Original Article

Cephalalgia 33(2) 101111 ! International Headache Society 2012 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0333102412466968 cep.sagepub.com

Randomized, double-blind, placebo-controlled, phase II trial of gabapentin enacarbil for migraine prophylaxis
Stephen Silberstein1, Stacey Goode-Sellers2, Colleen Twomey2, Jane Saiers3 and John Ascher2

Abstract Objective: The objective of this article is to evaluate the efficacy and safety of gabapentin enacarbil (GEn) for migraine prophylaxis. Methods: In this randomized, double-blind, parallel-group study, patients with International Headache Society-defined migraine who met criteria suggesting the need for prophylactic therapy were randomized 2:1:2:2:1 to one of the following five groups, designated according to target daily dose of study medication during the 20-week treatment period: placebo, GEn 1200 mg, GEn 1800 mg, GEn 2400 mg, or GEn 3000 mg. Results: The intent-to-treat population included 523 patients (n 128 placebo, n 66 GEn 1200 mg, n 134 GEn 1800 mg, n 133 GEn 2400 mg, n 62 GEn 3000 mg). No statistically significant difference between active treatment (the average of 1800 mg and 2400 mg treatment groups) and placebo was found for change from baseline in the number of migraine headache days during the last four weeks of treatment prior to taper (the primary endpoint). Results of analyses of the primary endpoint using the per protocol population, analyses using imputation methods different from those of the primary analysis, and nonparametric analyses were consistent with the primary analysis in showing no difference between active treatment and placebo. The pattern of results was similar for the secondary efficacy endpoints. Pharmacokinetic data demonstrate that patients had adequate estimated exposure to GEn. The adverse event profile of GEn was consistent with that in previous studies. Conclusion: GEn did not significantly differ from placebo for migraine headache prophylaxis. A high placebo effect should be considered when interpreting these results. Keywords Clinical trials, randomized controlled, headache, migraine
Date received: 16 November 2011; revised: 26 March 2012; 5 September 2012; accepted: 5 October 2012

Introduction
Prophylactic therapy for migraine is given to reduce the frequency, intensity, and duration of migraine attacks (14). Treatment guidelines advocate that prophylactic therapy be considered when patients quality of life and daily activities are substantially aected by migraine; when patients experience 2 or more migraine attacks monthly; for migraine attacks that do not respond to acute migraine medications; and for migraine attacks associated with frequent, long, or uncomfortable auras (3,4). Anticonvulsants including valproate and topiramate are among several classes of agents used for migraine prophylaxis (5). Their use in migraine prophylaxis is supported theoretically by their molecular and cellular eects on pain systems and systems putatively involved in migraine pathophysiology, but their mode of action in migraine prophylaxis is currently unknown (2,6,7).

1 2

Jefferson Headache Center, Philadelphia, PA, USA GlaxoSmithKline, Research Triangle Park, NC, USA 3 The WriteMedicine Inc., Chapel Hill, NC, USA Corresponding author: Stephen Silberstein, Jefferson Headache Center, 8130 Gibbon Building, 111 South 11th St., Philadelphia, PA 19107, USA. Email: stephen.silberstein@jefferson.edu

102
The anticonvulsant gabapentin has been used to treat various pain syndromes, including neuropathic pain (8), bromyalgia (9), cancer-related pain (10), and postoperative pain (11) and has been studied for migraine prophylaxis (1215). Results of open-label studies (12,13) as well as two randomized, doubleblind, placebo-controlled studies (14,15) are consistent with the ecacy of gabapentin for migraine prophylaxis; however, the ability to draw conclusions from the placebo-controlled studies is limited because of their methodological and analytical limitations. Citing the shortcomings of currently available evidence, recent reviews, including a Cochrane review, conclude that further evaluation of gabapentin in migraine prophylaxis is warranted in order to inform clinical practice (2,16). The randomized, double-blind, placebo-controlled, phase II trial reported herein was conducted to evaluate the ecacy of gabapentin enacarbil (GEn) for migraine headache prophylaxis. This phase II study had planned doses selected from the range of doses that could possibly oer ecacy and be safe and potentially well tolerated at the same time. However, there was no indication at the time of study planning which dose would be most eective. It was hypothesized that the higher doses may not show ecacy if patients experienced adverse events and dropped out, and both 1800 mg and 2400 mg doses seemed like good candidates to establish ecacy per the exposure curve. Thus, in order to preserve the alpha-level and allow a chance to test both middle doses in the case where only one of them may have been signicant, the average of the two was tested rst (if the average was signicant, then at least one pairwise comparison had to be signicant). The 1200 mg group was included to establish the minimum eective dose. GEn is a transported prodrug of gabapentin that provides sustained, dose-proportional exposure to gabapentin (17). GEn is rapidly converted to gabapentin upon absorption from the intestinal lumen. It has no known pharmacological activity of its own, and its activity is attributable to the release of the active form, gabapentin. In studies in which circulating prodrug levels were measured, including doses ranging from 600 mg to 6000 mg GEn, GEn exposure has consistently represented less than 0.5% of the corresponding gabapentin systemic exposure, based on the area under the curve (AUC).

