Efficacy and safety of mycophenolate mofetil therapy in neuromyelitis optica spectrum disorders: a systematic review and meta-analysis

Mycophenolate mofetil (MMF) is an immunosuppressive agent (IS) which is widely prescribed in neuromyelitis optica spectrum disorder (NMOSD) patients. We aim to assess the efficacy and safety of MMF in controlling relapse and disease severity. Eligible studies obtained from the EMBASE and Ovid MEDLINE databases were studies of NMOSD patients treated with MMF, which reported treatment outcomes as Annualized Relapse Rate (ARR) or Expanded Disability Status Scale (EDSS) before and after treatment. Fifteen studies included 1047 patients, of whom 915 (87.4%) were aquaporin-4 immunoglobulin seropositive. The total number of patients that received MMF was 799. A meta-analysis on ARR was conducted in 200 patients from 4 studies and on EDSS in 158 patients from 3 studies. The result showed a significant improvement with a mean reduction of 1.13 [95% confidence interval (CI) 0.60–1.65] in ARR, and a mean reduction of 0.85 (95% CI 0.36–1.34) in EDSS after MMF therapy. Adverse events occurred in 106 (17.8%) of 594 patients during MMF therapy. This systematic review and meta-analysis showed that using MMF as a preventive therapy in NMOSD patients can significantly reduce relapse rates and improve disease severity with acceptable tolerability.


Materials and methods
Study selection. Two investigators (S.S. and P.K.) independently searched for eligible published peerreviewed studies indexed in Ovid MEDLINE and EMBASE databases from inception to April 2020. The search terms "neuromyelitis optica spectrum disorder" and "mycophenolate mofetil" were used (supplementary data), and was limited to English-language human studies. Eligible studies could be either randomized-controlled trials or cohort studies/case series that investigated MMF's efficacy in NMOSD patients. Changes in either the annualized relapse rate (ARR) ratio or Expanded Disability Status Scale (EDSS) score before and after treatment must be reported. To avoid the non-representativeness of cases, case series that included fewer than 3 patients were excluded. The studies were reviewed in full-length to assess the appropriateness for their inclusion in the systematic review. Any differences in the determination of study eligibility between the two investigators mentioned above were re-evaluated, and the disagreement was resolved by discussion with other investigators (J.J. and P.U.). Data extraction. The extracted data included year and country of publication, study design, diagnostic criteria of NMO/NMOSD, demographic data of patients (patient population, female sex ratio, age of onset, follow-up duration, mean disease duration, ARR, EDSS score before and after MMF treatment, and AQP4-IgG serostatus), MMF treatment protocols (dose and duration of MMF treatment, and previous or concurrent therapy) and outcome measures.
Efficacy and safety measures. For the primary outcome on efficacy, differences in ARR, and the mean and median EDSS scores before and after MMF treatment were assessed. Safety outcomes included the proportion of deaths, drug withdrawals due to toxic effects, and the ADRs related to MMF use such as infection, malignancy, and abnormal laboratory results if available. Detail of adverse events in 15 studies on mycophenolate mofetil in NMOSD was displayed in Table 3.
Quality assessment and statistical analysis. Quality assessment for the included observational studies was performed using the Newcastle-Ottawa quality assessment scale, which consists of 3 domains: (a) selection of the participants; (b) comparability between the groups; and (c) ascertainment of the outcome in cohort studies 12 . Differences in the assessment were discussed and resolved with consensus among investigators.
