Eculizumab in Aquaporin-4–Positive Neuromyelitis Optica Spectrum Disorder

Trial Design and Oversight

From April 2014 through October 2017, we enrolled patients at 70 sites, primarily hospital clinics, in 18 countries. At baseline (day 1), we randomly assigned patients in a 2:1 ratio to receive either eculizumab or placebo. The patients were stratified across sites according to the score on the Expanded Disability Status Scale (EDSS) on day 1 (≤2.0 or 2.5 to 7.0 on a scale ranging from 0 [no disability] to 10 [death]) and the use of concomitant immunosuppressive therapy (no previous immunosuppressive therapy, except for glucocorticoids alone; continuing the same immunosuppressive therapy after the last relapse before screening, although doses may have changed; or receiving new immunosuppressive therapy or discontinuing an existing therapy after the last relapse). Patients, site staff members, and the representatives of the sponsor (Alexion Pharmaceuticals) were unaware of the trial-group assignments, and trial agents and treatment kits appeared identical. (Details regarding the design of the trial are provided in the protocol and in Fig. S1 in the Supplementary Appendix, both of which are available with the full text of this article at NEJM.org.)

The trial was designed to continue until 24 patients had a relapse of NMOSD, as adjudicated by an independent panel. After a review of blinded data, representatives of the sponsor terminated the trial after 23 patients had had an adjudicated relapse, given the uncertainty in estimating when the final event would occur. The patients who completed the trial could enter an extension trial and receive open-label treatment with eculizumab.

The trial was conducted in accordance with the provisions of the Declaration of Helsinki,¹⁷ the International Conference on Harmonisation guidelines for Good Clinical Practice,¹⁸ and applicable regulatory requirements. The trial was approved by the institutional review board at each participating institution. All the patients provided written informed consent before participation.

Alexion Pharmaceuticals designed the trial in consultation with two academic authors, provided the trial agents, and analyzed the data. The sponsor paid for professional writing assistance. Confidentiality agreements were in place between the authors and Alexion Pharmaceuticals. All the authors vouch for the completeness and accuracy of the data, the reporting of adverse events as stipulated in the protocol, and the fidelity of the trial to the protocol.

Patients

Eligibility criteria included an age of at least 18 years; a diagnosis of neuromyelitis optica or NMOSD, according to 2006¹⁹ or 2007³ criteria, respectively; AQP4-IgG–seropositive status, as confirmed with the use of a commercially available cell-binding kit assay (Euroimmun) analyzed centrally at Mayo Medical Laboratories; a history of at least two relapses during the previous 12 months or three relapses during the previous 24 months, at least one of which had occurred within the previous 12 months; and a score of 7 or less on the EDSS.²⁰ Patients who were receiving immunosuppressive therapies for relapse prevention were eligible for inclusion if they were receiving stable-dose regimens.

Exclusion criteria included treatment with mitoxantrone or rituximab during the previous 3 months, the receipt of intravenous immune globulin during the previous 3 weeks, the receipt of prednisone doses greater than 20 mg per day or the equivalent for other glucocorticoids at screening, and unresolved meningococcal disease or systemic bacterial or other infection considered to be clinically significant or not treated with appropriate antibiotics. (A complete list of inclusion and exclusion criteria is provided in the Supplementary Appendix.)

Trial Procedures

Patients were vaccinated against Neisseria meningitidis before receiving a trial agent (Fig. S1 in the Supplementary Appendix). Eculizumab or matching placebo was administered intravenously over approximately 35 minutes (range, 25 to 45 minutes). Patients received 900 mg of eculizumab weekly for the first four doses. Starting the following week, patients received a maintenance regimen of 1200 mg every 2 weeks until relapse, trial discontinuation, or the end of the trial. Immunosuppressive therapies that patients were receiving at baseline were changed only if the treating physician determined that a relapse had occurred or had safety concerns; changes to immunosuppressive therapy after relapses were unrestricted. (Details regarding concomitant medications are provided in the Supplementary Appendix.)

Patients were evaluated within 48 hours after a possible relapse and again following intervals of 1 week, 4 weeks, and 6 weeks by the treating physician and by EDSS raters. (The EDSS raters, who were also unaware of trial-group assignments, were not involved in patient care.)

Treating physicians identified relapses according to the following criteria: a new onset of neurologic symptoms or worsening of existing neurologic symptoms with an objective change on neurologic examination that persisted for more than 24 hours, signs and symptoms attributable to NMOSD rather than other causes, and onset preceded by at least 30 days of clinical stability. Changes in imaging were not considered to be an indication of relapse without related clinical findings. Treating physicians determined the appropriate relapse therapy.

