Post-hoc analysis of open-label extension period of study MCI186-19 in amyotrophic lateral sclerosis

Abstract

Study MCI186-19 investigated the safety and efficacy of edaravone in the treatment of ALS. The 24-week, double-blind period was followed by a 24-week, open-label, active extension period. Patients originally receiving edaravone continued edaravone (E-E group, n = 65), and patients originally receiving placebo switched to edaravone (P-E group, n = 58). Because no statistical tests had been prospectively planned in the open-label period, we performed post-hoc analyses to assist in the interpretation of efficacy data. A mixed model for repeated measures (MMRM) and the Combined Assessment of Function and Survival (CAFS) were assessed. Additionally, slopes of time-dependent change between baseline in cycle 1 and the end of cycle 6 (24 weeks double-blind) and between the end of cycle 6 and end of cycle 12 (24 weeks open-label) were calculated using a random coefficient model including all available data during each period.

At week 48, the MMRM analysis showed significantly less decline in ALS Functional Rating Scale-Revised (ALSFRS-R) total score in the E-E group than in the P-E group (least-squares mean change from baseline ± standard error, 4.17 ± 1.40, p = 0.0037), meaning that the differences in the ALSFRS-R total score during the 24-week double-blind period were maintained in patients receiving edaravone for an additional 24 weeks. The CAFS endpoint (p = 0.0089) supported this finding. The slope analysis during the double-blind period showed a significant difference between the treatment groups, while there was no significant difference between the groups during the active extension period. These analyses suggest a potential benefit of early and continued edaravone treatment over delayed edaravone treatment.

Keywords:

• Amyotrophic lateral sclerosis
• edaravone
• open-label
• study MCI186-19
• post-hoc

Introduction

In the first phase III, double-blind 24-week trial of edaravone in the treatment of ALS (MCI186-16), edaravone did not statistically demonstrate efficacy evaluated by the ALS Functional Rating Scale-Revised (ALSFRS-R) as primary endpoint (1). Exploratory analyses were performed to identify a subpopulation in which efficacy of edaravone might be detectable within a 24-week study period. As such, four key criteria were identified: (i) a baseline score of 2 points or better on each individual item of the ALSFRS-R to ensure a scale sensitivity to detect functional deterioration in each ALSFRS-R item; (ii) predicted percent forced vital capacity (%FVC) ≥ 80% to reduce variability in disease progression since respiratory insufficiency may lead to very rapid progression that may potentially mask a drug effect; (iii) definite or probable ALS based on El Escorial revised criteria to increase probability of disease progression during the 24-week study period; and (iv) within two years of ALS onset to exclude patients who are more likely to show less marked progression of disease during the study period (2).

In the second phase III, 24-week, placebo-controlled, double-blind study (MCI186-19) that prospectively defined these entry criteria, edaravone demonstrated a significant difference from placebo in change of ALSFRS-R total score ([the least-squares mean ± standard error] 2.49 ± 0.76, p = 0.0013) (3). The study MCI186-19 had a 24-week, open-label active extension period and the efficacy and safety findings were described in the previous article without statistical analysis, as none was included in the study design (4). To assist in interpreting efficacy data from the extension period, we performed post-hoc analyses for the efficacy endpoints to determine if the difference we observed during the initial 24-week double-blind period could be maintained for an additional 24 weeks.

Methodology

Study design and participants

Details of study methodology, patient selection, ethical study conduct, endpoints, and prospective statistical analyses for the 24-week double-blind period and 24-week open-label active extension period are described in earlier publications (4). The study is registered with ClinicalTrials.gov (NCT01492686).

Briefly, patients were required to meet all the four key criteria as described in the Introduction. After a 12-week pre-observation period, during which patients were required to have an ALSFRS-R change between –1 and –4 to maintain eligibility for the study, patients began the 24-week double-blind treatment period. Eligible patients were allocated in a 1:1 ratio to receive either 60 mg edaravone or placebo, once daily, via intravenous infusion over 60 min. A minimisation method of dynamic randomisation was used to ensure overall balance across the prognostic factors (change of ALSFRS-R during the pre-observation period [–3/–4 or –2/–1], El Escorial revised Airlie House diagnostic criteria [definite or probable], and age [< 65 or ≥65 years]) (4). Study treatment consisted of 12 cycles of 60 mg edaravone or matching placebo. Each cycle consisted of a 14-d treatment period followed by a 14-d drug-free period. Infusions were given on 10 of the 14 d of each treatment period cycle except in cycle 1 where infusions were given all 14 d. The 24-week, open-label, active-treatment period began at the end of cycle 6, at which point edaravone was administered to patients from either treatment group who agreed to continue in the study for an additional six cycles up to cycle 12.