Cephalalgia 33(2) an onset before the age of 50 years were eligible if they had ! three migraine headache attacks and ! four migraine headache days (dened as calendar days with any occurrence of migraine headache pain of at least 30 minutes in duration) per month but <15 migraine or non-migraine headache days per month during each of the three months prior to the screening visit and during the baseline period (dened below) and if their migraines were consistent over time in incidence and severity. Females were eligible if they were unable to bear children or, if able to bear children, if they were not pregnant and using adequate contraception. Patients were ineligible if they were unable to discontinue prohibited medications (beta-blockers, tricyclic antidepressants, calcium channel blockers, antiepileptic drugs, bupropion, serotonergic noradrenergic reuptake inhibitors) during the two-week screening period and throughout the duration of the study (uoxetine, riboavin, magnesium and feverfew were allowed); had a history of ergotamine, triptan, opioid, or combination medication intake for !10 days per month or simple analgesic intake for !15 days per month for ! three months; had previously taken gabapentin or pregabalin for migraine headache prophylaxis; the patient reported experiencing lack of ecacy of two or more ! eightweek trials of prophylaxis of migraine headache; or had uncontrolled hypertension (i.e. sitting systolic blood pressure >160 mmHg or sitting diastolic blood pressure >90 mmHg) at the screening visit or at randomization. All patients provided written informed consent.

Procedures
The protocol for this randomized, double-blind, parallel-group, placebo-controlled trial (GlaxoSmithKline protocol MPX111381; NCT identier #NCT00 742209) was approved by ethics committee or institutional review boards for each of the 51 United States and Canadian study sites. The study comprised a twoweek screening period during which patients eligibility was determined; a six-week baseline period for conrming eligibility and establishing baseline measures; a 20week treatment period (consisting of ve weeks of exible titration to the assigned target dose or the maximum tolerated dose, 12 weeks of maintenance treatment at the assigned or maximum tolerated dose, and three weeks of tapered discontinuation of study medication); and a two-week post-treatment period during which data on adverse events were collected. Patients who were currently on prohibited medications were to discontinue these medications during the two-week screening period prior to enrolling in the study and beginning the six-week baseline phase. Patients meeting eligibility criteria at screening and during the baseline period were randomized 2:1:2:2:1

Methods Patients
Males and females !18 years old with International Headache Society (IHS) criteria-dened migraine headache with or without aura (18) for at least one year with

Silberstein et al. to one of the following ve groups, designated according to target daily dose of study medication during the treatment period: placebo, GEn 1200 mg, GEn 1800 mg, GEn 2400 mg, or GEn 3000 mg. Blocks of randomization code were assigned to each center. Treatment group allocation was conducted via an Interactive Voice Recognition System (IVRS) at the time of randomization. Doses were titrated according to a exible schedule to reach the following target doses or maximum tolerated doses within a treatment group: 600 or 1200 mg/day for the GEn 1200 mg group; 600, 1200, or 1800 mg/day for the GEn 1800 mg group; 600, 1200, 1800, or 2400 mg/day for the GEn 2400 mg group; or 600, 1200, 1800, 2400, or 3000 mg/day for the GEn 3000 mg group. The exible titration regimen was designed to allow patients to reach the target dose for maintenance treatment or, if unable to reach this target dose, to achieve a maximum tolerated dose for maintenance treatment. Patients had the opportunity to undergo a single downward adjustment in dose (by 600 mg/day) at the end of week two, three, four, or ve of the exible titration period in the event that a dose was not tolerated well. If the dose was adjusted downward once, no additional downward or upward dose adjustments were permitted. Study medication was administered twice daily (morning and evening) with food. Use of acute migraine medication was permitted for breakthrough migraine attacks. Each day during the baseline period and the treatment period, patients recorded information about the presence or absence of migraine or non-migraine headache and associated symptoms of nausea, vomiting, photophobia, and phonophobia in an electronic diary. For breakthrough migraines, patients also recorded the date and time of the start and the resolution of the migraine, pain severity (0 none; 1 mild; 2 moderate; 3 severe) before the use of medication for breakthrough migraine, and maximum pain severity. Additional migraine medications, if taken, were recorded in a paper diary.

