Alimentary Pharmacology and Therapeutics

Randomised clinical trial: arbaclofen placarbil in gastro-oesophageal reux disease insights into study design for transient lower sphincter relaxation inhibitors
N. B. Vakil*, F. J. Huff & K. C. Cundy

*University of Wisconsin School of Medicine and Public Health, Milwaukee, WI, USA. XenoPort, Inc., Santa Clara, CA, USA.

SUMMARY Background Arbaclofen placarbil is a pro-drug of the gamma-aminobutyric acid-B agonist R-baclofen that has been shown to reduce reux episodes in patients with gastro-oesophageal reux disease (GERD). Aim To evaluate the efcacy and safety of arbaclofen placarbil vs. placebo as adjunctive therapy in subjects with troublesome GERD symptoms despite therapy with once-daily doses of a proton pump inhibitor (PPI) and to identify the characteristics of patients who were responders. Methods Patients (n = 460) with symptomatic GERD experiencing troublesome symptoms on once-daily PPI therapy were enrolled in this phase II, randomised, multicentre, double-blind, placebo-controlled, dose-ranging study. Patients were randomised to receive placebo or arbaclofen placarbil (20 or 40 mg once daily, 20 or 30 mg twice daily) with their current PPI for 6 weeks. Patients recorded heartburn and other GERD symptoms in a daily diary and rated severity of each event. The primary endpoint was percent change from baseline in heartburn events per week. Results In the primary analysis, there was no signicant difference between arbaclofen placarbil and placebo. Post hoc analyses removing mild and very mild heartburn events resulted in greater percent reductions for all arbaclofen placarbil doses with nominal P values <0.05 for each dose compared with placebo. There was a dose-related increase for the most common adverse events. Conclusions Arbaclofen placarbil was not superior to placebo in the primary analysis. Post hoc analyses suggest that subjects with more clinically relevant moderate or severe symptoms are more likely to respond to arbaclofen placarbil (clinicaltrials.gov NCT00978016). Aliment Pharmacol Ther 2013; 38: 107117

Correspondence to: Dr N. B. Vakil, Aurora Summit Hospital, 36500 Aurora Drive, Summit, WI 53066, USA. E-mail: nimish.vakil@gmail.com

Publication data Submitted 27 March 2013 First decision 11 April 2013 Resubmitted 13 May 2013 Accepted 14 May 2013 Ev Pub Online 30 May 2013

2013 John Wiley & Sons Ltd doi:10.1111/apt.12363

107

N. B. Vakil et al. INTRODUCTION Despite the success of proton pump inhibitor (PPI) therapy in the management of gastro-oesophageal reux disease (GERD), a number of patients need additional therapy with other agents. In a study in the primary-care setting, only 36% of patients receiving prescription therapy reported that they were currently asymptomatic; 20.5% of patients were also taking at least one over-thecounter medication.1 To address some of these needs, an alternative approach has been explored, using drugs developed specically to inhibit the transient lower oesophageal sphincter relaxation. Results in preliminary clinical studies25 have been very promising. Despite these data, the results in larger clinical trials have been less than satisfactory. In a phase III, randomised controlled trial of lesogaberan as add-on therapy in patients with a partial response to PPI therapy, the additional benet of lesogaberan therapy on GERD symptoms was marginal.6 Similar results were seen with arbaclofen placarbil in a phase II, randomised controlled trial of patients with GERD symptoms.7 Despite these negative results, interest in the development of reux-modifying drugs remains high. Key questions remain regarding the selection of patients for trials with reux-modifying agents. Patients with persisting symptoms despite PPI therapy may have acid reux, weakly acidic reux, functional heartburn or functional dyspepsia. It has been suggested that patients who respond may be identied by the presence of reux symptoms, e.g. regurgitation, but little information is currently available from clinical trials.8 The aim of this study was to determine the safety and efcacy of adjunctive arbaclofen placarbil (XP19986) in relieving symptoms of heartburn in patients with an incomplete response to PPI therapy. A secondary aim of this study was to identify the characteristics of patients who were responders to better dene the population for future studies. Another goal was to demonstrate efcacy for arbaclofen placarbil using less frequent dosing than is required for racemic baclofen, which is dosed three or four times a day. Doses of arbaclofen placarbil given once (q.d.s.) and twice a day (b.d.) were selected based on the results of previous studies,4, 7 with the aim of identifying an effective dose that would be well tolerated. MATERIALS AND METHODS Study design This multicentre, parallel-group, randomised, doubleblind, placebo-controlled, dose-ranging study was con108

ducted in subjects with symptomatic GERD who continued to experience troublesome symptoms of any severity while taking a PPI at the allowable dose and regimen on a daily basis. This study was conducted in accordance with good clinical practice guidelines, all applicable regulatory requirements and the guiding principles of the Declaration of Helsinki on biomedical research involving human subjects. Written informed consent was obtained from each subject before any study-related activities were performed. The protocol was approved by institutional review boards at participating institutions. The study was registered: clinicaltrials.gov NCT00978016.