A patient was discontinued from the study if any of the following discontinuation criteria were met: (i) patient request; (ii) ineligible for the study; (iii) adverse event; (iv) tracheotomy was required; (v) respiratory support was required all day long; (vi) worsening of ALS; (vii) spinal surgery for cervical spondylosis, intervertebral disc hernia, etc.; (viii) %FVC ≤50%, and PaCO₂ (blood gas) ≥ 45 mmHg; (ix) creatinine clearance of ≤50 mL/min; or (x) any other case (at the discretion of the investigator). The primary reason for discontinuation was recorded by the investigator.

The primary efficacy endpoint was ALSFRS-R total score. A secondary endpoint was time to death or a specified state of disease progression, defined in the protocol as death, disability of independent ambulation, loss of upper limb function (apparent as ALSFRS-R scores of 0 in the ‘Handwriting’, ‘Eating motion’, and ‘Dressing and hygiene’ items), tracheotomy, use of respirator (except bi-level positive airway pressure), use of tube feeding, or loss of useful speech. As directed by the prespecified statistical analysis plan, these events were recorded from the start of treatment until two weeks after the last dose. Predicted %FVC, grip and pinch strength were measured at the end of each cycle. The Modified Norris Scale and the ALS Assessment Questionnaire-40 items (ALSAQ-40) were obtained at the end of cycle 6 and cycle 12.

Post-hoc statistical methods

Efficacy was analysed in the full analysis set (FAS), defined as all randomised patients with ALS who received study drug and had at least one efficacy data point after baseline. Especially for a statistical evaluation for long-term efficacy in a progressing disease with a substantial clinical trial discontinuation rate, the last observation carried forward (LOCF) method might not be adequate to impute missing data after a proportion of the patients discontinued. Therefore, to better characterise the potential effects of treatment, a mixed model for repeated measures (MMRM), including all available patient data without data imputation, was applied for the efficacy assessments of ALSFRS-R, %FVC, Modified Norris Scale, ALSAQ-40, and grip and pinch strength, from baseline in cycle 1 to the end of cycle 12 (48 weeks). Since the MMRM requires at least one post-baseline value, patients who did not have any post-baseline value were excluded from the analyses. Additionally, analysis of covariance (ANCOVA) was applied for the Combined Assessment of Function and Survival (CAFS) endpoint (5), which incorporates function and survival together. CAFS Rank Scores were derived based on change in ALSFRS-R total score from baseline to the end of cycle 12 and on deaths occurring during the study (including the period of safety follow-up until four weeks after the last dose).

Additionally, slopes of time-dependent change between baseline in cycle 1 and the end of cycle 6 (baseline to week 24) and between the end of cycle 6 and end of cycle 12 (week 24 to 48) were calculated using a random coefficient model, including all available ALSFRS-R score data during each double-blind treatment period and open-label period, respectively. This model potentially assumes linearity in ALSFRS-R change during each double-blind treatment period and open-label period, respectively (6).

For survival analysis of death or a specified state of disease progression, a Kaplan-Meier curve was generated with the censored date, which was used as the last observation date. For patients with multiple events, the onset date of the first event was used for survival analysis. The same statistical tests that had been prespecified for the initial 24-week double-blind period (3), a log-rank test and generalised Wilcoxon test, were applied here for the 48-week interval from cycle 1 to the end of cycle 12.

Results

Overall, 137 patients were initially randomised at baseline of the 24-week double-blind period to receive edaravone (n = 69) or placebo (n = 68). The majority (n = 123) continued into the open-label active extension period. Of these, 65 patients previously received edaravone (E-E) and 58 patients previously received placebo (P-E). The primary reasons for discontinuation, judged by the investigators, are shown in Figure 1. Twelve patients in the E-E group and 18 patients in the P-E group discontinued from the study before the completion of the active treatment period, and 53 patients in the E-E group and 40 patients in the P-E group completed the active treatment period (Figure 1).

Figure 1. Patient disposition. MMRM, mixed model for repeated measures.