Continuous and dichotomous data were both included in this study. Continuous data (ARR and EDSS) were reported as a median with range or mean with standard deviation (SD) depending on available data in the original articles. A number presented dichotomous data (i.e., number of ADRs, number of AQP4-IgG seropositive patients) with a percentage.
Meta-analysis was performed using Review Manager 5.3 software from the Cochrane Collaboration (London, United Kingdom). Mean pre-and post-treatment ARR, as well as mean pre-and post-treatment EDSS along with their SD, were extracted from each study, and the mean difference (MD) was calculated. If the study reported a 95% confidence interval (95% CI) instead of SD, the SD would be calculated from an upper limit of the 95% CI. Statistical heterogeneity was evaluated using Cochrane's Q test and was complemented with the I 2 statistic, which quantifies the proportion of the total variation across studies due to heterogeneity rather than chance. An I 2 value from 0 to 25% represents insignificant heterogeneity, 26-50% represents low heterogeneity, 51-75% Demographic and clinical characteristics. The characteristics and demographic data for the 15 studies are described in Table 1 16,19,27 with EDSS that reported the mean and standard deviation (SD), they were included in the meta-analysis. The remaining 8 studies did not report the mean and SD; therefore, they could not be statistically analyzed and were omitted from the meta-analysis. Nevertheless, we did a qualitative analysis of the 15 studies, which were displayed in Table 2.
Qualitative analysis. All 15 studies reported the median ARR before and after treatment. One study 22 reported ARR by using "the total number of relapses per patient-year" while the remaining studies defined ARR as the number of relapses per year. All but one study 22 demonstrated significant ARR reduction after receiving MMF (p < 0.05). The relapse-free rate was 60% (ranged from 27 to 80%).
For the 11 studies reporting EDSS as a treatment outcome, 7 studies revealed stabilization or improvement of disability in patients receiving MMF treatment measured by EDSS with the proportion varying from 87 to 100%. Eleven studies reported median EDSS before and after MMF treatment. Of those, 9 showed significant post-treatment median EDSS improvement (p < 0.05).
There were 2 studies 20,24 that categorized patients into seronegative and seropositive NMOSD groups and reported efficacy on MMF treatment separately in each group. Montcuquet et al. showed a reduction in the median pre-treatment ARR from 1 to post-treatment ARR of 0.21 in seropositive NMOSD and 0.9-0 in seronegative NMOSD and a relapse-free rate of 46.7% and 61.3%, respectively. However, no changes in EDSS was demonstrated 20 . The other study also revealed the reduction in the median pre-and post-treatment ARR of 1.79-0.29 in seropositive NMOSD and 1.45-0.30 in seronegative NMOSD, and a relapse-free rate of 64% and 77.8%, respectively 24 (Table 2).   16 showed a large decrease in disability measured by EDSS from baseline of moderately severe (EDSS 4.1) to full independence (EDSS 2.1). (Fig. 3b).
Safety. ADRs are summarized in Table 3     www.nature.com/scientificreports/ chronic dermopathy of the hand, rectal cancer, renal insufficiency, rash, hypotension, fatigue, easy bruising, anxiety, and sun sensitivity. Data on the discontinuation of MMF was available for 687 patients. Twenty-seven patients (3.9%) discontinued MMF due to ADRs such as rash, agranulocytosis, leukopenia, thrombocytopenia, arthromyalgia, GI side effects, and amenorrhea. All ADRs were reversible after discontinuation of MMF. One patient was discovered to have high serum CEA (carcinogenicembryonic antigen). Three patients died during the MMF treatment: one with EDSS 8.5 succumbed from immobilization complications, another developed disseminated varicella-zoster with acute respiratory distress syndrome, and the other had-according to death certificate documents-"cardiopulmonary failure; respiratory drive failure and Devic's disease" 11,20,25 .