Participants Subjects were between 18 and 65 years of age and had a diagnosis of GERD by a gastroenterologist prior to randomisation. Eligible subjects had experienced partial relief of GERD symptoms while receiving an approved regimen of PPI therapy, were taking an approved q.d.s. dose of PPI therapy for GERD for the 4 weeks prior to screening and had experienced GERD symptoms, dened as heartburn of any severity with or without regurgitation or other GERD symptoms, on 3 days during the week prior to screening. Patients on b.d. doses of a PPI were not permitted into the trial. Prior to randomisation, subjects should have been compliant with their PPI regimen during 3 weeks between the screening and baseline visits, and must have experienced 4 separate heartburn episodes of any severity on 3 days during the baseline week. Key exclusion criteria were evidence of erosive oesophagitis or erosive gastritis on endoscopy during screening; prior surgery of the upper gastrointestinal tract, including bariatric surgery; history of endoscopic therapy for GERD; structural abnormality of the gastrointestinal tract, including Barretts oesophagus, hiatus hernia >3 cm or oesophageal stricture; known oesophageal motility disorder, including oesophageal spasm or achalasia; delayed gastric emptying; active duodenal or gastric ulcer; recurrent pancreatitis or cholecystitis; inammatory bowel disease or scleroderma; antibiotic treatment for eradication of Helicobacter pylori within 30 days of screening or anticipated need for H. pylori treatment during the study; and chest pain that the investigator attributed to a motility disorder, such as oesophageal spasm. Other exclusion criteria were no improvement in GERD symptoms after an adequate course (2 weeks) of PPI treatment prior to screening, as this suggests that an aetiology other than GERD may be involved; abdominal pain or chest pain that the subAliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

Randomised clinical trial: arbaclofen placarbil in GERD

ject rated as more severe than retrosternal heartburn; and a score of 10 on either the anxiety or depression subscale of the Hospital Anxiety and Depression Scale (these criteria were intended to reduce the likelihood of enrolling patients with functional disorders). Subjects were required to discontinue and refrain from using histamine type 2 receptor antagonists, pro-kinetics and alginic agents for the duration of the study. Antacid tablets containing calcium carbonate (675 mg) and magnesium hydroxide (135 mg) were provided as rescue medication during the treatment period. provided as rescue medication during the treatment period. A subset of sites participated in an intense pharmacokinetic day on day 42 to include a moderate-fat meal, predose blood draw, investigational product dosing in clinic and hourly blood draws at 1, 2, 3, 4, 5, 6, 7 and 8 h post morning dose. The investigational treatment was then tapered during 0 to 4 further treatment days. A nal assessment occurred on day 49. GERD symptoms, morning and evening study-drug dosing (including investigational product and PPI) and night-time awakenings were recorded using an electronic diary. Subjects remained in the double-blind period for approximately 6 weeks, and then down titrated in a predetermined dosing regimen. Subjects returned to the clinic on a 2-week basis during the double-blind period.

Treatments The protocol called for randomising at least 425 subjects to 6 weeks of treatment to obtain approximately 72 evaluable subjects per treatment group. Four hundred and sixty subjects were randomised using a 1:1:1:1:1 allocation ratio to one of four active doses of arbaclofen placarbil + PPI or placebo + PPI. The screening period included collection of vital signs, safety labs, electrocardiogram (ECG), an oesophagogastroduodenoscopy and determination of frequency of GERD symptoms immediately preceding screening visit. The training week allowed subjects to train and familiarise themselves with the electronic diary. Baseline data regarding frequency and severity (very mild, mild, moderate, severe or very severe) of GERD symptoms while on q.d.s. PPI were recorded using an electronic diary. Approximately 3 weeks after the start of recording data in the electronic dairy, subjects returned to the clinic, at which time those who had acceptable safety assessments, showed electronic diary capability and had a minimum of four heartburn events (episodes) on 3 days were randomised to one of four active study arms or a placebo arm. During the study, subjects dosed the PPI in the morning, and dose time in relation to food followed the PPI label instructions. The investigational agent (arbaclofen placarbil or placebo) was taken with food, and it was recommended that doses be taken 1012 h apart. The treatment period included 04 days of blinded titration (depending on the double-dummy target dose), followed by 3642 days treatment at the target dose and then 04 days of blinded tapering from the target dose. Although higher dose levels had longer titration and tapering periods than lower dose levels, the primary efcacy evaluations occurred at steady state on the 42nd day of dosing in each treatment segment for all dose levels, and the total duration of treatment (including taper) was 46 days for all dose levels. Antacids were
Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

Efcacy measures Symptom diary. An electronic diary was used to capture the frequency and severity (very mild, mild, moderate, severe or very severe) of GERD symptoms of heartburn (dened as a burning discomfort or pain behind the breastbone), regurgitation (dened as a sensation of stomach contents rising into the throat or mouth), belching/burping and burning/pain in the throat. The diary also collected dosing information for the investigational product, PPI and rescue antacid. Other efcacy measures. Other efcacy measures included the Patient Assessment of Upper Gastrointestinal Disorders Quality of Life Questionnaire (PAGIQOL), SF-12 Health Survey and patient-rated Overall Treatment Effect Scale. Safety evaluations Safety was assessed based on the frequency, intensity and relationship of treatment-emergent adverse events (AEs), and on the results of clinical laboratory parameters, vital signs, ECG parameters and physical examination results. Sample size Approximately 85 subjects were to be randomised to each of ve treatment groups, for a total of approximately 425 subjects. Subjects were to be randomised using a 1:1:1:1:1 allocation ratio to four active doses of arbaclofen placarbil or placebo as adjunctive therapy to PPI to obtain 72 evaluable subjects per treatment arm. This allocation assumed a dropout rate of 15%. Randomisation was stratied by site and PPI dose level.
109