On July 1, 2016, after the enrollment of 88 patients, the protocol was amended to create an independent relapse adjudication committee consisting of two neurologists and one neuro-ophthalmologist who were unaware of trial-group assignments and who reviewed all physician-determined relapses; adjudications were performed retrospectively and with the same criteria that were used by treating physicians. (Details regarding relapse assessment and adjudication are provided in the Supplementary Appendix and the protocol.)

Neurologic function, including visual acuity, and disability assessments were performed at baseline, at weeks 4, 8, and 12 and every 12 weeks thereafter, and at the time of trial discontinuation or termination. Raters assessed patients using the EDSS. Treating physicians or appropriately trained site staff members assessed visual acuity with the Snellen chart; they also evaluated the patients using the modified Rankin scale,²¹ with scores ranging from 0 (no disability) to 6 (death), and the Hauser Ambulation Index,²² with scores ranging from 0 to 9, with higher scores indicating decreased independent ambulation. Patients completed the European Quality of Life 5-Dimension 3-Level (EQ-5D-3L) questionnaire²³ consisting of a visual analogue scale with scores ranging from 0 to 100 and a summary index score ranging from less than 0 to 1; higher scores on each component represent better health status.

Safety assessments included monitoring for adverse events, evaluation of vital signs, a physical examination, the performance of electrocardiography and clinical laboratory tests, and evaluation for the presence of suicidal ideation or behavior. Additional details regarding safety monitoring, including definitions of adverse events, are provided in the Supplementary Appendix.

End Points

The primary efficacy end point in this time-to-event trial was the first adjudicated relapse. The original primary efficacy end point was the first physician-determined relapse, which was changed to a sensitivity analysis by protocol amendment after 88 patients had been enrolled. There were six hierarchically ordered secondary efficacy end points: the adjudicated annualized relapse rate and changes from baseline in scores on the EDSS, modified Rankin scale, Hauser Ambulation Index, EQ-5D-3L visual analogue scale, and EQ-5D-3L summary index. In the original statistical analysis plan, the EDSS score was evaluated first in the hierarchy before the annualized relapse rate. After a reassessment of the relative clinical importance of the trial end points but before the trial data were unblinded, the sponsor changed the hierarchy in the final version of the statistical analysis plan. Adverse-event data were calculated with and without physician-determined relapses of NMOSD that met the definition of a serious adverse event. Investigators assessed the relationship between adverse events and trial regimen and recorded the relationship using predefined categories (not related and unlikely, possibly, probably, or definitely related).

Statistical Analysis

The sample size for this trial was based on the percentage of patients who we estimated would be relapse-free (80% for eculizumab; 40% for placebo) at 12 months (hazard ratio, 0.24). With 2:1 randomization for the trial groups, we calculated that 24 events in a population of 132 patients would provide a power of 90% to determine the prespecified between-group difference on the basis of a two-sided log-rank test at a 5% level of significance, assuming a 10% dropout rate.

We used a stratified log-rank test to analyze between-group differences for the primary end point in the modified intention-to-treat population, which included all the patients who had undergone randomization and received at least one dose of a trial agent. (In our trial, the modified intention-to-treat population was the same as the intention-to-treat population because all the patients who had undergone randomization also received at least one dose of a trial agent.) We used a stratified Cox proportional-hazards model with the trial group as a covariate to estimate the hazard ratio for the primary end point. Data for patients who did not have an adjudicated relapse were censored at the end of the trial period, including those who had a physician-determined relapse that was adjudicated as not being a relapse and those who discontinued the trial regimen early. Sensitivity analyses for the primary end point are described in the statistical analysis plan and the Supplementary Appendix.

To analyze the adjudicated annualized relapse rate, we performed a Poisson regression analysis with the trial group, historical annualized relapse rate, and randomization strata as covariates. For the between-group differences in the remaining secondary efficacy end points, we used a stratified, randomization-based nonparametric analysis of covariance that was adjusted for the baseline score on each scale. We calculated differences in least-squares means and 95% confidence intervals using analyses of covariance after adjustment for the baseline score on each scale and randomization strata. Missing data were replaced by the last-observation-carried-forward method, as described in the statistical analysis plan and the statistical information section in the Supplementary Appendix.

To control for the overall type-I error at 0.05 for multiple hypothesis tests,²⁴ we used a fixed-sequence hierarchical testing procedure, with the primary end point followed by the six secondary efficacy end points in the order described earlier. Point estimates and 95% confidence intervals are provided for the first outcome with a P value greater than 0.05 and then for all subsequent outcomes in the hierarchy. No inferences can be drawn from results after the failure of statistical significance in the hierarchy.

Patients

Figure 1. Figure 1. Enrollment, Randomization, and Follow-up.

