Article Text

Download PDFPDF

Research paper
Dimethyl fumarate: a possible exit strategy from natalizumab treatment in patients with multiple sclerosis at risk for severe adverse events
  1. Massimiliano Calabrese1,
  2. Marco Pitteri1,
  3. Gabriele Farina1,
  4. Albulena Bajrami1,
  5. Marco Castellaro2,
  6. Roberta Magliozzi1,3,
  7. Salvatore Monaco1
  1. 1 Neurology B, Department of Neurosciences, Biomedicine and Movement, University of Verona, Verona, Italy
  2. 2 Department of Information Engineering, University of Padova, Padova, Italy
  3. 3 Division of Brain Sciences, Faculty of Medicine, Imperial College London, London, UK
  1. Correspondence to Professor Massimiliano Calabrese, Neurology B, Department of Neurosciences, Biomedicine and Movement Sciences, University of Verona, Policlinico “G.B. Rossi” Borgo Roma, Piazzale L A Scuro, Verona 1037134, Italy; massimiliano.calabrese{at}


Introduction Among disease-modifying treatments for multiple sclerosis, natalizumab (NTZ) is highly effective, well tolerated and generally safe. Major concerns regard the risk of developing progressive multifocal leukoencephalopathy (PML), and the occurrence of rebounds or disease activity after its discontinuation. The aim of this study was to explore the efficacy of dimethyl fumarate (DMF) in preventing disease reactivation after NTZ discontinuation.

Methods Thirty-nine patients with relapsing remitting multiple sclerosis, at high risk of PML, were switched from NTZ to DMF and underwent neurological and 3T MRI monitoring for 2 years. Clinical and MRI data regarding the 2-year period preceding NTZ treatment, the 2 years of NTZ treatment and the 2 years of DMF were collected.

Results During the DMF phase, among the 39 patients, one or more relapses occurred in five patients (12.8%), increased disability progression in 4 (10.3%) and MRI activity in 8 (20.5%). Post-NTZ rebound effect was observed only in one patient. Overall, only two dropouts (one rebound activity and one gastrointestinal side effect) were registered and almost 80% of the patients have still no evidence of disease activity at the end of DMF treatment. The multiple linear regression model revealed that the number of relapses and MRI parameters before DMF treatment were good predictors of disease activity during treatment with DMF.

Discussion DMF appeared generally safe and no carryover PML among investigated cases was observed. Although DMF did not eliminate the possibility of disease reactivation, it seems anyway a promising drug for those patients who shall discontinue NTZ. The clinical and radiological activity preceding the DMF treatment might be used as a prognostic marker of therapy response.

Statistics from


Natalizumab (NTZ) is a therapeutic option particularly effective for treating patients with highly active multiple sclerosis (MS) disease.1–3 NTZ is a recombinant humanised monoclonal antibody that targets the α4 subunit of α4β1 leucocyte integrin,1 an adhesion molecule which interacts with the endothelial and extracellular matrix ‘vascular cell adhesion molecule 1’ and mediates the passage of lymphocytes into the central nervous system (CNS). NTZ suppresses leucocyte adhesion and entry in the CNS, which explains its therapeutic effects in MS.

Although several studies demonstrated the efficacy and the good tolerability profile of NTZ in patients with MS, its long-term safety is burdened by the risk of progressive multifocal leukoencephalopathy (PML), an uncommon but potentially fatal opportunistic infection caused by JC virus (JCV), which most likely results from compromised brain immunosurveillance.4 5 Among JCV-seropositive patients, the risk of PML is particularly high,6 and interruption of NTZ therapy has to be considered, especially in patients with cumulative NTZ treatment longer than 24 months, and previous exposure to immunosuppressive therapies or anti-JCV antibody index >1.5.5

A major concern with NTZ discontinuation is the risk of disease reactivation7–9; indeed, several studies have shown that clinical and neuroradiological disease activity increased shortly after stopping NTZ, peaking at 3–4 months after the last infusion.10 11