N. B. Vakil et al.
The sample size of 72 subjects per treatment arm was determined to detect a 20% difference in mean percent change from baseline to end of treatment in weekly heartburn frequency between one of the arbaclofen placarbil doses and placebo as adjunctive therapy to PPI, assuming a common standard deviation of 42.5%, a twosided t-test and a Type I error of 0.05 with at least 80% power. A post hoc statistical analysis plan was developed, which included analysis of heartburn events of at least moderate severity (based on the exclusion of mild and very mild heartburn events without eliminating subjects who had fewer than four heartburn events of at least moderate severity at baseline), and separate analyses of subjects with at least one regurgitation event of at least moderate severity at baseline. The prespecied statistical analysis methodology for symptom-related endpoints was used to analyse the post hoc symptom-related endpoints.

Statistical methods The main null hypothesis to be tested for the primary endpoint was that the mean percent change from baseline in heartburn events per week at week 6 is not different between the subjects treated with arbaclofen placarbil doses and placebo as adjunctive therapy to PPI. Hypotheses similar to the main null hypothesis were tested for the secondary endpoints. The primary efcacy endpoint was analysed using a repeated-measures analysis of covariance model with xed effects for treatment group, pooled site, PPI dose level, treatment week and treatment week by treatment group interaction as xed effects and baseline heartburn events per week as covariate. Least square (LS) means and differences between the arbaclofen placarbil doses and placebo and the corresponding 95% condence intervals (CIs) were calculated. The intent-to-treat (ITT) population was dened as all subjects who were randomised during the double-blind treatment period of the study. The full-analysis (FA) population was dened as the subset of the ITT population who had no major entry criteria violations, took at least one scheduled investigational product and had at least one available efcacy assessment collected post-randomisation. The FA population was the primary-analysis population for the efcacy analyses. Efcacy analyses The primary efcacy endpoint evaluated was percent change from baseline in heartburn events per week. The secondary endpoints included change from baseline in regurgitation events per week; percent change from baseline in the sum of heartburn and regurgitation events per week; proportion of responders with complete relief of heartburn (dened as no episodes of heartburn during week 6 of treatment); proportion of responders with complete relief of heartburn and regurgitation during week 6 of treatment; and change from baseline in weekly average of severity of heartburn and regurgitation, patient-rated Overall Treatment Effect Scale, PAGI-QOL and SF-12 Health Survey.
110

RESULTS Subject disposition and demographics There were 460 subjects included in the ITT population. There were 442 subjects included in the FA population, 455 subjects in the safety population and 328 subjects in the per-protocol population. There were 358 subjects who completed 6 weeks of treatment (subjects who completed visit 9), and 353 (76.7%) subjects who completed the study. Of the 107 who did not complete the study, 65 subjects discontinued due to an AE, 22 subjects terminated from the study due to the subjects decision, six subjects were lost to follow-up and six subjects discontinued due to protocol violation. There were 89 subjects randomised to the placebo group, 94 to the 20-mg q.d.s. group, 93 to the 40-mg q.d.s. group, 90 to the 20-mg b.d. group and 94 subjects to the 30-mg b.d. group. Of the 460 randomised subjects, 346 (75.2%) subjects used low-dose PPI (the approved dose for treatment of symptomatic GERD), and 114 (24.8%) subjects used high-dose PPI (the approved dose for treatment of erosive oesophagitis). Figure 1 shows the diagram of the progress through the phases of this study. Subjects were randomised across 58 investigative sites. Randomisation was stratied by site and PPI dose level (low or high). There was a statistically signicant difference among the treatment groups in the number of discontinued subjects (P = 0.037). The placebo group had 15.7% discontinued subjects, while the arbaclofen placarbil treatment groups had at least 20% discontinued subjects (ranging from 20.2% discontinuations in the 20-mg q.d.s. group to 28.7% discontinuations in the 30-mg b.d. group). An overview of the baseline demographics by treatment group for the ITT population is presented in Table 1. There were no statistically signicant differences
Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

Randomised clinical trial: arbaclofen placarbil in GERD

Randomised (n = 460)

Allocated to 20 mg q.d.s.* Allocated to 40 mg q.d.s.* Allocated to 20 mg b.d.* (n = 94) (n = 93) (n = 90) Low PPI dose 71 Low PPI dose 71 Low PPI dose 70 High PPI dose 23 High PPI dose 22 High PPI dose 20 Reason for withdrawal Adverse event 6 Lack of efficacy 1 Lost to follow-up 3 Non-compliance 0 Subject withdrawal 8 Protocol violation 0 Other 1 Completed study (n = 75) Reason for withdrawal Adverse event 18 Lack of efficacy 0 Lost to follow-up 2 Non-compliance 1 Subject withdrawal 4 Protocol violation 1 Other 0 Completed study (n = 67) Reason for withdrawal Adverse event 16 Lack of efficacy 0 Lost to follow-up 0 Non-compliance 1 Subject withdrawal 2 Protocol violation 1 Other 1 Completed study (n = 69)