Definitions of the analysis populations are provided in the Methods section (modified intention-to-treat population) and the Supplementary Appendix (safety and per-protocol populations, including the reasons for exclusion from the per-protocol population). Randomization was stratified according to patients’ scores on the Expanded Disability Status Scale (EDSS) (≤2.0 or 2.5–7.0) and their status with respect to previous immunosuppressive therapy (IST) (no previous therapy, unchanged therapy, or changed therapy). IST was considered to have been unchanged if no therapy had been started or discontinued after the last relapse before screening, although doses may have changed; patients who had previously received only glucocorticoid therapy were considered to have received no previous IST. Patients were considered to have completed the trial if the treating physician determined that they had had a relapse or if the trial had ended because 23 positively adjudicated relapses had occurred in 23 patients. AQP4-IgG denotes aquaporin-4 IgG.

Table 1. Table 1. Demographic and Clinical Characteristics of the Patients at Baseline.

A total of 96 patients received eculizumab and 47 received placebo (Figure 1). Overall, 91% of the patients were women. At baseline, the mean (±SD) annualized relapse rate during the previous 24 months was 1.99±0.94, and the median scores on the EDSS, modified Rankin scale, and Hauser Ambulation Index indicated moderate-to-severe disability (Table 1, and Table S1 in the Supplementary Appendix). Of the 143 patients, 46 (32%) had received rituximab previously but not within the 3 months before screening. A total of 34 patients (24%) did not receive any concomitant immunosuppressive therapy during the trial. The baseline characteristics of the patients were well balanced between the two groups.

A higher percentage of patients receiving eculizumab (17%) than placebo (6%) discontinued their participation in the trial; the corresponding rates of discontinuation were 9.3 and 5.6 per 100 patient-years, respectively (Figure 1, and Table S2 in the Supplementary Appendix).

Efficacy

Table 2. Table 2. Efficacy End Points.

Figure 2. Figure 2. First Relapse of Neuromyelitis Optica Spectrum Disorder (NMOSD) during the Trial.

Shown are Kaplan–Meier plots of relapse-free survival among patients who were receiving eculizumab or placebo for NMOSD in the analysis of the primary end point (Panel A), in the prespecified subgroup of patients who were not receiving concomitant immunosuppressive therapy (Panel B), and in the post hoc subgroup of patients who were receiving concomitant immunosuppressive therapy (Panel C). Data are from the modified intention-to-treat population. The tick marks indicate censoring of data. Data for patients who did not have an adjudicated relapse were censored at the end of the trial period, including those who had a physician-determined relapse that was adjudicated as not being a relapse and those who discontinued the trial regimen early. Concomitant immunosuppressive therapies are listed in Table 1.

The primary end point of adjudicated relapse occurred in 3 of 96 patients (3%) in the eculizumab group and in 20 of 47 (43%) in the placebo group (hazard ratio, 0.06; 95% confidence interval [CI], 0.02 to 0.20; P<0.001) (Table 2 and Figure 2A). The median time until the first adjudicated relapse was not reached in the eculizumab group and was reached at 103 weeks in the placebo group. Most relapses were of myelitis (Table S3 in the Supplementary Appendix).

The results of the time-to-event sensitivity analyses were consistent with those in the primary analysis (Fig. S2 in the Supplementary Appendix). In the sensitivity analysis of the first physician-determined (nonadjudicated) relapse (the original primary end point), relapses occurred in 14 patients (15%) in the eculizumab group and 29 (62%) in the placebo group (hazard ratio, 0.18; 95% CI, 0.10 to 0.34; P<0.001) (Fig. S3 in the Supplementary Appendix). Of 45 physician-determined relapses overall, 24 (53%) were adjudicated as relapses. Disagreements in determinations between treating physicians and the adjudication committee were most frequently because the committee considered that there had been insufficient objective change in the neurologic examination (in 18 of 21 relapses adjudicated as not being a relapse) (Table S4 in the Supplementary Appendix).

In the prespecified subgroup analysis involving patients who were not receiving concomitant immunosuppressive therapy, adjudicated relapses occurred in none of the 21 patients in the eculizumab group and in 7 of the 13 patients (54%) in the placebo group. In a post hoc analysis, the corresponding numbers for patients receiving concomitant immunosuppressive therapy were 3 of 75 patients (4%) and 13 of 34 patients (38%), respectively. The time course of relapses in each subgroup suggested a treatment effect consistent with that in the overall population (Figure 2B and 2C).

