusion for the dosage of IgG-NMO. Three techniques were used: indirect immunofluorescence on a substrate of murine cerebellum and brainstem,18 cell culture assay on HEK 293 cells transfected by the AQP4-M1 plasmid,19 and fluorescent immunoprecipitation.20 A patient was considered seropositive if the results were positive in at least one of the three techniques.

Criteria for treatment efficacy

The main measure to assess for MTX efficacy was the annualised relapse rate (ARR), before and after treatment. The secondary criteria were the evolution of the EDSS score under treatment that was assessed in the whole NMOS cohort, the median relapse time, the predictive factors of early relapse, as well as the neuroradiological features 1 year after treatment onset.

Statistical analysis

Treatment analysis was done on intention to treat for the whole cohort of 51 NMOS patients. The clinical markers of disease progression (ARR and EDSS) were computed at the end of each follow-up year, up to 5 years, and compared to the value corresponding to the same patients and calculated during the year prior to treatment by the Wilcoxon rank test. The median time to the first relapse after the first MTX infusion was estimated with the Kaplan–Meir method, and the predictive factors for relapse were estimated by univariate analysis using the Mantel–Cox test. The χ2 test of MacNemar was used to compare the percentage of patients presenting major lesions and the effect of the contrast on MRI (Gad+ lesions) at baseline and after 1 year of MTX treatment. All statistical analyses were done with Stat View 4.0 software using a threshold of significance at a p value <0.05.


Patient characteristics

The characteristics of NMOS patients treated with MTX are presented in table 1. There were 94 relapses in the year prior to MTX treatment (median 2 months, range 0–12). Twenty-two patients (43.1%) experienced a relapse within 30 days before MTX treatment. Thirty-eight patients (74.6%) were naïve of disease modifying therapies. The remainding patients had received either interferon-β (n=7) or cyclophosphamide (n=6) prior to administration of MTX.

Table 1

Characteristics of the cohort of patients presenting a with NMOS

Clinical efficacy of the treatment

The ARR was significantly reduced from 1.82 (during the year prior to MTX treatment) to 0.37 one year after starting MTX (p<0.0001), corresponding to an 80% reduction in the frequency of relapses (figure 1). The ARR decreased to 0.55 at 6 months of treatment (p<0.0001). The lower frequency of relapses was maintained up to 5 years of follow-up. The mean EDSS score was significantly improved by 1.3 points, going from 5.8 at inclusion to 4.5 at 1 year (p<0.0001). The significant improvement of the mean EDSS score compared to baseline was maintained up to 5 years of follow-up (the mean EDSS measured was 4.5 at 2 years, p=0.0002; 4.7 at 3 years, p=0.0009; 4.7 at 4 years, p=0.018; and 4.8 at 5 years, p=0.0012). After exclusion of patients who had a relapse within 1 month before starting MTX treatment, significant improvement of EDSS was observed at 1 year (5.6 to 4.6; p=0.002) that was maintained up to 3 years of follow-up. The median time to the first relapse after starting MTX treatment was 18 months (SD=5.1) for the whole cohort. The number of patients showing relapse at 1 year of treatment was 15 (29.4%) without correlation, with an incomplete protocol of MTX treatment (figure 1). Attacks in these patients had mild severity except for two patients whose EDSS increased at 1 year of follow-up. The predictive factors of relapse were an IgG-NMO seropositivity (median time of relapse of 14 months in case of positivity versus 48 months in case of negativity; p=0.006, figure 2). Table 2 shows that the differences in term of frequency of relapses between the IgG-NMO positive and IgG-NMO negative patients occurred significantly 2 years after MTX induction. There was a trend for an intake of Gad on spinal cord MRI at baseline to predict therapeutic failure (p=0.07). No clinical factor (gender, NMOS subtype, typology of the first relapse, recovery from the first relapse, time of progression, age at time of treatment, association with another autoimmune pathology, ARR before treatment) was related to a therapeutic failure. Among patients with limited forms NMO, the disease converted into definite NMO in five patients despite IgG-NMO seronegativity in two of them. Among the 18 patients (35.3%) who experienced optic attacks during follow-up, blindness occurred in a single case. Four patients died: two from respiratory failure due to a lack of control of relapses that occurred 2 years and 4 years, respectively, after starting MTX; one from pulmonary embolism during a severe spinal cord relapse just after completion of MTX treatment; and one from acute myeloid leukaemia related to MTX treatment.

Table 2

Impact of MTX treatment according to IgG-NMO status

Figure 1

Reduction in annualised relapse rate (ARR) up to 5 years after treatment onset. The ARR was significantly reduced by 80% during the first year of follow-up compared to 12 months prior to mitoxantrone (MTX) treatment. The magnitude of this reduction was maintained up to 5 years. n=number of patients followed during 1, 2, 3, 4 and 5 years; *p<0.0001.