103
from baseline in the number of migraine headache days for all study phases except for the last four weeks of treatment prior to taper (which was the primary endpoint), the number of migraine attacks, the number of migraine headache periods, mean migraine attack duration, the mean peak migraine pain severity (utilizing the four-point pain scale), acute migraine medication use (type of medication, number of days of medication use, number of doses) compared with placebo, and percentage of migraine attacks with each of the following migraine symptoms: aura, nausea, vomiting, photophobia, phonophobia. Baseline calculations for all endpoints used the last four weeks of the baseline period. A migraine headache period was dened as a 24-hour segment of time with migraine. Additional secondary ecacy endpoints included the percentages of patients with !50% reductions in migraine headache days, migraine attacks, and migraine headache periods. Health outcomes measures (Clinician Global Impression of Change, Patient Global Impression of Change, Migraine-Specic Quality of Life Questionnaire, Headache Impact Test (HIT)-6 survey, productivity assessments, satisfaction assessments) were obtained, but health outcomes results are not reported herein since the primary and secondary endpoints lacked signicance, and the authors concluded that the data would be dicult to interpret. Based on the randomization ratio of 2:1:2:2:1 for placebo and GEn 1200, 1800, 2400, and 3000 mg/day, respectively, 112 subjects in each of the placebo, GEn 1800 mg and GEn 2400 mg groups were determined to provide 80% power to detect a dierence in the dose response when the true dierence between placebo and the GEn 1800 mg and GEn 2400 mg groups combined in the change from baseline in the number of migraine headache days was 1.3 days. This estimate was based on a two-sided analysis of variance (ANOVA), an alpha level of 0.05, and a standard deviation (SD) of 4.0. Ecacy data were analyzed for the intent-to-treat population, dened as randomized patients who received at least one dose of study medication and provided at least one post-baseline ecacy assessment. For the last four weeks prior to taper for the primary endpoint, missing data were handled in three ways. Method 1 extended the calendar earlier into the treatment period until 28 days of non-missing data contributed to the count of migraine headache days. Method 2 proportionately adjusted the number of migraine headache days (multiplied by 28 and divided by the number of non-missing days). Method 3 treated all missing days as non-migraine headache days. An analysis of covariance (ANCOVA) model was used to test the null hypothesis of no dierence between placebo and the average of the values for the GEn 1800 mg group and

Measures and data analysis

Efficacy. The primary ecacy endpoint was the change from baseline in the number of migraine headache days during the last four weeks of treatment prior to taper. A migraine headache day was dened as a calendar day with any occurrence of migraine headache pain of at least 30 minutes in duration. Secondary ecacy endpoints were calculated separately for the ve-week exible titration phase; the rst, second, and third fourweek segments of the maintenance phase; the entire maintenance phase; the entire treatment period; and the last four weeks of the treatment period prior to taper. Secondary ecacy endpoints included changes

104
the GEn 2400 mg group for the primary endpoint. The main eects model included baseline number of migraine headache days and IHS Headache Classication for presence or absence of aura. Additional linear contrasts of each active treatment group versus placebo were tested using the nal model. The results of each contrast were summarized using the adjusted mean and standard error (SE) for each treatment group, a 95% condence interval (CI) for the change from baseline for each treatment group, a model estimate of the dierence between each active treatment group and placebo, a 95% CI for the dierence, and an associated p value and adjusted p value for the dierence. The analysis of the primary endpoint was carried out using a combination of a sequential method and a Hochberg procedure (19) to maintain the experiment-wise alpha level of 0.05. The rst step in the testing sequence compared the average of the GEn 1800 mg group and the GEn 2400 mg group with placebo. If the comparison was signicant at  0.05, then the second step compared each of these treatment groups with placebo separately. If both comparisons were signicant at  0.05, then testing of other dose groups continued; otherwise, formal testing stopped. For the third step in the testing sequence, if both GEn 1200 mg and GEn 3000 mg compared with placebo were signicant at  0.05, then testing was completed. If one comparison was >0.05, then the other comparison was conducted at  0.025. Adjustment for multiplicity was not made for other ecacy analyses. For the primary endpoint, sensitivity analyses were performed to conrm the ndings of the primary analysis. Parametric ANCOVAs were performed using imputation methods dierent from those of the primary analysis. Additionally, a nonparametric ANCOVA using adjusted residuals and a second nonparametric ANCOVA using the residuals of the ranks were conducted. Finally, the primary analysis was repeated using the per protocol population, dened as patients in the intent-to-treat population, excluding those who had <80% compliance with the study medication; had < two weeks of non-missing diary data during the rst four weeks of maintenance treatment; used prohibited medications during the baseline or treatment periods; who did not meet inclusion criteria for the minimum number of attacks or days with migraine during baseline; or who had <14 days of non-missing diary data during the last 28 days of the baseline period. Formal hypothesis testing of secondary endpoints was not conducted. Pharmacokinetics. For pharmacokinetic analyses, four blood samples were taken from a subset of patients at various time points post-dose at treatment weeks ve,

Cephalalgia 33(2) nine, 13, and 17 (or premature withdrawal). The sampling times were evenly distributed across patients via a randomization method for the post-dose time points of two, four, eight, 10, and 12 hours. GEn exposures at steady state for each patient were estimated via nonlinear mixed-eect modeling analysis to verify whether the GEn plasma concentrations were as expected and to estimate the individual exposures for relation to ecacy or safety endpoints if needed.