¹One patient in the placebo group who did not have any post-baseline efficacy value was excluded from MMRM analyses. ²A total of 2 patients required tracheotomy. Of these, 1 tracheotomy in the edaravone group that was required within 2 weeks after the last dose was counted as an event for secondary efficacy analysis in Table 3. One tracheotomy in the placebo group that was required more than 2 weeks after the last dose was not counted as an event for secondary efficacy analysis in Table 3. ³A total of 6 patients (2 patients in the E-E group and 4 patients in P-E group) died during the study. Of these, 3 deaths occurring within 2 weeks after the last dose (1 E-E patient discontinuing due to adverse events (AEs), 1 P-E patient discontinuing due to AEs, and 1 P-E patient discontinuing due to respiratory support all day) were counted as events for the secondary efficacy analysis in Table 3. Three deaths occurring more than 2 weeks after the last dose (1 E-E patient discontinued due to worsening of ALS and 2 P-E patients discontinued due to respiratory support all day) were not counted as events for secondary efficacy analysis in Table 3. All 6 deaths were incorporated into the CAFS analysis.

As the MMRM required patients to have a post-baseline assessment for the ALSFRS-R, the post-hoc analysis population included 69 (E-E) and 67 (P-E) patients, respectively.

In general, the demographics and baseline characteristics were balanced, and no relevant differences between the treatment groups were evident, with the exception of a lower ALSFRS-R score at the start of the open-label active extension period (E-E, 37.8 points; P-E, 34.8 points) (4).

Post-hoc statistical assessments

Patients treated with edaravone for 48 weeks had significantly less decline in ALSFRS-R total score than patients who were treated with placebo for 24 weeks before receiving edaravone (LS mean ± SE 4.17 ± 1.40, p = 0.0037). Incorporating six deaths (two in E-E and four in P-E), the CAFS analysis showed positive effect for the E-E compared to P-E (p = 0.0089) (Table 1).

Table 1. MMRM analysis of ALSFRS-R and CAFS analysis from baseline in cycle 1 to the end of cycle 12, FAS.

		Number of patients		Adjusted mean	Between-group difference
	Treatment group	Baseline	End of cycle 12	LS mean ± SE	LS mean ± SE (95% CI)	p Value
MMRM	E-E	69	51	−10.16 ± 0.97	4.17 ± 1.40(1.39, 6.95)	0.0037
	P-E	67	37	−14.33 ± 1.02
CAFS^*	E-E	69	–	–	32.51 ± 12.24(8.29, 56.73)	0.0089
	P-E	68	–	–

CAFS: the Combined Assessment of Function and Survival; CI: confidence interval; E-E: edaravone group in double-blind period followed by edaravone group in active-treatment period; FAS: full analysis set; LS: least-squares; MMRM: mixed model for repeated measures; P-E: placebo group in double-blind period followed by edaravone group in active-treatment period; SE: standard error.

^* An analysis of covariance was performed with baseline ALSFRS-R score, duration of ALS, riluzole use, initial symptom (bulbar or limb), and treatment group.

Observed cases of ALSFRS-R scores are displayed in Figure 2. Since this observed case analysis may be affected by any imbalance between groups in patients discontinuing from the study (in the current study more discontinued from the P-E group) and by any outliers (in the absence of joint modelling for repeated measures data), LS mean changes from baseline modelled by MMRM are displayed for comparison in Figure 3.

Figure 2. Mean ALSFRS-R score (observed cases, FAS).

ALSFRS-R: ALS Functional Rating Scale-Revised; BL: baseline; FAS: full analysis set; SD: standard deviation of the mean.

Figure 3. LS mean change (Δ) from baseline in ALSFRS-R score (MMRM, FAS).

One patient in the placebo group who did not have any post-baseline efficacy value was excluded from MMRM analyses. ALSFRS-R: ALS Functional Rating Scale-Revised; FAS: full analysis set; LS: least-squares; MMRM: mixed model for repeated measures; SE: standard error.

The slopes of the time-dependent change in ALSFRS-R (per cycle) calculated with a random coefficient model during the double-blind period were significantly different between treatment groups (E: –0.74 ± 0.08 vs. P: –1.21 ± 0.11, p = 0.0005). In contrast, the slopes did not differ between the groups during the open-label active treatment phase (E-E: –0.95 ± 0.09 vs. P-E: –1.08 ± 0.11, p = 0.3687).

Additionally, the difference in treatment effect between edaravone and placebo in the initial 24-week double-blind period appeared to be sustained through 48 weeks of treatment when assessed by %FVC, Modified Norris Scale, and ALSAQ-40 (Table 2, Figure 4).

Figure 4. Efficacy comparisons of edaravone and placebo from baseline in cycle 1 to the end of cycle 12 of treatment (FAS, MMRM).