Eculizumab was associated with a lower adjudicated annualized relapse rate after adjustment (the first secondary end point) than placebo, with rates of 0.02 and 0.35, respectively (P<0.001) (Table 2). No inferences can be made for the remaining secondary end points because the difference between groups for the next end point in the hierarchy, the change in the EDSS score, was not significant (Table 2). (Additional data for disability and quality-of-life outcomes are provided in Figs. S4 and S5 in the Supplementary Appendix.)

Safety

Table 3. Table 3. Adverse Events.

The rate of adverse events was 745 per 100 patient-years in the eculizumab group and 1127 per 100 patient-years in the placebo group (Table 3). Higher rates of upper respiratory tract infection and headache were reported in the eculizumab group than in the placebo group, with 31 and 19 events of upper respiratory tract infection per 100 patient-years and 55 and 38 headache events per 100 patient-years, respectively. The rate of adverse events that were categorized as being related to a trial regimen by investigators was 212 and 164 per 100 patient-years, respectively.

The rate of serious adverse events was 27 and 55 per 100 patient-years in the eculizumab group and the placebo group, respectively, with the exclusion of relapses (Table 3). One patient in the eculizumab group who was receiving concomitant azathioprine died from pulmonary empyema after 108 weeks in the trial. The associated cultures yielded Streptococcus intermedius and Peptostreptococcus micros. Two patients discontinued placebo because of adverse events (pneumonia in one patient and prerenal failure and pancytopenia in the other). Serious adverse events that were categorized by investigators as being related to a trial regimen are described in Table S6 in the Supplementary Appendix.

No cases of meningococcal infection were reported during the trial. The rates of other serious infections were 8 and 15 events per 100 patient-years in the eculizumab group and placebo group, respectively.

Among patients with AQP4-IgG–positive NMOSD, those who received eculizumab had a significantly lower risk of relapse than those who received placebo. Approximately three quarters of the patients were receiving concomitant immunosuppressive therapy during the trial.

Immunosuppressive therapies are currently used to prevent NMOSD relapses.^4,5 In an open-label, randomized trial involving 86 patients who had NMOSD with or without AQP4-IgG antibodies, those who received the monoclonal antibody rituximab had a significantly greater reduction from baseline in the annualized relapse rate at 12 months than those who received azathioprine,²⁵ although neither drug is currently approved for this use. In our trial, rituximab was not permitted as concomitant immunosuppressive therapy because of incompatibility between its mechanism of action and that of the terminal complement inhibitor eculizumab. Specifically, rituximab selectively depletes B cells, mainly through complement-dependent cytotoxicity, and B-cell lysis is inhibited by 90% in the presence of eculizumab.²⁶ Approximately one third of the patients in our trial had previously received rituximab but not within 3 months before enrolling in the trial.

By blocking the terminal complement system, eculizumab increases the risk of meningococcal and encapsulated bacterial infection.^27,28 In our trial, all the patients received meningococcal vaccination, and no cases of meningococcal infection were reported. The microorganisms that were implicated in our trial patient who died from pulmonary empyema (i.e., S. intermedius and P. micros) are not associated with complement deficiency. A greater proportion of patients in the eculizumab group than the placebo group withdrew from the trial, even allowing for differences in on-trial time, but only two patients (both of whom were receiving placebo) discontinued owing to adverse events.

This trial has several limitations. First, only patients with AQP4-IgG antibodies were included, so the findings cannot be extrapolated to patients without AQP4-IgG antibodies, a potentially heterogeneous group that may also include those with antibodies against myelin oligodendrocyte glycoprotein.²⁹ Second, because the trial design precluded follow-up beyond 6 weeks after a single relapse and because the confidence interval for the between-group difference in the change from baseline in the EDSS score included zero in hierarchical testing (implying there was no benefit of the drug on disability progression during the short period of the trial), no formal inferences could be made about other disability and quality-of-life outcomes overall. Third, the trial was terminated after 23 of 24 prespecified adjudicated relapses but retained at least 80% power to detect a between-group difference for the primary end point based on original assumptions for the sample-size calculations. Fourth, on the basis of the same criteria, treating physicians identified more relapses than the adjudication committee. However, the discordance was mostly due to the interpretation of the objective change in the neurologic examination, which we assessed as reflecting concern among treating physicians that relapses should not be overlooked. Despite the discordance, a significantly lower risk of relapse with eculizumab than with placebo was apparent in the sensitivity analysis of physician-determined relapses. Fifth, although the trial was undertaken with a placebo comparator, the use of concomitant immunosuppressive therapy, except rituximab, was permitted.

In conclusion, eculizumab was associated with a lower risk of relapse than placebo among patients with AQP4-IgG–positive NMOSD. Since there was no significant between-group difference in measures of disability progression, the long-term effect of eculizumab in patients with NMOSD warrants further study.