Several strategies have been advocated to reduce the risk of disease reactivation after NTZ discontinuation, and the requirement for a washout period before switching to an alternative treatment has even been questioned. Available studies have shown that but neither IFN-β nor glatiramer acetate (GA) yielded positive results.12 Even the alternative treatment with monthly prednisolone followed by GA was ineffective in preventing the reemergence of inflammatory activity,13 whereas the combination of GA plus pulsed steroid administration was associated with worse clinical and radiological outcome than the GA alone.12 Several studies14 15 have analysed the efficacy of fingolimod (FTY) after NTZ discontinuation, showing that FTY does not completely stop disease reactivation and its effect seems to be influenced by the different length of the washout period.14–16 There are also studies showing better effectiveness and tolerability of anti-B-cell therapy, for example, rituximab compared with FTY in stable patients with relapsing remitting (RR) MS switching from NTZ due to JCV antibody positivity.17

Dimethyl fumarate (DMF) is an orally administered drug, which has been recently approved for the treatment of RRMS. DMF acts through stabilisation and nuclear translocation of the transcription factor Nrf2, which leads to downstream activation of a cascade of several cytoprotective and antioxidant pathways, as well as to the shift from proinflammatory to anti-inflammatory immune responses.18–20 The high efficacy of DMF in downregulating T-cell, B-cell and myeloid cell proinflammatory responses has been shown to highly contribute to the efficacy of this therapy in RRMS, since early disease stages.21–24 However, the mode of action of DMF is still not fully understood and its efficacy, as compared with that of existing neuroprotective and immune-modulating therapies, remains unsettled.

Despite the encouraging data of DMF in MS treatment,25 26 little is known about its efficacy when transitioning from NTZ to an oral agent is dictated in patients at risk of developing PML. To the best of our knowledge, the only previous attempt concluded that DMF would not be protective against disease reactivation after NTZ discontinuation, since 27% of patients with RRMS experienced a relapse within 3 months after NTZ cessation.27 However, the lack of any other similar studies, the small sample size and the long NTZ washout period before DMF therapy leave the question of feasibility of DMF after NTZ discontinuation still open.

The aim of the present longitudinal study was to assess the efficacy of DMF in reducing disease activity after a short period of NTZ discontinuation in a homogeneous cohort of NTZ-treated patients at high risk of developing PML.

Materials and methods


Thirty-nine patients with RRMS at high risk for developing PML, having high titre of JCV antibodies (anti-JCV antibody index more than 1.5), in treatment with NTZ for more than 2 years, and, in seven cases, with a pre-NTZ history of immunosuppressive therapy, have been enrolled in this study (see table 1 for details).

Table 1

Demographic, clinical and MRI characteristics of the studied population during different treatment periods

Each patient has been switched from NTZ to DMF after a washout period of about 1 month (mean 33.7±14.0 days), and underwent longitudinal, clinical and radiological follow-up through 24 months of DMF treatment (mean 24.0±4.5 months). Patients received DMF at a dose of 120 mg twice daily for 1 week, with subsequent escalation to 240 mg twice daily. All relevant clinical and MRI information about the 2 years pre-NTZ treatment and the first 2 years of NTZ treatment was retrospectively collected and analysed. Treatment preceding NTZ consisted in IFN-β1a (18 patients), IFN-β1b (5 patients), GA (11 patients), FTY (6 patients) and immunosuppressive therapy (cyclophosphamide, 7 patients).

For each patient, the number of relapses and the confirmed progression of disability, defined as an increase of at least 1.0 point on the Expanded Disability Status Scale (EDSS)28 score, sustained over 6 months, were evaluated. We considered a time period of 6 years overall, divided into three time periods (time points): the 2 years preceding NTZ treatment (T0), the 2 years of NTZ treatment (T1) and the 2 years of DMF treatment (T2).

This study was approved by the local Ethic Committee and the informed consent was obtained from all patients.

MRI acquisition and analysis

All patients underwent 3T MRI scan (Philips Achieva, Philips Medical Systems, Best, The Netherlands) at the beginning of DMF, after 3 months and then every 6 months during the DMF treatment.

The 1.5T brain MRI assessment (Philips Achieva, Philips Medical Systems), obtained every 6 months before and during NTZ treatments, was retrospectively collected.

No major hardware upgrades were carried out on the MRI scanners (both 1.5T and 3T) during the course of the study.