LS mean differences in scores are shown together with 95% confidence intervals. Because low ALSAQ-40 scores correspond to better health status, LS mean ALSAQ-40 differences that showed a favorable effect of edaravone had negative values, and here are transformed to positive values for comparison alongside the other secondary endpoints. CI: confidence interval. ALSAQ-40: ALS Assessment Questionnaire-40 items; ALSFRS-R: ALS Functional Rating Scale-R; FAS: full analysis set; %FVC: % forced vital capacity; LS: least-squares; MMRM: mixed model for repeated measures.

Table 2. MMRM analysis of %FVC, ALSAQ-40, Modified Norris Scale Score, grip and pinch strength assessments from baseline in cycle 1 to the end of cycle 12, FAS.

		Number of Patients		Adjusted Mean	Between-Group Difference
Endpoint	Treatment Group	Baseline	End of cycle 12	LS mean ± SE	LS mean ± SE (95% CI)	p Value
%FVC	E-E	68	50	−28.24 ± 3.52	11.88 ± 5.05(1.85, 21.91)	0.0207
	P-E	67	36	−40.12 ± 3.72
ALSAQ-40 score	E-E	68	51	32.69 ± 3.54	−10.71 ± 4.51(−19.66, −1.76)	0.0195
	P-E	61	37	43.40 ± 3.92
Modified Norris Scale Score (total)	E-E	68	51	−26.70 ± 2.71	10.28 ± 3.71(2.93, 17.64)	0.0066
	P-E	61	37	−36.98 ± 2.97
Grip strength	E-E	69	51	−7.45 ± 0.73	1.55 ± 1.03(−0.49, 3.59)	0.1350
	P-E	67	37	−9.00 ± 0.77
Pinch strength	E-E	69	51	−1.62 ± 0.18	0.09 ± 0.25(−0.40, 0.59)	0.7109
	P-E	67	37	−1.71 ± 0.19

ALSAQ-40: ALS Assessment Questionnaire-40 items; CI: confidence interval; E-E: edaravone group in double-blind period followed by edaravone group in active-treatment period; FAS: full analysis set; %FVC: % forced vital capacity; LS: least-squares; MMRM: mixed model for repeated measures; P-E: placebo group in double-blind period followed by edaravone group in active-treatment period; SE: standard error.

There was also a significant difference in death or a specified state of disease progression between the E-E and P-E treatment groups at 48 weeks; 10 patients in the E-E group and 19 patients in the P-E group had these events, respectively (log-rank test, p = 0.0193; generalised Wilcoxon text, p = 0.0347) (Table 3).

Table 3. Death or specified disease progression at the end of cycle 12.

Event	E-E	P-E
Death	1	2
Disability of independent ambulation	3	2
Loss of upper limb function	4	6
Tracheotomy	1	0
Use of respirator	0	0
Use of tube feeding	2	5
Loss of useful speech	4	5
Total events	15	20
Number of patients with events	10	19
Between-group comparison (post-hoc)	p = 0.0193, log-rank testp = 0.0347, generalised Wilcoxon test

Note: If multiple events occurred in one patient, the date of the first event was counted as the date of event. The date of censoring was the last observation date for each patient.

According to the prespecified protocol, events were captured until two weeks after the last dose.

E-E = 6 cycles of edaravone in double-blind period followed by 6 cycles of edaravone in open-label period.

P-E = 6 cycles of placebo in double-blind period followed by 6 cycles of edaravone in open-label period.

Discussion

The entry criteria of the current study were designed so that treatment would be initiated when functions are still retained and also to reduce variability in the rate of disease progression. Moreover, the patients recruited into the study would be predicted to show disease progression on placebo treatment during the 24-week double-blind study period. The initial 24-week double-blind period of Study MCI186-19 demonstrated a significant difference between edaravone and placebo in the change of ALSFRS-R total score ([LS mean ± SE] 2.49 ± 0.76, p = 0.0013).

After the additional 24 weeks of open-label active treatment, patients who received edaravone for a total of 48 weeks had less functional loss as measured by the ALSFRS-R than patients who began 24 weeks of edaravone treatment after 24 weeks on placebo. The observed differences in the ALSFRS-R total score were maintained in patients who continued edaravone treatment for an additional 24 weeks. The CAFS analysis over the 48 weeks, which included the effect of six deaths in the treatment comparison, also showed a significant difference between E-E and P-E groups.

The difference in slopes between the P-E and E-E groups from cycles 7 to 12 was less than the difference in slopes between placebo and edaravone groups from cycles 1 to 6. The slope for the P-E group from cycles 7 to 12 decreased less than the slope for the placebo group from cycles 1 to 6. These findings may suggest that there is some effect of edaravone from cycles 7 to 12. The previous studies MCI186-16 and MCI186-17 also showed this trend in a post-hoc analysis (7).