Allocated to 30 mg b.d.* (n = 94) Low PPI dose 71 High PPI dose 23 Reason for withdrawal Adverse event 20 Lack of efficacy 1 Lost to follow-up 0 Non-compliance 0 Subject withdrawal 5 Protocol violation 0 Other 1 Completed study (n = 67)

Allocated to placebo (n = 89) Low PPI dose 63 High PPI dose 26 Reason for withdrawal Adverse event 5 Lack of efficacy 0 Lost to follow-up 1 Non-compliance 1 Subject withdrawal 3 Protocol violation 4 Other 0 Completed study (n = 75)

Figure 1 | Subject disposition. b.d., twice a day; PPI, proton pump inhibitor; q.d.s., once a day. * Arbaclofen placarbil.

Table 1 | Demographic characteristics (ITT population)

Arbaclofen placarbil Demographic characteristic Age (years), mean (s.d.) Sex (%) Male Female Race (%) White All others Weight (kg), mean (s.d.) BMI (kg/m2), mean (s.d.) Placebo (n = 89) 42.8 (13.32) 48.3 51.7 82.0 18.0 83.3 (15.32) 28.2 (4.21) 20 mg q.d.s. (n = 94) 42.5 (12.78) 46.8 53.2 80.9 19.1 83.4 (15.67) 28.4 (4.35) 40 mg q.d.s. (n = 93) 45.4 (11.68) 38.7 61.3 87.1 12.9 79.6 (13.65) 27.7 (3.66) 20 mg b.d. (n = 90) 45.6 (11.92) 50.0 50.0 78.9 21.1 83.6 (16.69) 28.6 (4.21) 30 mg b.d. (n = 94) 44.6 (11.62) 40.4 59.6 86.2 13.8 81.5 (15.19) 28.3 (3.93) Total (n = 460) 44.2 (12.29) 44.8 55.2 83.0 17.0 82.2 (15.34) 28.2 (4.07) P value 0.225 0.447

0.539 0.328 0.590

b.d., twice a day; BMI, body mass index; ITT, intent-to-treat; q.d.s., once a day; s.d., standard deviation.

in demographic characteristics among the treatment groups. A summary of selected baseline characteristics is given in Table 2. In the ITT population, all 460 (100%) subjects had dyspepsia/heartburn and GERD. Four hundred and one (87.2%) subjects had regurgitation. The other major gastrointestinal disorders included eructation (80.7%), abdominal distension/bloating (56.3%), fullness after eating (47.2%), gastrointestinal sounds abnormal (46.7%), atulence (44.1%), nausea (40.7%), hiatus hernia (38.5%), breath odour (27.2%) and gastritis (20.2%). The most frequently used concomitant medications (10% of subjects in any treatment group) in the safety
Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

population were the allowed rescue antacid (71.4%), paracetamol (16.9%) and multivitamins (16.9%). A subject was considered to be compliant if the subject had taken 80120% of the expected tablets of study medication. Average compliance rates ranged from 97.6% in the 30-mg b.d. group to 99.5% in the 20-mg q.d.s. group. A subject was considered to be compliant to be eligible for randomisation with PPI regimen if the subject had taken 80120% of the expected PPI tablets, had followed the dosing regimen correctly for 80% of the calculated time period and had been on the PPI for at least 18 days and less than 42 days at visit 4. Average total PPI dosing compliance rates ranged from 99.2% in
111

N. B. Vakil et al.
Table 2 | Baseline characteristics (ITT population)
Arbaclofen placarbil Variable Weekly frequency of heartburn events Weekly heartburn events related at least moderate in severity Weekly frequency of regurgitation events Weekly regurgitation events rated at least moderate in severity Weekly frequency of heartburn and regurgitation events Weekly heartburn and regurgitation events rated at least moderate in severity Placebo (n = 89) 11.4 (7.16) 8.3 (5.95) 20 mg q.d.s. (n = 94) 12.2 (7.06) 8.6 (6.12) 40 mg q.d.s. (n = 93) 12.3 (8.30) 8.8 (7.08) 20 mg b.d. (n = 90) 11.6 (7.98) 8.1 (6.71) 30 mg b.d. (n = 94) 13.7 (8.52) 9.8 (8.43) Total (n = 460) 12.3 (7.84) 8.7 (6.92)

6.8 (4.93) 4.8 (4.46)

7.5 (6.44) 5.6 (6.01)

8.4 (7.10) 5.9 (5.52)

7.4 (8.29) 5.7 (7.54)

9.1 (8.82) 7.0 (8.19)

7.9 (7.27) 5.8 (6.51)

18.2 (9.60) 13.1 (8.63)

19.7 (11.88) 14.2 (11.20)

20.7 (13.09) 14.7 (10.89)

19.1 (14.84) 13.8 (13.43)

22.9 (16.05) 16.8 (15.78)

20.1 (13.35) 14.5 (12.27)

b.d., twice a day; ITT, intent-to-treat; q.d.s., once a day; s.d., standard deviation. Values represent mean (s.d.).

LS mean (S.E.) % change from baseline

the 40-mg q.d.s. group to 101.6% in the 30-mg b.d. group. Average total PPI regimen compliance rates ranged from 93.2% in the placebo group to 98.6% in the 40-mg q.d.s. group.