Figure 2

Median time to the first relapse after mitoxantrone treatment depending on immunoglobulin G (IgG)-neuromyelitis optica (NMO) seropositivity (method of Kaplan–Meier) in the whole cohort of NMOS. The median time to the first relapse was 14 months in the case of IgG-NMO seropositivity and 48 months in the case of IgG-NMO seronegativity.

Efficacy of the treatment on neuroradiological lesions

During the inclusion of MTX treatment, the most frequent lesions on the spinal cord MRI were the major lesions of type A1 and A2. There were almost no minor lesions of B1 and B2 type. One year after treatment onset, the most frequent lesions at baseline, that is, 77.6%, significantly decreased to 51% at 1 year of treatment (p=0.002, figure 3). The number of patients presenting Gad+ spinal cord lesions at inclusion, that is, 46.9%, dropped to 10.6% (a 77.4% drop) at 1 year of MTX treatment (p=0.02). Figure 4 shows MRI features of two patients at baseline and after 1 year of treatment with MTX.

Figure 3

Progression of spinal lesions at MRI. A1, ≥3 vertebral segments with oedema; A2, ≥3 vertebral segments without oedema; A3, cavitary lesions; A4, atrophic lesions; B1, multifocal ≤3 vertebral segments; B2, monofocal ≤3 vertebral segments; C, absence of lesions. The major lesions included lesions of type A1, A2, A3 and A4. Minor lesions or no lesions included lesions of type B1, B2 and C. *p=0.002, **p=0.02.

Figure 4

Spine MRI of two patients showing the response to mitoxantrone (MTX). Gad indicates post-gadolinium image; T2, T2-weighted image. (A) Images at far left were obtained immediately before the start of MTX therapy. Note the extensive enhancing lesion in the cervical spinal cord associated with oedema (major lesion of type A1). Images obtained 1 year after initiation of therapy show no oedema but the lesion is still ≥3 vertebral segments (major lesion of type A2); Gad enhancement is minimal (arrow). (B) Images at baseline show extensive oedematous lesions in the cervical cord and in the thoracic cord (major lesion of type A1); Gad enhancement is seen at C5–C6 level (arrow). Follow-up MRI images show minor lesion of type B1 (multifocal ≤3 vertebral segments: arrows); however, there is still Gad enhancement at C4, T1 and T6 levels (arrows).

Side-effects of the treatment during follow-up

One case of reduced left ventricular ejection fraction below 50% was observed in this population on myocardial scintigraphy performed annually for 5 years. MTX side-effects were observed in 15 patients (29.4%). Moderate emesis during the courses of treatment occurred in four cases (7.8%). The most frequent side-effect was an affectation of appendages, including alopecia grade 1 in eight cases (15.6%), and leading to premature termination of treatment in one case and bluish nails coloration in five cases. Six female patients (13%) had secondary amenorrhoea, transitory in two cases and persistent in four cases (mean age 44 years), leading to discontinuation of treatment in two cases. A case of deep neutropenia (<500/mm3) was observed without an associated infectious event; it only delayed the following MTX injection by a few days. A skin infection with Staphylococcus aureus necessitated cessation of treatment. A case of type 1 acute myeloid leukaemia was observed. The patient was 23 years old, seropositive, and presented with a multi-refractory fulminating form of NMO despite MTX, cyclosphosphamide and rituximab treatment. Leukaemia appeared 30 months after treatment onset and was related to 17–19 translocation, causing death of the patient by septicaemia during induction of spinal cord aplasia, despite attempted bone marrow xenotransplantation.


During this multicentre observational study including 51 patients with NMOS, MTX treatment with six injections of 12 mg/m2, combined with 1 g MP resulted in a significant reduction of 80% ARR after 1 year. Moreover, after 1 year of follow-up, MTX treatment resulted in a decrease of the EDSS score by 1.3 points (5.8 to 4.5), and a reduction of Gad+ lesions (77.4%) and of extensive lesions on the spinal cord MRI. The therapeutic benefit of MTX was maintained on the clinical parameters at 5 years of treatment. Our study is to date the most important achievement during NMOS by an immunosuppressive agent following a standardised protocol, neuroradiological monitoring and significant follow-up. In two previous studies,8 ,21 MTX administered during NMO had beneficial effects on the clinical parameters following heterogeneous protocols. Our results are similar to those from a recent retrospective study including 20 NMOS patients who received MTX by monthly cycles of 12 mg/m2 from 3 to 6 months, and then by quarterly cycles of 6–12 mg/m2. In this study, after an average follow-up of 41 months, the ARR was reduced by 75% and the EDSS score dropped from 5.6 to 4.4.9

MTX exerts an inhibitory action on antigen presenting cells22 and induces a preferential apoptosis of B cells. However, it exerts a different type of action on B-lymphocyte subtypes, by favouring the production of the IL-10 regulating cytokine by naive B-lymphocytes, and a decrease of production of pro-inflammatory cytokines, such as TNF-α and lymphotoxine by memory B cells.23 Given the prevailing role of B-type immunity during NMO, it becomes the molecule of choice for this condition.