Tolerability and safety. Tolerability and safety were assessed via adverse event monitoring, clinical laboratory tests, vital sign assessment, and electrocardiograms (ECGs). An adverse event was dened as any untoward medical occurrence, regardless of its suspected cause, that was reported by a patient or noted by a clinician during the study. For each adverse event, investigators recorded whether they considered it related to study medication. Tolerability measures included the incidences of adverse events regardless of suspected cause, adverse events leading to premature withdrawal from the study, and serious adverse events (dened as adverse events that resulted in death, disability, or incapacity; were life threatening; required prolonged hospitalization; or were a congenital anomaly or birth defect). Adverse-event data were summarized for the safety population, dened as randomized patients who were administered at least one dose of study medication. Inferential statistics were not calculated on the adverse-event data.

Results Patients
The number of patients randomized to treatment was 526 (n 129 placebo, n 67 GEn 1200 mg, n 134 GEn 1800 mg, n 134 GEn 2400 mg, n 62 GEn 3000 mg). Three patients were randomized but did not take study medication and therefore were not included in the intent-to-treat or safety populations. All remaining patients (n 128 placebo, n 66 GEn 1200 mg, n 134 GEn 1800 mg, n 133 GEn 2400 mg, n 62 GEn 3000 mg) were included in both the intent-to-treat population and the safety population. Across treatment groups, 60% to 74% of patients completed the study (Table 1). The most common reason for premature withdrawal for all patients randomized was adverse events (Table 1). Demographics and baseline clinical characteristics were similar among treatment groups (Table 1). The median daily dose was the target dose in each GEn group (Table 1).

Table 1. Patient disposition, demographics, baseline clinical characteristics, and study medication exposure.

Patient disposition

Silberstein et al.

Randomized patients

Placebo n 129

GEn 1200 mg n 67

GEn 1800 mg n 134

GEn 2400 mg n 134

Gen 3000 mg n 62

Completion status, n (%) 95 (74) 34 (26) 11 (9) 8 (6) 6 (5) 3 (2) 6 (5) 0 Placebo n 128 41.1 (11.72) 111 (87) 108 (84) 85 (66) 43 (34) 9.3 (2.59) 415 414 9.1 (2.76) 26 (39) 40 (61) 54 (82) 52 (79) 39.4 (9.74) 37.7 (11.75) 115 (86) 107 (80) 99 (74) 35 (26) 9.2 (2.89) 114 Gen 1200 mg n 66 GEn 1800 mg n 134 0 5 (4) 4 (6) 1 (1) 5 (4) 1 (<1) 5 (7) 4 (3) 5 (4) 5 (4) 3 (2) 1 (<1) GEn 2400 mg n 133 39.0 (12.04) 105 (79) 112 (84) 93 (70) 40 (30) 9.2 (2.82) 115 4 (6) 14 (10) 7 (5) 4 (6) 17 (13) 16 (12) 18 (27) 46 (34) 37 (28) 49 (73) 88 (66) 97 (72) 37 (60) 25 (40) 13 (21) 4 (6) 3 (5) 3 (5) 1 (2) 1 (2) GEn 3000 mg n 62 39.1 (11.78) 46 (74) 53 (85) 42 (68) 20 (32) 9.0 (2.97) 014

Completed

Withdrawn

Primary reason for withdrawal, n (%)

Adverse event

Withdrew consent

Protocol deviation

Lost to follow-up Lack of efficacy

Investigator discretion

Demographics and baseline clinical characteristics

Intent-to-treat/safety populationsa

Mean (SD) age, years

Female, n (%)

White race, n (%)

Body mass index, n (%)

30 kg/m2

>30 kg/m2

Historical number of migraine headache days/monthb Mean (SD)

Range

Exposure to study medication na na na na na na na 1077.8 (243.9) 1200 11 (20) 43 (80) na na na 1701.8 (263.7) 1800 3 (3) 12 (11) 95 (86) na na 2203.5 (463.7) 2400 4 (4) 9 (8) 7 (6) 93 (82) na 2776.5 (562.4) 3000 1 (2) 2 (4) 4 (8) 1 (2) 43 (84)

Mean (SD) daily dose during maintenance period

Median daily dose

Maximum tolerated dose, n (%)

600 mg 1200 mg

1800 mg

2400 mg

3000 mg

Three patients were randomized but did not take study medication and therefore were not included in the intent-to-treat or safety populations. Historical number of migraine headache days per month is the number reported by the patient at screening. GEn: gabapentin enacarbil; na: not applicable.

105

106
Table 2. Summary of primary endpoint based on the last four weeks of treatment before taper. Change from baseline in number of migraine headache daysa Adjusted mean difference of GEn vs placebo (95% CI) p value
a

Cephalalgia 33(2)

Gen 1800/2400 mg n 261 0.3 (0.6, 1.1) 0.579

Gen 1800 mg n 131 0.0 (1.3, 1.3) >0.999

Gen 2400 mg n 130 0.5 (0.8, 1.8) 0.783

Gen 1200 mg n 63 0.6 (1.0, 2.2) 0.775

Gen 3000 mg n 62 0.3 (1.4, 1.9) 0.987

Migraine headache days: calendar days with any occurrence of migraine headache pain of at least 30 minutes in duration. GEn: gabapentin enacarbil; CI: confidence interval. For the primary endpoint, three patients had missing values at baseline and eight patients had missing values after applying Method 1 imputation, resulting in a total of nine patients with missing change from baseline.