Discussion
Our analysis, including 15 retrospective studies, showed that treatment with MMF for 13-95 months in NMOSD patients had significantly reduced ARR with a relapse-free rate of approximately 60% (ranged from 27 to 80%) and most of the studies showed EDSS stabilization or improvement varying from 87 to 100%. After excluding pediatric NMOSD patients from one study 26 , the median age of onset of our study varied from 28.7 to 56.0 years old. For those who underwent the meta-analysis, the post-treatment reduction of ARR decreased approximately 1.13 times a year, and EDSS reduction was 0.85 points, compared to those before treatment initiation in NMOSD patients. The degree of disability measured by EDSS depends mainly on ambulation and EDSS at 3 or 4 is defined as full ambulation. The small change in EDSS at a higher level has a greater disable impact than the same amount of EDSS change at a lower scale. At the pre-treatment state, our analysis composed of 12 studies (9 studies individually and 3 studies in meta-analysis) with a moderately severe disability of median EDSS between 3.0 and 4.0, one 11 with severe pre-treatment disability with a median EDSS of 6.0, and one study with mild to moderate disabilities with a baseline EDSS of 2.0 19 . Therefore, the reversibility of permanent damages may not be obvious since most of the patients in our analysis could ambulate. Our study also showed that 46-80% of the NMOSD patients were free from relapses. Since disability in NMOSD patients is related to attacks and the accumulation of incomplete recovery, reducing the number of attacks should result in fewer neurological deficits 28 . The findings in our study suggested that MMF exerted positive effects in preventing future relapses and considerably decreasing disability measured by EDSS.
AQP4-IgG autoantibody was present in 84.6% of the total NMOSD patients (range 36.4% to 100%). Two out of the 15 studies 20,24 evaluated efficacy on MMF treatment separately between seronegative and seropositive NMOSD groups. Although with only 35 seronegative NMOSD, it seemed to show no different in treatment response, with a relapse-free rate around 60% between the two groups; however, no changes in EDSS were demonstrated.
Novel medications such as eculizumab, satralizumab, and inebilizumab have been recently approved for maintenance treatment in NMOSD. Although they showed higher efficacy in relapse reduction, varying from 74-94% in AQP4-positive NMOSD, the benefit was not seen across all of the studies in the AQP4-Ab negative NMOSD group [4][5][6][7] . The relapse-free rate is high, around 76.5-97.9% in AQP4-positive NMOSD; however, it varies around 56-84.4% during the 48-96 weeks treatment period for AQP4-Ab negative NMOSD [4][5][6][7] . Therefore, the efficacy of the new medications in AQP4-negative groups needs further studies. To date, the data for treatment of seronegative NMOSD patients has no robust evidence. Our study suggests that MMF may be useful in this group of patients.
A large international cohort study revealed that race affected the clinical phenotype, the age at onset, and the severity of attacks 29 . MMF was used as a first-line treatment in 23% for Asians, 13% for Caucasians, and 27% for Afro-America/Afro-European NMOSD patients with a relapse-free rate of approximately 48%; 54% in Asian and 41% in Caucasian. However, the overall outcome is most dependent on early and effective immunosuppressive treatment 29 . Therefore, treatment response in specific groups of patients and further studies on pharmacogenomics are needed to understand the effect of racial difference and response to IS 29 .
A previous study suggested that MMF's efficacy with or without low dose steroids is not statistically different 16 . Adding supraphysiologic doses of steroids may increase MMF's efficacy and increase the risk of infections 30 . Most of the studies in our analysis reported that one-to two-thirds of patients had concomitant use of corticosteroids. Four studies reported concomitant use of MMF with corticosteroid in almost all of the patients 19,21,25,26 . However, details regarding the dose and duration of steroid treatment used in each study were not available. Future study on the benefit or risk of MMF with and without steroids is needed.
The efficacy of MMF was comparable to that of AZA but with fewer ADRs [17][18][19]21,31,32 . Besides, MMF therapy has been escalated when treatment with AZA showed a suboptimal response, or patients cannot tolerate AZA's side effects 25,27 . Huang et al. compared MMF's efficacy and safety with RTX, AZA, CYP, and cyclosporine A (CyA) and found that MMF was superior to AZA and CYP but inferior to RTX and CyA 33 . However, MMF had the highest tolerability among all IS in the study 33 . The present study demonstrated that common ADRs were infections, abnormal liver function tests, hair loss, GI symptoms, and bone marrow suppression. ADRs reported from other studies included infections, bone marrow suppression, and malignancy 34 . Increased risk of malignancy has not been proven in another study 35 . ADRs from MMF was not severe nor life-threatening. Only 27 patients (3.9%) discontinued MMF due to ADRs, 3 cases with fatalities, one from infections 20 , and the other two seem to be related to NMOSD 11,25 . These findings are also consistent with other studies reporting tolerable side effects of MMF compared to other IS, e.g., AZA, CYP, or RTX, which led to better drug compliance 33 .
So far, the novel drug reports no serious side effects but particular caution is needed with regards to respiratory tract and urinary tract infection; however, long term side effects require evaluation.
Even though the new medications showed very high efficacy, the medication and accessibility cost hampers the use of the new drug. Furthermore, it is still unclear whether the new drug should be used as a first-line treatment or escalated when patients do not respond to other IS. Currently, MMF is among the first-line maintenance therapy for NMOSD, especially in Asia, where it is widely available for use and at affordable prices.

Limitation
Our analysis has several limitations. Firstly, the studies included in this review were mostly observational cohort studies subjected to particular bias. Secondly, the study populations' heterogeneity, particularly racial differences, pre-treatment disability, and frequency of relapse, reflected the severity of the disease, various MMF dosage use, and concomitant corticosteroid use before using other IS, which could contribute to different treatment outcomes and ADRs. MMF's efficacy in NMOSD patients should be cautiously interpreted and need further studies; nevertheless, it seems to show reasonable effects for relapse prevention in NMOSD patients.

Conclusion
This systematic review and meta-analysis indicate that receiving MMF as a preventive therapy in NMOSD patients is associated with a reduction in ARR and EDSS compared to pre-treatment use. It also has acceptable ADRs and low rates of discontinuation.