Study design
This post hoc analysis of Phase 3, randomized, double-blind, placebo-controlled studies in adult patients utilized data from two, 6-month episodic migraine studies pooled (EVOLVE-1 and EVOLVE-2) and one 3-month chronic migraine study (REGAIN) to analyze the impact of galcanezumab 120 mg once-monthly injection relative to placebo on total pain burden among patients with episodic migraine or chronic migraine [12-14]. Patients enrolled in the studies were 18 to 65 years of age and had a diagnosis of migraine (with or without aura) for at least one year prior to enrollment and onset prior to 50 years of age [12-15].
In all studies, patients entered a prospective baseline period of 30 to 40 days in which they completed an electronic patient-reported outcome diary to record the occurrence of headaches, headache duration, headache features, severity of headache, and use of headache medication. Eligible patients were randomized (1:1:2) to subcutaneous injections of galcanezumab 120 mg/month (following an initial 240 mg loading dose), galcanezumab 240 mg/month, or placebo. For the episodic migraine studies, patients were to have discontinued the use of medication or other treatments for the prevention of migraine for at least 30 days, and use of botulinum toxin A and B for at least 4 months, prior to the prospective baseline period. For the chronic migraine trial, patients were allowed to continue using topiramate or propranolol if they were on a stable dose in the 2 months prior to the prospective baseline period and remained on that dose throughout the baseline and double-blind periods. During all three studies, patients were permitted to continue acute migraine medications including triptans, ergots, nonsteroidal anti-inflammatory drugs, aspirin, and acetaminophen. patients were permitted to continue acute migraine medications including triptans, ergots, nonsteroidal anti-inflammatory drugs, aspirin, and acetaminophen. Opioid- and barbiturate-containing medications were limited to 3 days/month, and only one corticosteroid injection was allowed during any period. Exclusion criteria of note for all three studies included, but was not limited to, prior exposure to any CGRP antibody or any therapeutic antibody 12 months prior to screening, using opioids or barbiturates more than twice per month, persistent daily headache, cluster headache, head or neck trauma within the past 6 months, possible posttraumatic headache, primary headache, or a medical or psychiatric illness that would preclude study participation [12-14]. The study protocols were reviewed and approved by the appropriate institutional review board for each of the study sites. The studies were conducted according to Good Clinical Practice and the Declaration of Helsinki guidelines. Patients provided written informed consent before undergoing study procedures. The studies are registered with ClinicalTrials.gov (NCT02614183, NCT02614196, and NCT02614261).
Outcomes and statistical methods
The primary outcome of the EVOLVE-1, EVOLVE-2, and REGAIN studies was the overall mean change from baseline in the number of monthly migraine headache days during the double-blind treatment phase. This post hoc analysis evaluated the galcanezumab 120 mg and placebo treatment arms of the 3 studies. The objective of these analyses was to derive and compare between treatment groups, a “total pain burden” measure that incorporates: frequency of migraine headache days in a month, duration of migraine headache on a given day, and maximum severity of migraine headache on a given day. The total pain burden for a given month (severity-weighted duration) was calculated by multiplying duration (hours) of migraine headache and maximum pain severity (0=none, 1=mild, 2=moderate, 3=severe) for each migraine headache day and summing these over the days in a month. As an example, consider a patient who has 2 days of migraine headache in a month. The patient reports 2 hours of migraine headache on Day 1, which is of mild severity (score = 1) and 3 hours of migraine headache on Day 2, which is of moderate severity (score = 2). The total pain burden score for that month would be calculated as the sum of (2 hours x 1) and (3 hours x 2) which equals 8 severity-weighted hours of total pain burden. The change from baseline in monthly total pain burden measure over the double-blind period was analyzed for both episodic (6 months) and chronic migraine (3 months) studies using a mixed-model repeated measures (MMRM) model. The analysis of the episodic migraine studies included the following fixed effect variables: baseline, treatment, month, study indicator (EVOLVE‑1 or EVOLVE‑2), pooled region/country (nested within study), and the treatment by month, and baseline by month interaction effects. For the chronic migraine study, the fixed effect variables included: baseline, treatment, month, pooled country, baseline medication overuse (yes/no), concurrent prophylaxis use (yes/no), the interaction effects of treatment by month, and baseline by month. A marginal unstructured covariance structure was assumed to account for the correlation induced due to repeated measures on patients. The mean percentage change in total pain burden was analyzed using an MMRM model, as described earlier, for the change from baseline analyses. A similar MMRM modeling approach was also used to analyze change from baseline in each of the individual components of frequency of migraine headache days, duration per migraine headache day, and severity.
The Type II squared semi-partial correlation of each individual component within the total pain burden measure at baseline was obtained from the following regression model on the log transformed variables: log (total pain burden) = log (monthly migraine headache days), log (mean hours per migraine headache day), and log (severity of remaining migraine headache days). These provide the additional or unique proportion of variability explained by each component in its ability to predict total pain burden after adjusting for the other two components.
To explore the construct validity of the total pain burden measure, the Spearman’s rank correlation was used to determine the degree of correlation of the Migraine Specific Quality of Life Questionnaire (MSQ) and Migraine Disability Assessment Scale (MIDAS) to total pain burden at baseline. The MSQ is a self-administered instrument that evaluates the physical and emotional limitations of specific concern to patients with migraine [16]. The MIDAS quantifies days with headache-related disability across 5 areas over the last 3 months (90 days) [17].
Sensitivity analyses
The pain severity scores (0=none, 1=mild, 2=moderate, 3=severe) can be thought of as arbitrary. Therefore, a sensitivity analysis was performed by calculating (and comparing between groups) the total pain burden score using the square root of severity score in the calculation instead of severity (that is, using scores of 1=mild, 1.414=moderate, and 1.732=severe) and using the square of the severity score in the calculation instead of severity (that is, using scores of 1=mild, 4=moderate, and 9=severity). To further understand whether the changes observed in total pain burden can be fully explained by the change observed in frequency of migraine headache days, the original MMRM model analyzing change from baseline in total pain burden was repeated after including an additional time-varying covariate of the change from baseline in the number of monthly migraine headache days.
General considerations
Only patients in the intent-to-treat (ITT) population were considered for inclusion in this post-hoc analysis. Baseline demographics and descriptive summaries were provided for continuous variables using means and standard deviation and/or median and quartiles; for categorical variables, frequency and percentages were used. Treatment effects from MMRM models were expressed as least squares (LS) means and 95% confidence intervals (CI). No specific methods, such as multiple imputation, were employed to handle missing data. Furthermore, all repeated measures models included patients that had data at baseline and at least one time point during the double-blind phase. All statistical tests done were two-sided assuming a significance level of 5%. All statistical analyses were done using SAS® Enterprise Guide version 7.1. All analyses done were post hoc and results should be considered exploratory.