Efficacy
No statistically signicant dierence between active treatment (the average of 1800 mg and 2400 mg treatment groups) and placebo was found for change from baseline in the number of migraine headache days during the last four weeks of treatment prior to taper (the primary endpoint). The mean dierence between active treatment and placebo was 0.3 days (95% CI 0.6, 1.1; p 0.579) (Table 2). Given the hierarchal testing structure, formal hypothesis testing was stopped once this nonsignicant result was identied, but nominal p values for the comparison of each GEn dose versus placebo are shown in Table 2 for descriptive purposes. Results of analyses of the primary endpoint using the per protocol population, analyses using imputation methods dierent from those of the primary analysis, and nonparametric analyses of the primary endpoint were consistent with the primary analysis in showing no dierence between active treatment and placebo. The pattern of results was similar for the secondary ecacy endpoints (see Table 3 for results for selected endpoints during the last four weeks of treatment before taper).

Pharmacokinetics
Steady-state plasma exposures of GEn were as expected (Table 4). The average daily plasma concentration of GEn at steady state (Cavgss) and the 24-hour area under the plasma concentration-time prole for GEn at steady state AUC(024)ss appeared proportional to dose.

Safety and tolerability

The most common adverse events reported during the treatment period were dizziness, fatigue, nausea and somnolence (Table 5). In the safety population, adverse events led to premature withdrawal of 11 patients in the placebo group (9%), four in the GEn 1200 mg group (6%), 16 in the GEn 1800 mg group (12%), 15 in the GEn 2400 mg group (11%), and 13 in the GEn 3000 mg

group (21%). The only adverse event that led to premature withdrawal in >2% of patients in a treatment group was dizziness, which was the reason for the premature withdrawal of one patient in the placebo group, 0 patients in the GEn 1200 mg and GEn 3000 mg groups, four in the GEn 1800 mg group, and three in the GEn 2400 mg group. Two deaths occurred during the study. Both occurred during the taper phase of the treatment period. The rst death, bronchopneumonia related to drug use other than study medication, occurred in a 42year-old female randomized to GEn 1800 mg/day. The investigator did not consider this event to be related to the study medication. Among the substances found in this patients blood as listed in the toxicology report were cocaine metabolites, oxycodone, alprazolam, carisoprodol, citalopram, clonazepam, and mirtazapine. The second death, ruled an accident attributed to combined toxicity of multiple drugs, occurred in a 31-yearold male randomized to GEn 2400 mg/day. The investigators judgment about relatedness of this event to study medication was undetermined at the conclusion of the study. The patient had a history of prescription drug and alcohol abuse and on the day of his death was found with several prescription bottles containing oxycodone and alprazolam as well as several empty liquor bottles and an open pack of beer. Nine non-fatal serious adverse events were reported in eight patients during the treatment period. Eight of the nine events (appendicitis, cholecystitis, cholelithiasis, metastatic malignant melanoma, muscle spasms, convulsion, pneumonia, and pharyngitis) were not considered to be related to study medication; the remaining event, reported as conversion disorder, was considered possibly to be related to study medication. The conversion disorder was moderate pseudoseizure occurring on two occasions after sexual intercourse. The pseudoseizures were manifested as focal left upper extremity jerking and turning of the head to the left side. No bowel or bladder incontinence, tongue biting, or alteration in consciousness occurred with the pseudoseizure. The patient was withdrawn from the study for this event, which resolved without further action.

Table 3. Summary of secondary endpoints based on the last four weeks of treatment before taper. Placebo n 128 67/123 (54) 67/123 (54) 2.2 (2.8, 1.7) 3.7 (8.8, 5.9) 0.0 (0.1, 0.0) 2.1 (2.4, 1.8) 3.0 (3.6, 2.3) GEn 2400 mg n 133 0.7 (3.0) 0.5 (3.5) 3.7 (3.1) 5.5 (2.6) 7.4 (2.9) 0.6(6.5, 5.7) 0.0 (0.3, 0.0) 2.2 (2.7, 1.8) 3.0 (3.9, 2.0) GEn 1200 mg n 66 3.4 (4.8) 2.0 (5.7) 0.9 (3.0) 3.5 (4.1) 1.2 (3.1) 2.3 (3.1, 1.5) 31/59 (53) 26/59 (44) 38/58 (66) 39/58 (67) 2.1 (2.9, 1.3) 4.5(9.7, 11.4) 0.0 (0.3, 0.0) 2.6 (3.0,2.1) 3.2 (4.2, 2.2) GEn 3000 mg n 62 1.2 (4.4) 5.0 (4.7) 0.4 (4.8) 0.1 (2.5) 3.6 (1.9) GEn 1800 mg n 134 GEn 2400 mg n 133 GEn 1200 mg n 66 GEn 3000 mg n 62

Silberstein et al.