3.0T MRI protocol

  1. Three-dimensional fluid attenuated inversion recovery (FLAIR) repetition time (TR)/ echo time (TE)=5500/292 ms, inversion time (TI)=1650 ms, voxel dimension of 1×1×1 mm, matrix=256×256

  2. Three-dimensional double inversion recovery (DIR) TR/TE=5500/292 ms, TI1/TI2=525/2530 ms voxel dimension of 1×1×1 mm

  3. Three-dimensional T1-weighted fast field echo (FFE) TR/TE=8.4/3.7 ms, voxel dimension of 1×1×1 mm, matrix=256×256

  4. Two-dimensional T1-weighted spin echo post gadolinium: TR/TE=550/10 ms, 50 contiguous axial slices with a thickness=3.0 mm, matrix=256×256

1.5T MRI protocol

  1. Three-dimensional DIR: TR/TE=15 631/25 ms, TI=3400 ms, delay=325 ms, echo train lenght (ETL)=17, 50 contiguous axial slices with a thickness=3 mm, matrix size=130×256, and field of view (FOV)=250×200 mm

  2. Three-dimensional FLAIR: TR/TE=10 000/120 ms, TI=2500 ms, ETL=23, 50 contiguous axial slices with a thickness=3.0 mm, matrix size=172×288, and FOV=250×200 mm

  3. Three-dimensional T1-weighted FFE: 120 contiguous axial slices with the off-centre positioned on zero, TR/TE=25/4.6 ms, flip angle=30°, slice thickness=1.2 mm, matrix size=256×256, and a FOV=250×250 mm

  4. Two-dimensional T1-weighted spin echo post gadolinium: TR/TE=550/10 ms, 50 contiguous axial slices with a thickness=3.0 mm, matrix=256×256

MRI image analysis

All MRI images were assessed by consensus by two experienced observers, who had no major disagreement in image interpretation. The number of new or newly enlarging T2 MRI lesions including cortical lesions (evaluated on DIR images) and the number of Gd-enhancing lesions were evaluated for each MRI scan.

No evidence of disease activity (NEDA-3) evaluation

No evidence of disease activity (NEDA-3) during DMF treatment was also considered. NEDA-3 is a composite score obtained from three related measures of disease activity: (1) no evidence of relapses, (2) no confirmed disability progression as assessed by an increase of the EDSS score by at least one point sustained over 6 months and (3) no MRI activity (new or newly enlarging lesions in T2 derived sequences including DIR).29 30

In addition to NEDA-3, in this study we also calculate a ‘cumulative activity score’, assigning to each measure of disease activity (the evidence of relapse, the disability progression and any MRI activity) the value of 1.

Therefore, we identified four classes of patients with scores from 0 (i.e., NEDA-3) to score 3 (evidence of disease activity in all of the three parameters). With this classification, we evaluated the severity of disease activity using a more sensitive scale with respect to the classification of NEDA-3 dichotomous scale.

Statistical analysis

Differences in terms of number of relapses, EDSS change and MRI metrics have been evaluated among the three different treatment periods: the 2 years before NTZ treatment (T0), the 2 years of NTZ treatment (T1) and the 2 years of DMF treatment (T2). Post hoc analysis was considered between pre-NTZ and NTZ treatment, and between NTZ and DMF treatment.

We used the Friedman test to evaluate the difference among the three time points (pre-NTZ, NTZ, DMF) of the selected variables (EDSS, number of relapses and MRI activity). Post hoc analysis with the Wilcoxon-signed rank test was applied to assess multiple comparisons.

A χ2 test was applied to test the difference between NTZ and DMF treatment in terms of number of patients with MS who had relapses or new MRI lesions or who were considered NEDA-3.

Multiple linear regression analysis was used to test the role of different variables (number of relapses, MRI activity, EDSS, actual age, disease duration) before DMF as predictors of severity of disease activity (severity of disease activity score or severity score) at the end of DMF treatment.


Number of relapses

During the 2 years before NTZ treatment, 39 out of 39 patients (100%) had at least one relapse; on the contrary, during NTZ treatment 3 out of 39 patients (7.7%) had at least one relapse (all of which occurred within the first 6 months of the NTZ treatment). During DMF treatment, 5 out of 39 patients (12.8%) had one or more relapses (χ2 between NTZ and DMF treatment p=0.287). Moreover, among the total six relapses observed, only three caused a disability progression confirmed at 6 months. For a comprehensive overview of the number of active patients, see figure 1.