20 30 40 50 60 70 80 90

Placebo Arbaclofen placarbil 20 mg q.d.s. Arbaclofen placarbil 40 mg q.d.s. Arbaclofen placarbil 20 mg b.d. Arbaclofen placarbil 30 mg b.d.

Primary analysis Figure 2 displays the LS mean percent change from baseline in heartburn events per week over 6 weeks of treatment by treatment group. Results showed that LS mean for placebo at week 6 was estimated as a 68% decrease from baseline in heartburn events. For the arbaclofen placarbil dose groups, the LS means ranged from 69.4% to 77.7% decreases from baseline. None of the comparisons of arbaclofen placarbil doses with placebo reached statistical signicance. The largest mean differences in percentage units compared with placebo in change from baseline to week 6 were observed for the 20-mg q.d.s. (mean, 9.0%; P = 0.094) and 30-mg b.d. (mean, 9.8%; P = 0.077) dosages. Separate analyses of the primary endpoint within low-dose and high-dose PPI subjects did not reveal statistically signicant differences between any active dose and placebo. Post hoc analysis results for primary efcacy endpoint. Figure 3 displays the LS mean percent change
112

Week 1 Week 2 Week 3 Week 4 Week 5 Week 6

Figure 2 | LS mean percent change from baseline in heartburn events per week over 6 weeks of treatment (FA population). b.d., twice a day; FA, full-analysis; LS, least squares; q.d.s., once a day; S.E., standard error.

from baseline in heartburn events with at least moderate severity per week over 6 weeks of treatment. In this analysis, the LS mean differences in units of percent change from baseline for the arbaclofen placarbil dose groups relative to placebo ranged from 12.6% to 19.9%. All arbaclofen placarbil treatment
Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

Randomised clinical trial: arbaclofen placarbil in GERD

Arbaclofen placarbil 40 mg 20 mg 30 mg q.d.s. b.d. b.d.

50

48

Placebo

20 mg q.d.s.

46

46

LS mean (S.E.) % change from baseline in heartburn events

40
10 20 30 40 50 60 70 80 90 100

34
Percent

33

34

35 31

30

27 24

20

10

P = 0.022 P = 0.045

P = 0.010

P = 0.002

n = 23 29 Placebo

30 42 20 mg q.d.s. 7

22 30 40 mg q.d.s.

30 39 20 mg b.d.

28 39 30 mg b.d. 8

Figure 3 | LS mean percent change from baseline in heartburn events with at least moderate severity per week of treatment (FA population). b.d., twice a day; FA, full-analysis; LS, least squares; q.d.s., once a day; S.E., standard error.

NNT

Arbaclofen placarbil 9

FA population

FA population excluding mild/very mild events

groups displayed statistically signicant superiority vs. placebo.

Figure 4 | Proportion of subjects with complete relief of heartburn. b.d., twice a day; FA, full-analysis; NNT, number needed to treat; q.d.s., once a day.

Secondary endpoints Analysis of change from baseline in regurgitation events per week over 6 weeks of treatment showed nonsignicant differences, with the 20-mg q.d.s. treatment group having the largest difference from placebo (mean, 0.9 events/week; P = 0.065) at week 6. Post hoc analysis of subjects with at least one regurgitation event of moderate severity at baseline showed nonsignicant differences, with the 20-mg q.d.s. treatment group showing the largest difference (mean, 1.0 events/week; P = 0.063) with regard to change from baseline in regurgitation events. Analysis of percent change from baseline in sum of heartburn and regurgitation events with at least moderate severity at week 6 showed greater percentage reductions for each arbaclofen placarbil dose group compared with placebo: The mean differences in percentage units vs. placebo were 13.6% in the 20-mg q.d.s. group (P = 0.033), 11.8% in the 40-mg q.d.s. group (P = 0.071), 15.1% in the 20-mg b.d. group (P = 0.022) and 19.0% in the 30-mg b.d. group (P = 0.004). Figure 4 displays the proportion of subjects with complete relief of heartburn in the FA population and the FA population excluding mild and very mild events. For the FA population, the proportion of subjects with complete relief of heartburn in the placebo group was 26.7%
Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

at week 6, while the proportion in the arbaclofen placarbil dose groups ranged from 24.2% to 34.9% at week 6. There were no statistically signicant differences between the arbaclofen placarbil dose groups and placebo for the proportion of subjects with complete relief of heartburn. For the FA population excluding mild and very mild events, the proportion of subjects with complete relief of heartburn was not statistically signicantly different between treatment groups. The largest differences were observed for the 20-mg q.d.s. group [odds ratio (OR) = 1.9; 95% CI for OR, 1.0, 3.6; P = 0.059] and the 30-mg b.d. group (OR = 1.8; 95% CI for OR, 1.0, 3.5; P = 0.065) relative to placebo. Figure 5 displays the proportion of subjects with complete relief of heartburn and regurgitation in the FA population and the FA population excluding mild and very mild events. For the FA population, the proportion of subjects with complete relief of heartburn and regurgitation per week over 6 weeks of treatment was not statistically signicantly different between the arbaclofen placarbil dose groups and placebo. The largest difference was observed for the 20-mg b.d. group compared with placebo at week 6 (OR = 1.9; 95% CI for OR, 0.9, 3.7; P = 0.084). For the FA population excluding mild and very mild events, the proportion of subjects with complete relief of heartburn and regurgitation, dened as no
113

N. B. Vakil et al.
50

47* 44 44*

40

31
Percent 30

33 29 29 28

21
20

20

10

n = 18 26 Placebo

26 42 20 mg q.d.s. 6

18 26 40 mg q.d.s.