Test sequence/endpoint (unit)

!50% reduction in migraine headache days (days)a Proportion (%) 65/120 (54) 68/114 (60) !50% Reduction in migraine headache attacks (attacks)b,c Proportion (%) 64/120 (53) 67/114 (59) Change from baseline in number of days of acute migraine medication use (days) Adjusted mean (95% CI) 2.0 (2.5, 1.4) 2.7 (3.3, 2.2) Change from baseline in mean migraine attack duration (attacks)d Median (95% CI) 4.4 (8.7, 2.5) 4.8 (10.6, 0.0) Change from baseline in mean peak migraine pain severity (attacks)e Median (95% CI) 0.0 (0.2, 0.0) 0.0 (0.2, 0.0) Change from baseline in number of migraine attacks (attacks) Adjusted mean (95% CI) 2.2 (2.5, 1.8) 2.3 (2.6, 2.0) Change from baseline in number of migraine headache periods (periods)f Adjusted mean (95% CI) 3.3 (3.9, 2.6) 3.6 (4.3, 2.9) Placebo n 128 7.4 (3.2) 5.2 (3.9) 0.4 (3.0) 2.1 (2.9) 0.7 (3.4) GEn 1800 mg n 134

Other secondary endpoints (unit)

Change from Mean (SE) Change from Mean (SE) Change from Mean (SE) Change from Mean (SE) Change from Mean (SE)

baseline in % migraine attacks with aura (attacks) 7.0 (3.6) baseline in % migraine attacks with nausea (attacks) 8.0 (3.7) baseline in % migraine attacks with vomiting (attacks) 0.0 (2.6) baseline in % migraine attacks with photophobia (attacks) 1.9 (2.8) baseline in % migraine attacks with phonophobia (attacks) 5.4 (3.1)

Migraine headache days: calendar days with any occurrence of migraine headache pain of at least 30 minutes in duration. This secondary endpoint was not a part of the testing sequence. c Migraine attacks: a migraine headache of at least 30 minutes in duration and can include recurring non-migraine or migraine headaches. d Mean peak migraine attack duration: the total duration of the attack, which could comprise several headache events. e Mean peak migraine severity: maximum pain severity across all headache events considered to be one attack. f Migraine headache period: 24-hour block of time that begins at the onset of a migraine event. GEn: gabapentin enacarbil; CI: confidence interval. The sample size for the intent-to-treat population is reported, but the actual sample size varies depending on the number of patients with non-missing data for each endpoint.

107

108
Table 4. Estimated individual pharmacokinetic parameters. Parameter AUC(024)ss (ug*hr/ml) Dose (mg/day) 600 1200 1800 2400 3000 600 1200 1800 2400 3000 600 1200 1800 2400 3000 600 1200 1800 2400 3000 600 1200 1800 2400 3000 n 12 39 72 60 25 12 39 72 60 25 12 39 72 60 25 12 39 72 60 25 12 39 72 60 25 Geometric mean 58.0 94.2 149 188 223 2.42 3.92 6.19 7.85 9.28 4.25 4.77 8.68 9.54 12.4 1.00 3.03 4.01 6.07 6.43 4.23 1.57 2.16 1.57 1.93 Median 58.0 89.8 154 190 226 2.42 3.74 6.41 7.93 9.42 4.13 4.55 8.82 9.64 12.9 0.886 2.88 4.15 5.88 7.26 4.39 1.57 2.11 1.53 1.92

Cephalalgia 33(2)

CV% 20.8 26.0 27.3 28.8 33.9 20.8 26.0 27.3 28.8 33.9 21.3 24.4 23.3 27.9 31.5 37.0 29.8 36.4 31.5 39.3 36.9 11.1 18.2 9.73 12.9

Cavgss (ug/ml)

Cmaxss (ug/ml)

Cminss (ug/ml)

Peak-trough ratio

AUC(024)ss: 24-hour area under the plasma concentration-time profile for gabapentin enacarbil (GEn) at steady state; Cavgss: average daily plasma concentration of GEn at steady state; Cmaxss: maximum daily plasma concentration of GEn at steady state; Cminss: minimum daily plasma concentration of GEn at steady state; CV%: coefficient of variation.

No signicant changes in laboratory values or ECGs were observed among patients receiving GEn. Likewise, no notable changes in vital signs were observed.