A retrospective study in nine NMO patients24 reported a significant reduction of relapses with daily use of at least 10 mg of oral corticosteroids after a median follow-up of 19 months. In addition, the superiority of the combination monthly MP+MTX versus MP on inflammatory focal lesions within the CNS was dramatically demonstrated on clinical and MRI grounds in MS patients with very active disease.25 In NMO, this comparison has not been made formally, but for these reasons we decided to combine MTX and MP for six courses in our observational study.

Nevertheless, MTX in our study did not allow a remission of the NMOS in all patients, and median time of relapse was 18 months in the whole cohort of NMOS. Interestingly, IgG-NMO seropositivity was a predictive factor of early relapse in the whole cohort. Several research studies outlined an increased severity of seropositive NMO.26 ,27 Our study thus shows that IgG-NMO is not only a pejorative prognostic factor, but also a resistance factor to treatment. In this observational study, patients did not receive an alternative immunosuppressant drug after the 1 year treatment with MTX except for a minority of patients that were very poor responders. However, given differences in ARR that occurred at 2 years of follow-up between IgG-NMO positive and IgG-NMO negative patients, we recommend using systematically, in NMOS seropositive patients, maintenance therapy following an induction treatment of MTX with mycophenolate5 or azathioprine,6 ,28 which have been shown to be partially effective in reducing relapses in NMO. Our study lacks a control group, so we are unable to state without ambiguity that clinical and neuroradiological efficacy is related to MTX. However, a study against placebo is ethically difficult to carry out in a life-threatening disease like NMO. In addition, the reduction of frequency of relapses (−80%) induced by MTX is of a magnitude too large and superimposed on that obtained during controlled trials in aggressive forms of MS,25 to be attributed to the natural decrease of the progressive character of NMO over time. Short-term improvement of EDSS scores might be confounded by attacks that occurred near the beginning of MTX but could also be a direct consequence of MTX treatment or the treatment of relapses by either steroids or plasma exchanges. In NMO, the residual disability is the consequence of the accumulation of relapses and not partially (as in MS) the consequence of a progressive neurodegenerative process. We therefore speculate that long-term improvement of EDSS (ie, up to almost 5 years) is linked in our study to the ability of MTX induction treatment to prevent long-term relapse occurrence. The 77.4% reduction in Gad+ lesions on the spinal cord MRI in our study is equally consistent with the results observed for encephalic lesions where MTX is used for treatment of aggressive forms of MS,25 but can hardly be discussed in the lack of sequential longitudinal data during NMO and LEM.

Minor side-effects of MTX—nausea, leucopenia, alopecia and secondary amenorrhoea29—were found in our study. The two major side-effects—cardiotoxicity, including a 0.4% incidence of congestive heart failure30—were not observed in our study. Two recent studies assessed the risk of MTX induced leukaemia at 0.8%31 and 0.93%,32 with a mean time of 18.5 months and 49 months, respectively. However, the only prospective study gave a lower rate (0.25%33), with a mean follow-up of 60 months and a cumulative dose of 72 mg/m2. In our case of acute myeloid leukaemia that occurred 30 months after MTX treatment onset, administration of cyclophosphamide might have increased this risk.

In conclusion, we have shown in an observational open multicentre study that MTX can be offered as a reasonable treatment in patients with NMOS, resulting in a significant reduction of its clinical and radiological activity parameters, and that the risk–benefit ratio advocates its use in this severe disease, which threatens the functional and vital prognosis in patients who are usually young.


The authors thank Professor Herve Deramon for MRI reevaluation and analysis of MRI follow-up investigations; Annie Lannuzel MD, PhD and Anne Landais MD, for giving access to their patients and for creating facilities for patient investigation; the Department of Neuroradiology of Pierre Zobda-Quitman Hospital for MRI imaging; and Sophie Domingues PhD for help in writing the manuscript. Statistical analysis was performed by Dr S. Olindo at the Department of Neurology, CHRU Pierre Zobda-Quitman, 97261, Fort de France, Martinique; French West Indies.



  • Contributors PC: study concept and design, acquisition of data, interpretation of results, and writing of manuscript. SO: statistical analysis, interpretation of results, and approving final version. RM: laboratory determination of aquaporin-4 antibodies and approving final version. SJ: acquisition of data and approving final version. HM: acquisition of data and approving final version. DS: study concept and design, interpretation of results, and approving final version.

  • Competing interests PC has received funding for travel from Biogen Idec and receives research support from the Pierre Zobda-Quitman Hospital Research Fund.

  • Ethics approval Committee of protection of individuals of the Caribbean Guyana region.

  • Provenance and peer review Not commissioned; externally peer reviewed.