28 37 20 mg b.d.

25 38 30 mg b.d.
7

NNT
FA population

Arbaclofen placarbil 8

FA population excluding mild/very mild events

Figure 5 | Proportion of subjects with complete relief of heartburn and regurgitation. b.d., twice a day; FA, fullanalysis; NNT, number needed to treat; q.d.s., once a day. *P < 0.05 vs. placebo (logistic regression).

more than mild episodes of heartburn and regurgitation at week 6, showed a statistically signicant result for the 20-mg q.d.s. group compared with placebo (OR = 2.2; 95% CI for OR, 1.1, 4.2; P = 0.018) and for the 30-mg b.d. group compared with placebo (OR = 2.1; 95% CI for OR, 1.1, 3.9; P = 0.026). Analysis of patient-rated Overall Treatment Effect score of 5 or greater (a good deal better, a great deal

better or a very great deal better) at week 4 and week 6 of treatment showed statistically signicant differences for the 20-mg q.d.s. group compared with placebo (OR = 2.0; 95% CI for OR, 1.0, 4.0; P = 0.038) and for the 40-mg q.d.s. group compared with placebo (OR = 2.4; 95% CI for OR, 1.2, 5.0; P = 0.013). Analysis of symptom severity across events of all severity showed that both the 20-mg b.d. group and the 30-mg b.d. group were statistically signicantly different compared with placebo in change from baseline at week 6 for weekly average severity of heartburn and regurgitation (mean, 0.5; P = 0.049 and mean, 0.5; P = 0.025 respectively). For weekly average severity of heartburn and regurgitation with at least moderate severity, there were statistically signicant differences for the 20-mg q.d.s., 20-mg b.d. and 30-mg b.d. groups compared with placebo in change from baseline at week 6 (mean, 0.6; P = 0.030, mean, 0.6; P = 0.040 and mean, 0.7; P = 0.010 respectively). In the FA population, the change from baseline in PAGI-QOL relationship scores showed a statistically signicant difference at week 6 for the 30-mg b.d. group compared with placebo (mean, 0.2; 95% CI for the difference, 0.3, 0.0; P = 0.037). Analysis of change from baseline in SF-12 Physical Component Summary scores showed no statistically signicant differences. Analyses of change from baseline in belching or burping events per week and throat burning or pain events per week showed no statistically signicant differences at week 6 comparing the arbaclofen placarbil doses with placebo.

Table 3 | TEAEs occurring in 5% of subjects in 1 treatment group in the safety population

Arbaclofen placarbil Preferred term (%) Any TEAE Somnolence Dizziness Nausea Fatigue Headache Diarrhoea Vomiting Dry mouth Nasopharyngitis Placebo (n = 87) 52.9% 2.3 3.4 5.7 6.9 1.1 1.1 1.1 0 3.4 20 mg q.d.s. (n = 93) 47.3% 7.5 6.5 7.5 4.3 1.1 3.2 2.2 2.2 5.4 40 mg q.d.s. (n = 93) 64.5% 15.1 17.2 12.9 7.5 10.8 1.1 1.1 1.1 2.2 20 mg b.d. (n = 90) 68.9% 22.2 5.6 3.3 5.6 4.4 2.2 3.3 1.1 1.1 30 mg b.d. (n = 92) 67.4% 18.5 20.7 21.7 14.1 12.0 7.6 6.5 5.4 0

b.d., twice a day; q.d.s., once a day; TEAEs, treatment-emergent adverse events.
114 Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

Randomised clinical trial: arbaclofen placarbil in GERD Safety and tolerability There were no clinically important changes in mean haematology, clinical chemistry and urinary parameter values from baseline to any visit time point. Individual and mean values of the safety ECG parameters showed no clinically meaningful trends over the evaluated time points. Treatment-emergent serious AEs occurred in ve subjects. None was assessed as related to arbaclofen placarbil. While receiving treatment with arbaclofen placarbil 40-mg q.d.s., one subject experienced possible food poisoning, one had cholecystitis, one suffered a myocardial infarction and one died due to arteriosclerotic cardiovascular disease. One subject on placebo suffered an ankle fracture. There was a dose-related increase for the most common AEs, which were somnolence, dizziness and nausea (see Table 3). Most reported AEs were mild or moderate in severity. Withdrawals due to AEs were 6% in subjects receiving placebo and 16% in subjects receiving arbaclofen placarbil. The most common reasons for withdrawal were nausea, somnolence, dizziness and headache, none of which exceeded 5%. DISCUSSION The results of our study show that arbaclofen placarbil was not superior to placebo in reducing the number of heartburn or regurgitation events of very mild or greater severity in a population of GERD patients who were symptomatic despite q.d.s. PPI therapy. Post hoc exploratory analyses suggest that there may be benet in reducing moderate to severe symptoms in these patients. Another factor that may have diminished sensitivity for detection of a treatment effect was the high percent change from baseline in heartburn events (68%) in the placebo group. Figure 2 shows that there was a decrease in heartburn events on all arbaclofen placarbil doses and placebo through week 3. After week 3, there was minimal further change for placebo, while further decreases were observed for active treatments. The initial decrease in symptoms could be due to some subjects having enrolled after only the minimum required 4 weeks on PPI, then experiencing further improvement in symptoms extending through the 4-week screening/baseline period and into the study treatment period. This could be remedied by requiring subjects to have been on PPI for 8 weeks at the time of screening. Of the 442 subjects in the full-analysis population, only 2 were on their PPI for less than 8 weeks, and 4 others were on their PPI for approximately 8 weeks at the time of screening. It is also
Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