Discussion
This randomized, double-blind, phase II clinical trial did not demonstrate a statistically signicant benet of GEn over placebo in the prophylaxis of migraine. No statistically signicant dierence between GEn 1800/2400 mg and placebo was observed in the change in the number of migraine headache days from baseline during the last four weeks of treatment in the intent-totreat population (primary endpoint) (p 0.579). Changes from baseline in the number of migraine headache days during the last four weeks of treatment prior to taper were of similar magnitude across treatment groups that received GEn (1200, 1800, 2400, or

3000 mg/day) and placebo. The results were consistent across three methods of imputation applied to the primary endpoint and in an analysis of the primary endpoint for the per protocol population. Although the withdrawal rate was higher in the treatment groups than for placebo (34/129, 26%), this dierence is not likely to be disparate enough to cause the lack of signicance observed for the primary analysis. In addition, the primary endpoint was dened in a robust manner, using the last four weeks of treatment prior to taper in the calculation of the change from baseline. Thus, only nine patients had missing data for the primary analysis. Formal hypothesis testing of the secondary ecacy endpoints was not undertaken, but numerical data show that GEn failed to separate from placebo on these endpoints, like the primary endpoint. The placebo response in the present study fell at the upper end of the range of placebo responses seen

Silberstein et al.
Table 5. Treatment-emergent adverse events reported in !5% of patients in a treatment group. Placebo n 128 GEn 1200 mg n 66 GEn 1800 mg n 134 n (%) Any event Dizzinessa Fatigue Nausea Somnolence Weight increased Upper respiratory tract infection Constipation Dry mouth Nasopharyngitis Diarrhea Vomiting Influenza Insomnia Peripheral edema Sinusitis Balance disorder Abdominal pain Back pain Cough
a

109

GEn 2400 mg n 133

GEn 3000 mg n 62

87 (68) 8 (6) 9 (7) 12 (9) 6 (5) 7 (5) 9 (7) 3 (2) 3 (2) 8 (6) 8 (6) 5 (4) 4 (3) 1 (<1) 4 (3) 3 (2) 1 (<1) 1 (<1) 0 0

44 (67) 16 (24) 10 (15) 3 (5) 6 (9) 4 (6) 4 (6) 4 (6) 4 (6) 3 (5) 1 (2) 1 (2) 1 (2) 4 (6) 4 (6) 4 (6) 2 (3) 2 (3) 1 (2) 3 (5)

99 (74) 43 (32) 12 (9) 15 (11) 7 (5) 8 (6) 4(3) 7 (5) 6 (4) 4 (3) 1 (<1) 3 (2) 3 (2) 1 (<1) 1 (<1) 3 (2) 2 (1) 2(1) 6 (4) 1 (<1)

101 (76) 35 (26) 14 (11) 12 (9) 14 (11) 9 (7) 9 (7) 8 (6) 5 (4) 4 (3) 7 (5) 7 (5) 4 (3) 6 (5) 3 (2) 3 (2) 6 (5) 3 (2) 1 (<1) 0

49 (79) 11 (18) 3 (5) 6 (10) 9 (15) 4 (6) 5 (8) 5 (8) 3 (5) 2 (3) 1 (2) 2 (3) 3 (5) 2 (3) 2 (3) 1 (2) 1 (2) 3 (5) 3 (5) 0

Numerically, dizziness occurred more frequently in all treatment groups compared to placebo. No statistical analyses were conducted. GEn: gabapentin enacarbil.

in previous migraine prophylaxis studies (20) and needs to be considered when interpreting the treatment eect that could be achieved. There could be several factors that contributed to the placebo response. In the literature, a meta-analysis of placebo response in the prophylactic treatment of migraine showed that the pooled estimate of the placebo response (percentage of placebo-treated subjects who had a reduction in migraine attacks greater than 50% at endpoint) was 21% but there was a large range across individual studies (655%) (20). The placebo response rate seen in the current study (53% reduction for migraine attacks) fell within this range. This meta-analysis also reported that placebo response rates were signicantly higher in parallel group studies compared with crossover studies (20). The authors suggest that one possible explanation is the information given to the patients in the content of the informed consent. In parallel studies, the patients are aware that they will receive either study medication or placebo, by chance. The individuals expectation is that he or she will be on active treatment.

Therefore, ecacy as well as adverse events are maximized (20). In this study there were a high number of active treatment arms (four arms active treatment and one arm placebo). Additionally the informed consent contained the number of treatment arms, and elaborated that patients had a 4:5 chance of getting active treatment. This could have had an impact on how the patients responded to the subjective assessments in the study. Suboptimal exposure to GEn does not appear to explain the failure of GEn to separate from placebo as the estimated exposure to GEn in this study was consistent with that expected based on previous ndings for restless legs syndrome (RLS) studies (21,22). The adverse event prole of GEn was consistent with that in other studies with GEn or gabapentin (21,23). No new safety or tolerability ndings were observed. In conclusion, the results of this randomized, double-blind, placebo-controlled, phase II study did not demonstrate a dierence between GEn and placebo in migraine prophylaxis. The large placebo eect needs to be considered when interpreting the lack of eect of GEn over placebo seen in this study.

110 Clinical implications

Cephalalgia 33(2)

. The results of this randomized, double-blind, placebo-controlled, phase II study did not demonstrate a dierence between GEn and placebo in migraine prophylaxis. . The large placebo eect needs to be considered when interpreting the lack of eect of GEn over placebo seen in this study.

