The study design has been reported previously . Briefly, this was a phase 2, four-period, randomized, double-blind, PBO-controlled study of GMB in Japanese patients with EM (NCT02959177). Following protocol approval by local independent ethics review boards and written informed consent from the participants, the study was conducted in accordance with the Declaration of Helsinki, the International Council for Harmonisation Guideline for Good Clinical Practice, and applicable laws and regulations at 40 sites in Japan between December 2016 and January 2019.
The four periods of the study comprised a screening period, including full clinical assessment and washout of preventive treatments for migraine for at least 30 days; a baseline period to confirm patient eligibility and establish baseline number of migraine headache days; a 6-month, randomized, double-blind, PBO-controlled treatment phase; and a 4-month washout and follow-up phase. This analysis focuses on patient satisfaction during the 6-month treatment phase.
Eligible patients had to have onset of migraine with or without aura before age 50 years and occurring at least 1 year before entering the study, as well as a migraine frequency of 4–14 monthly migraine headache days with at least two attacks per month. Monthly migraine headache days were defined as a calendar day on which a migraine headache or probable migraine headache occurred. Patients currently taking preventive treatments for migraine were not eligible for enrollment in the study unless they stopped prior preventive treatments and went through a washout period of at least 30 days. Other ineligible patients were those who had a higher monthly frequency of headache days (at least 15 monthly migraine headache days during the 3 months before the screening period) or had chronic migraine. Additional exclusion criteria have been previously reported .
As previously described , patients were randomized (2:1:1) to one of three treatment groups, receiving PBO, GMB 120 mg, or GMB 240 mg. Treatments were administered once a month by subcutaneous injection. A loading dose of 240 mg was given at the first injection (month 0) to patients receiving GMB 120 mg.
Patients’ experience of GMB treatment was measured using three scales. The Patient Global Impression of Severity (PGI-S) and Patient Global Impression of Improvement (PGI-I) scales have been developed and used in many conditions/diseases to assess patient-reported severity and health improvements [7, 8, 24,25,26,27]. The PGI-S scale ranges from a score of 1 (“not at all ill”) to 7 (“extremely ill”), and the PGI-I scale ranges from a score of 1 (“very much better”) to 7 (“very much worse”). The PGI-S was measured at baseline (month 0) and at each monthly visit (months 1–6). The PGI-I was measured at each monthly visit (months 1–6).
The third scale was the Patient Satisfaction with Medication Questionnaire-Modified (PSMQ-M), which assesses patients’ levels of satisfaction with medication . The PSMQ-M is a self-rated scale, which has been used in one previous open-label study of GMB . It assesses three items related to clinical treatment over the previous 4 weeks: satisfaction, preference, and side effects. Satisfaction with the current medication was rated from “very unsatisfied” to “very satisfied”. Preference, in relation to previous medications used, was rated from “much rather prefer my previous medication” to “much rather prefer the medication administered to me during the study”. Side effects, in relation to previous preventive medications used, were rated from “significantly more side effects” to “significantly less side effects.” The PSMQ-M was administered at months 1 and 6.
Mean changes from baseline to each visit for PGI-S were assessed using a restricted maximum likelihood (REML)-based mixed-model repeated measures (MMRM) technique and are reported as least squares (LS) means ± standard error (SE). The PGI-S MMRM analysis included fixed categorical effects (treatment, month, the baseline number of monthly migraine headache days [< 8, ≥ 8], and treatment-by-month interaction) and continuous fixed covariates (baseline value and baseline-by-month interaction). The PGI-I raw values for each visit were also analyzed using a MMRM technique, including the same fixed categorical effects (treatment, month, the baseline number of monthly migraine headache days [< 8, ≥ 8], and treatment-by-month interaction), with PGI-S baseline value and PGI-S baseline-by-month interaction included as continuous fixed covariates. PGI-I results are reported as LS means ± SE. For each treatment group, frequency tables of PGI-I score and PGI-S change from baseline at each month were generated and summarized as frequency heatmaps.
An analysis of positive PGI-I responses explored the relationship between PGI-I and monthly migraine headache days (the primary efficacy measure ) using conditional expectation. Positive PGI-I responses were defined as “very much better” and “much better.” The analysis translated the difference in the primary efficacy measure (monthly migraine headache days for months 1–6) between treatment groups into the difference in the positive PGI-I response rate (month 6), resulting in a measure of clinical importance (improvement in symptoms) that allowed comparison of treatment benefit. The estimated positive PGI-I response rate in month 6 was calculated as the conditional expectation E[Y120 − Y0|(− 1)[X120 − X0] = Δ] for GMB 120 mg, and E[Y240 − Y0|( − 1)[X240 − X0] = Δ] for GMB 240 mg. Y0, Y120, and Y240 are the positive PGI-I response rates at month 6 for PBO, GMB 120 mg, and GMB 240 mg, respectively. X0, X120, and X240 are overall changes from baseline in monthly migraine headache days (average of months 1–6) for PBO, GMB 120 mg, and GMB 240 mg, respectively. Δ is the prespecified difference in monthly migraine headache days between GMB 120 mg (or GMB 240 mg) and PBO. For computational purposes, we set Δ as integer values. For GMB 120 mg, monthly migraine headache days were reduced on average by 3.0 per month vs. PBO ; therefore, Δ was set at 3. For GMB 240 mg, monthly migraine headache days were reduced on average by 2.8 per month vs. PBO ; therefore, Δ was set at 2 or 3. The 95% confidence intervals (CIs) for this analysis were generated using a nonparametric Monte Carlo bootstrap (1000 replications). This analysis was previously used for the tadalafil benign prostatic hyperplasia new drug application in Japan .
Three analyses of the PSMQ-M data were conducted: a descriptive summary of each item (satisfaction, preference, and side effects) for months 1 and 6; a descriptive summary of the positive PSMQ-M responses in months 1 and 6; and a logistic regression analysis of the month 6 positive responses. Positive PSMQ-M responses were defined as follows: for satisfaction, reports of “somewhat satisfied” or “very satisfied”; for preference, reports of “much prefer study medication” or “prefer study medication”; and for side effects, reports of “significantly less side effects” or “less side effects” than their previous medication. The logistic regression model included the categorical effects of treatment and baseline number of monthly migraine headache days (< 8, ≥ 8).
No adjustments for multiplicity were made among arms, time points, and analyses. Significance was based on a two-sided alpha level of 0.05. All statistical analyses were performed using SAS® software version 9.4 (SAS Institute Inc., Cary, NC, USA).