Figure 1

The diagram shows the clinical course of the examined patients with MS: for each patient with MS, the clinical events/assessments including stable MRI scan (green square), active MRI scan (yellow square) and clinical relapse (red oval) are reported. Furthermore, blue asterisks are associated to each non-NEDA3 patient and continuous lines represent stable EDSS. In contrast, the different degree of the line step represents the different EDSS increase. DMF, dimethyl fumarate; EDSS, Expanded Disability Status Scale; MS, multiple sclerosis; NEDA, not evidence of disease activity; NTZ, natalizumab; YRS, years. This image is produced in colour online as part of the issue.

Disability progression confirmed at 6 months

In the 2 years before NTZ, 23 out of 39 patients (58.9%) had an increase in disability; on the contrary, during NTZ only 1 out of 39 patients (2.6%) had an increase in disability; finally, during DMF, 3 out of 39 patients (7.7%) had disability progression. The increase in disability was higher during pre-NTZ period than during NTZ or DMF treatment (p<0.001), but not between NTZ and DMF treatment (p=0.180).

MRI activity

Before NTZ treatment, 39 out of 39 patients (100%) had new or enlarged lesions in T2-derived sequences (including DIR) or T1-Gd+lesions (mean=4.2±3.5); during NTZ, 6 out of 39 patients (15.4%) showed new or enlarged lesions (mean=0.31±0.80); during DMF treatment, 8 out of 39 patients (20.5%) showed new or enlarged lesions (mean=0.7±2.1; χ2 between NTZ and DMF treatment p=0.555). The post hoc analysis on the number of new or enlarged lesions showed a significant difference between pre-NTZ and NTZ treatment (p<0.001), but not between NTZ and DMF treatment (p=0.282).

Dropouts and rebounds

Among the 39 patients, two dropouts were observed. One of these overshot the disease activity of pre-NTZ treatment (ie, rebound activity) showing two clinical relapses in few months with disability progression and more than 10 new MRI lesions. The other one dropped out for gastrointestinal side effects.

NEDA-3, cumulative activity score and regression analysis

No NEDA-3 patients were observed before starting NTZ treatment; 33 out of 39 (84.6%) patients were NEDA-3 at the end of NTZ treatment and 31 out of 39 (79.5%) patients were NEDA-3 at the end of DMF treatment (NTZ vs DMF p=0.555) (see figure 1).

Among the eight patients who were not NEDA-3 at the end of DMF treatment, one (2.6%) showed a cumulative active score of 1, five (12.8%) showed a cumulative active score of 2 and two patients (5.2%) showed a cumulative active score of 3. Moreover, the mean number of relapses in the 2 years preceding the NTZ treatment was significantly higher in the non-responder to DMF (3.4±0.5) compared with the group of responders (2.3±0.9, p=0.002).

The multiple linear regression model, with the severity score as depending variable, revealed that the number of relapses (p<0.001) and the MRI activity (p<0.001) before DMF treatment were significant predictors of disease activity also during the DMF treatment.


The primary and compelling goal of this study was to assess the efficacy of DMF in reducing disease activity after a short period of NTZ discontinuation in a homogeneous cohort of NTZ-treated patients at high risk of developing PML. As such, these patients were at high risk for the foregoing opportunistic infection, especially in regard to the subset with prior immunosuppressive drug use. Although it was not the aim of the study, as the small sample size and the short follow-up, we would like to point out that as far as we can see by our 3T MRI scans no PML signs were observed among investigated cases; if these data would be confirmed during the follow-up and also in a larger sample size, it could suggest that treatment sequencing from NTZ to DMF does not represent a further and synergic risk for PML.3

However the main result from this longitudinal study on 39 patients with RRMS is the evidence that DMF could be a promising exit strategy for JCV-seropositive patients under treatment with NTZ and at high risk for severe adverse events. Although DMF does not seem to entirely control disease reactivation, after 2 years of treatment almost 80% of the patients were still NEDA-3 and only one patient, after NTZ discontinuation, showed a disease activity suggestive of a rebound. The paucity of clinical and radiological reactivations (as also highlighted by the cumulative activity score) suggests that DMF has a good efficacy in maintaining low levels of disease activity after NTZ discontinuation. It has to point out that despite no significant differences in disease activity and disability progression between NTZ and DMF phases can be observed, this can be partially explained by the low number of patients included in the study. Indeed, there is a consistent trend indicating less efficacy in both clinical and MRI measures in the DMF phase.