Acknowledgments
This protocol was posted on the Clinicaltrials.gov database, identier #NCT00742209. The authors thank Alison Wentz, PharmD, for her contributions to the protocol design and study conduct.

Funding
GlaxoSmithKline funded the research described in this manuscript. Author C.T. conducted the statistical analysis.

Conflict of interest
This study was sponsored and conducted by GlaxoSmithKline. The team from the sponsor (S.G.S., C.T., J.A.) in collaboration with coauthor S.S. led the design and conduct of the study; the management, analysis, and interpretation of the data; and preparation, review, and approval of this manuscript in conjunction and consultation with the investigators. All authors had full access to the data. S.S. has consulted for or conducted research funded by GlaxoSmithKline and other pharmaceutical companies. S.G.S., C.T., and J.A. are employees of GlaxoSmithKline. The work of J.S., a professional medical writer, on this manuscript was funded by GlaxoSmithKline.

References
1. Rapoport AM. Acute and prophylactic treatments for migraine: Present and future. Neurol Sci 2008; 29(Suppl 1): S110S122. 2. Mulleners WM and Chronicle EP. Anticonvulsants in migraine prophylaxis: A Cochrane review. Cephalalgia 2008; 28: 585597. 3. Evers S, Afra J, Frese A, et al. EFNS guideline on the drug treatment of migraine: Report of an EFNS task force. Eur J Neurol 2006; 13: 560572. 4. Silberstein SD. Practice parameter: Evidence-based guidelines for migraine headache. Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2000; 55: 754762. 5. Stark RJ and Stark CD. Migraine prophylaxis. MJA 2008; 189: 283288. 6. Cutrer RM, Limmroth V and Moskowitz MA. Possible mechanisms of valproate in migraine prophylaxis. Cephalalgia 1997; 17: 93100.

7. Rogawski MA. Common pathophysiologic mechanisms in migraine and epilepsy. Arch Neurol 2008; 65: 709714. 8. Gilron I. Gabapentin and pregabalin for chronic neuropathic and early postsurgical pain: Current evidence and future directions. Curr Opin Anaesthesiol 2007; 20: 456472. 9. Tzellos TG, Toulis KA, Goulis DG, et al. Gabapentin and pregabalin in the treatment of fibromyalgia: A systematic review and a meta-analysis. J Clin Pharm Ther 2010; 35: 639656. 10. Bar Ad V. Gabapentin for the treatment of cancer-related pain syndromes. Rev Recent Clin Trials 2010; 5: 174178. 11. Straube S, Derry S, Moore RA, et al. Singledose oral gabapentin for established acute postoperative pain in adults. Cochrane Database Syst Rev 2010; CD008183. 12. Jimenez-Hernandez MD, Torrecillas Narvaez MD and Friera Acebai G. Effectiveness and safety of gabapentin in the preventive treatment of migraine. Rev Neurol 2002; 35: 603606. 13. Vukovic V, Lovrencic-Huzjan A, Bosnar-Puretic M, et al. The efficacy of gabapentin in migraine prophylaxis: An observational open-label study. Acta Clin Croat 2009; 48: 145151. 14. Di Trapani G, Mei D, Marra C, et al. Gabapentin in the prophylaxis of migraine: A double-blind randomized placebo-controlled study. Clin Ter 2000; 151: 145148. 15. Mathew NT, Rapoport A, Saper J, et al. Efficacy of gabapentin in migraine prophylaxis. Headache 2001; 41: 119128. 16. Vikelis M and Rapoport AM. Role of antiepileptic drugs as preventive agents for migraine. CNS Drugs 2010; 24: 2133. 17. Cundy KC, Sastry S, Luo W, et al. Clinical pharmacokinetics of XP13512, a novel transported prodrug of gabapentin. J Clin Pharmacol 2008; 48: 13781388. 18. Headache Classification Committee of the IHS. International Classification of Headache Disorders (ICHD-II), 2nd edition. Cephalalgia 2004; 24(Suppl 1): 1160. 19. Hochberg Y. A sharper Bonferroni procedure for multiple tests of significance. Biometrika 1988; 75: 800802. 20. Macedo A, Banos JE and Farre M. Placebo response in the prophylaxis of migraine. Eur J Pain 2008; 12: 6875. 21. Lal R, Sukbuntherng J, Luo W, et al. Pharmacokinetics and tolerability of single escalating doses of gabapentin enacarbil: A randomized-sequence, double-blind, placebo-controlled crossover study in healthy volunteers. Clin Ther 2009; 31: 17761786.

Silberstein et al.
22. Merlino G, Serafini A, Young JJ, et al. Gabapentin enacarbil, a gabapentin prodrug for the treatment of the neurological symptoms associated with disorders such as restless legs syndrome. Curr Opin Investig Drugs 2009; 10: 91102.

111
23. Irving G, Jensen M, Cramer M, et al. Efficacy and tolerability of gastric-retentive gabapentin for the treatment of postherpetic neuralgia: Results of a double-blind, randomized, placebo-controlled clinical trial. Clin J Pain 2009; 25: 185192.