possible that some subjects who had not been fully compliant with their PPI regimen prior to the study became more compliant during the study, and consequently received an incremental benet that extended beyond the time of initiating study treatment. This could be remedied by extending the baseline period prior to initiating study treatment, possibly reducing the placebo response that was observed in the present study. To enhance the generalisability of the study results, we elected to study patients on the PPI and dose they were taking at the time of study entry. That treatment had been determined clinically for each patient prior to study entry, and the study population was intended to be relevant to current clinical practice. It is possible that, for some patients, a different dose or a different PPI could have resulted in improved symptom response. The alternatives of restricting enrolment to a specic PPI or converting subjects to a single PPI regimen would have had their own operational and inferential disadvantages. For example, if patients were converted to a different dose of their PPI or to a different PPI than they were taking at study entry, an extended observation period would be required to establish that they had stable residual symptoms prior to starting the study medication. Such a study would leave open the question of relevance beyond the specic PPI dose regimen used. Patients with GERD who remain symptomatic despite therapy with PPIs are an important problem in clinical practice. These patients do not uniformly have reuxrelated symptoms. In one study of patients with GERD who were symptomatic on b.d. PPI therapy, 60% of patients had no correlation between symptoms and reux episodes.9 Symptomatic reux episodes were weakly acidic (pH 47) in the majority of cases, and reux episodes reaching the proximal oesophagus were more likely to be symptomatic.9 Patients whose symptoms do not demonstrate any correlation with reux episodes may have functional dyspepsia or visceral hypersensitivity as the mechanism for their symptoms. Baclofen has been shown to reduce the rate of reux episodes by 60% in healthy volunteers and to augment pressures in the lower oesophageal sphincter.10 Baclofen was effective in patients with or without hiatus hernia.11 In pH-impedance studies, baclofen decreased both acidic and weakly acidic reux episodes.12 In patients with GERD, baclofen decreased upright reux and improved symptoms of belching, regurgitation and the overall GERD symptom score.13 Arbaclofen placarbil, a transported pro-drug of the active R-isomer of baclofen, was developed to compensate for some of the limitations
115

N. B. Vakil et al.
of baclofen.14 It can be administered q.d.s. or b.d. rather than with every meal, and side effects such as somnolence have been less frequent than reported for baclofen.7 The development of new drugs to assist patients with symptoms despite PPI therapy is complicated by additional problems. The widespread availability of low-cost PPIs and the success of these agents in a large number of patients with GERD have limited the development of new drugs for individuals who fail to respond to PPIs. The widespread use of PPIs complicates trial design, because the investigational agent must be added to existing therapy, making changes in symptoms more difcult to quantify. Study design is further complicated by the inability to identify the physiological basis for symptoms without diagnostic tests, such as pH-impedance testing, which are not widely available in the community. Although the results of our study show that arbaclofen placarbil was not different from placebo in decreasing heartburn events across a severity spectrum from very mild to severe, post hoc analyses suggest that there may be benet in reducing moderate to severe symptoms. Studies with other reux inhibitors have been disappointing in terms of efcacy. In a clinical trial of symptomatic patients with GERD, lesogaberan had marginal efcacy compared with placebo and was associated with liver-chemistry abnormalities in a small number of patients,6 despite showing good efcacy in preclinical studies. Preliminary studies with another reux inhibitor, ADX10059, a metabotropic glutamate receptor 5 (mGluR5) negative allosteric modulator that inhibits transient sphincter relaxations, suggested efcacy.15 However, subsequent studies showed an unacceptable frequency of elevated transaminase values that led to a halt in the development of the drug.16 The failure of reux inhibitors to show a substantial benet in clinical trials in patients who are still symptomatic on PPIs remains unexplained. One potential explanation is that the patient selection may not be optimal in the trials carried out to date. Our study provides some insight in this regard. Limiting studies to patients with more severe symptoms may provide a more clinically meaningful target for therapy and allow separation of the effects from placebo. Another approach might be to limit the trials to patients with symptoms associated with reux episodes determined based on pH-impedance studies. This group has the highest likelihood of responding. The obvious limitations of such a trial design are that pH-impedance studies are not widely
116

available, and that the studies are difcult to perform in large multicentre trials. Another approach might be to focus on patients who report symptoms of regurgitation or belching.8 Both might be indicators of transient lower oesophageal sphincter relaxations as the potential mechanism for the persisting symptoms. Although our study did not demonstrate statistically signicant reductions in regurgitation or belching/burping events, the sensitivity to detect such effects was limited because subjects were not required to have those symptoms for study entry. Future studies are necessary to further characterise subgroups of this population based on regurgitation, belching and other symptoms. In conclusion, symptomatic patients with GERD on PPI therapy remain a difcult clinical problem for whom new therapies need to be developed. The results for arbaclofen placarbil in the present study may warrant further study in the relevant population. The analyses presented suggest that, in future studies, selecting patients with moderate or severe GERD symptoms may improve the efcacy of reux inhibitors, such as arbaclofen placarbil. New drugs continue to be evaluated for this group of patients. A selective 5HT4 agonist is currently under study,17 and other drugs are under development. Appropriate patient selection will be critical to their success.