Nevertheless, the good results of DMF could be explained by similar activity to NTZ. In particular, it may be suggested that the reduction of Th1/Th17, myeloid and B cell which is one of the immunomodulatory effects of NTZ,24 27 31 might be further maintained by DMF, which was shown to be able to favour the Th2 regulatory and myeloid cell activity and to reduce the levels of circulating T and B cells.21–24 32 It may therefore be hypothesised that DMF may perpetuate the effect of NTZ either in the modulation of adhesion molecule expression on human leucocytes and their rolling features in vivo (directly affecting the leucocyte extravasation that can occur after NTZ suspension) or in maintaining the shift of the balance of the immune responses towards anti-inflammatory components.

Our results partially differ from those of the only previous study27 which has evaluated the possible efficacy of DMF following NTZ discontinuation. In that study, almost 30% of patients had relapses after switching to DMF, 17% suffered severe relapses with multifocal clinical and radiological findings, and new MRI lesions were observed in 35% of patients. However, the majority of reactivations occurred within the first 3 months after NTZ cessation. On the contrary, in this study the washout period between NTZ and DMF was significantly lower (about 1 month). Our hypothesis is that a short washout time can play a crucial role in reducing the risk of restoring leucocyte trafficking into the CNS, and, in turn, focal inflammatory activity, ensuing clinical and MRI reactivation. In accordance with previous authoritative recommendation,33 the results of this study suggest that a washout period of about 1 month should be the gold standard when switching from NTZ to an alternative disease modifying drug, in order to prevent MS reactivation or an immune reconstitution effect.

The analysis of the clinical and MRI predictors of the DMF treatment revealed that the disease activity (number of relapses or MRI activity) before and during NTZ treatment might represent the best predictor factor of the possible failure of DMF treatment. Patients with the most active disease before and during NTZ treatment also showed disease activity during DMF treatment. In this regard, recent studies suggested that DMF can influence the expression of adhesion molecules, changing the capability of lymphocytes to enter the blood–brain barrier.34 This could explain why good responders patients to NTZ are also good responders to DMF, especially if the washout period is appropriate in assisting NTZ-to-DMF ‘handover’. This finding suggests at least two important considerations: (1) a very active disease before NTZ or a suboptimal response to NTZ should encourage clinicians to switch to another, more aggressive treatment; (2) although the main mechanisms of action of NTZ and DMF are different, some similarities can be found.

We are aware that this study is not without limitations: among these, the most important depends on the small sample size, the short follow-up, and the retrospective collection of the NTZ and pre-NTZ data. Considering these limitations, additional data on larger populations with a longer follow-up are required to substantiate this approach in more standardised way in clinical practice. Finally, a possible limitation could be the different MRI protocol applied to monitor the NTZ and DMF treatment; however, the more accurate high-field 3T MRI protocol has been applied to monitor the efficacy of DMF while the 1.5T MRI protocol has been applied to monitor the NTZ phase. Therefore, considering the higher sensitivity and specificity of the 3T MRI in detecting disease activity, this should reinforce rather than weaken the present results.

In conclusion, for both modulating immune activity and regulating leucocyte migration into the CNS, DMF may represent an appropriate exit strategy after NTZ discontinuation at least in those patients showing low-to-moderate disease activity preceding or during NTZ treatment. Patients with very high disease activity should be encouraged to switch to other drugs.



  • Contributors MC was responsible for the study conception; MC, MP, GF and AB implemented and executed the data collection; MP and MC contributed to the methodology and data analysis. All the authors contributed to the manuscript preparation and revision.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Comitato etico per la Sperimentazione Clinica delle Province di Verona e Rovigo con sede presso AOUI di Verona.

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

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.