In line with the results from the ALSFRS-R as an efficacy endpoint, the effect of treatment in this study appeared to be sustained when patients were assessed using the efficacy variables of %FVC, the Modified Norris Scale and the ALSAQ-40. It was unclear why grip strength and pinch strength did not show significant differences, in contrast to the other measurements. Grip strength or pinch strength as a single score measuring physical force might be less sensitive than the ALSFRS-R, Modified Norris Scale or ALSFRS-40 that are composite endpoints combining scores from multiple items.

Approximately half as many patients receiving edaravone for all 48 weeks experienced specified disease progression events as did patients who received 24 weeks of placebo followed by 24 weeks of edaravone. Although the latter 24 weeks was open label, it should be noted that with the exception of one death, all the disease progression events were identified prior to unblinding of patient treatment assignment for the double-blind period.

These findings that efficacy persists in the patients allocated to edaravone treatment would appear to support the early initiation of edaravone treatment in patients with ALS. This concept appears to be in line with the European Federation of Neurological Societies (EFNS) guidelines on the Clinical Management of Amyotrophic Lateral Sclerosis, which outline the importance of early diagnosis and early initiation of treatment with the neuroprotective agent because early diagnosis/treatment may provide opportunities for treatment with neuroprotective agents at a time when fewer cells are irreversibly compromised (8).Therefore, while high oxidative stress has been reported in patients at all stages of ALS (9), a potential neuroprotective effect of edaravone might be more apparent when fewer cells are irreversibly compromised and functions of patients are still retained.

The analyses presented in this article have limitations in interpretability. First, this extension period was an open-label study where both groups received edaravone. To minimise duration of placebo exposure in patients with ALS, a devastating and serious disease, the double-blind placebo period with placebo treatment was limited to 24 weeks, then followed by open-label edaravone treatment. Therefore, interpretations of MMRM analyses applied to the P-E group that received different treatments over time are necessarily limited. Secondly, the study was not primarily designed to assess survival, and few deaths occurred during the 48-week study. At initial study entry, the population of enrolled patients was documented to have high levels of general baseline functionality, and was also within two years of initial symptom onset. The findings of the CAFS analysis favouring edaravone over placebo appear to be driven primarily by retained functional ability as contributed by the ALSFRS-R total score. Survival of patients receiving edaravone will continue to be assessed by means of a post-approval registry of edaravone-treated ALS patients in Japan.

The third limitation is the exploratory nature of post-hoc analyses without adjustment for statistical multiplicity. Applying multiple post-hoc statistical methods for the same endpoint increases the type 1 error rate. Despite this, our results are consistent with those from study MCI186-16/-17 (7) and the study MCI186-19 double-blind/open-label extension (4) for the changes in the ALSFRS-R. Furthermore, in this post-hoc analysis of the open-label extension of study MCI186-19, the changes in the ALSFRS-R were consistent with the changes in the %FVC, Modified Norris Scale, and the ALSAQ-40.

In conclusion, patients who received edaravone treatment for 48 weeks had less functional loss than patients who received 24 weeks of edaravone treatment after 24 weeks of placebo. The observed differences in the ALSFRS-R total score, %FVC, Modified Norris Scale, and ALSAQ-40 were maintained in patients receiving edaravone for an additional 24 weeks, suggesting a potential clinical benefit from early initiation of edaravone treatment.

Declaration of interest

KT and KT are employees of Mitsubishi Tanabe Pharma Development America (MTDA). FT is an employee of Mitsubishi Tanabe Pharma Corporation (MTPC). JP is an employee of MTDA and MTPC.

Funding information

The edaravone MCI-186 clinical trials were funded by Mitsubishi Tanabe Pharma Corporation. This and the other articles in the ALSFTD Supplement, Edaravone (MCI-186) in ALS (Amyotrophic Lateral Sclerosis), were funded by Mitsubishi Tanabe Pharma America, Inc.

Acknowledgements

The authors take full responsibility for the content of and the decision to submit this manuscript but thank David E. Hartree, under contract with Mitsubishi Tanabe Pharma America, and Teresa A. Oblak of Covance Market Access Services Inc., for providing research support, coordination assistance, and editorial contributions.

Additional information

Funding

Mitsubishi Tanabe Pharmaceutical Corporation

MTPC

Mitsubishi Tanabe Pharmaceutical Corporation

MTPCMitsubishi Tanabe Pharma Corporation