AUTHORSHIP Guarantor of the article: N. B. Vakil. Author contributions: NBV drafted the manuscript and contributed to the study design, analysis and interpretation of the data. FJH and KCC co-drafted the manuscript and contributed to planning the study and collecting and interpreting data. All authors have approved the nal version of the manuscript, including the authorship list. ACKNOWLEDGEMENTS The authors thank Dr Daniel Bonzo, Yongmei Zhou and Amy Bian for statistical analyses and review. The following investigators, in addition to NBV, enrolled subjects into the study: Dr Michael Adams, Dr Armen Arslanian, Charles F. Barish, Dr Magued Beshay, Dr Bal Raj Bhandari, Dr David Bird, Dr Duane Bohman, Dr Donald Brandon, Dr Glenn Davis, Dr Donna De Santis, Dr David Dulitz, Dr Richard Dobrusin, Dr Michael Epstein, Dr David Eskreis, Dr Shaban Faruqui, Dr Bernard Feldman, Dr Nelson Ferreira, Dr Michael Gaspari, Dr David Gatof, Dr C. Allen Goetsch, Dr John Goff, Dr Paul Goldberg, Dr Michael J. Goldstein, Dr David Grant, Dr Robert Holbrok, Dr Scott Horn, Dr Robert Kaplan, Dr
Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

Randomised clinical trial: arbaclofen placarbil in GERD

Seymour Katz, Dr Whiteld Knapple, Dr Milton Koch, Dr Richard Krause, Dr Steven Lawrence, Dr David Limauro, Dr John Lowe, Dr Fadel Nammour, Dr Daniel Pambianco, Dr John Pappas, Dr Arthur L. Poch, Dr Edward Portnoy, Dr Theadore Ptak, Dr Andrea Ramsay, Dr Harvey Resnick, Dr Donato Ricci, Dr Dennis Riff, Dr Mark Ringold, Dr Joel Rubenstein, Dr Barry Schneider, Dr Michael Schwartz, Dr Gerald Shockey, Dr Sylvain Sidi, Dr Bob Souder, Dr Joseph Soufer, Dr Elhan Suley, Dr Martin Throne, Dr Vivaik Tyagi, Dr Lawrence Wruble and Dr Douglas Young. Declaration of personal interests: Dr Nimish B. Vakil has served as a consultant for AstraZeneca, Takeda, Orexo, Otsuka, Ironwood and XenoPort, and has stock ownership in Meridian and Orexo. Dr F. Jacob Huff has been an employee of, and is currently a consultant for, XenoPort, and has stock options/ownership in XenoPort. Dr Kenneth C. Cundy is an employee for XenoPort and has stock options/ownership in XenoPort. Declaration of funding interests: This project was funded by XenoPort, Inc. Employees of, and consultants for, XenoPort were involved in the study design, collection, analysis and interpretation of the data and in the writing of the manuscript.

REFERENCES
1. Jones R, Armstrong D, Malfertheiner P, Ducrott e P. Does the treatment of gastroesophageal reux disease (GERD) meet patients needs? A survey-based study. Curr Med Res Opin 2006; 22: 65762. 2. Keywood C, Wakeeld M, Tack J. A proof-of-concept study evaluating the effect of ADX10059, a metabotropic glutamate receptor-5 negative allosteric modulator, on acid exposure and symptoms in gastro-oesophageal reux disease. Gut 2009; 58: 11929. 3. Boeckxstaens GE, Beaumont H, Mertens V, et al. Effects of lesogaberan on reux and lower esophageal sphincter function in patients with gastroesophageal reux disease. Gastroenterology 2010; 139: 40917. 4. Gerson LB, Huff FJ, Hila A, et al. Arbaclofen placarbil decreases postprandial reux in patients with gastroesophageal reux disease. Am J Gastroenterol 2010; 105: 126675. 5. Kim GH. Arbaclofen placarbil decreases reux with good tolerability in patients with gastroesophageal reux disease. J Neurogastroenterol Motil 2010; 16: 4445. 6. Shaheen NJ, Denison H, Bj orck K, Karlsson M, Silberg DG. Efcacy and safety of lesogaberan in gastrooesophageal reux disease: a randomised controlled trial. Gut 2012; 61: 15019. 7. Vakil NB, Huff FJ, Bian A, Jones DS, Stamler D. Arbaclofen placarbil in GERD: a randomized, double-blind, placebo-controlled study. Am J Gastroenterol 2011; 106: 142738. 8. Kahrilas PJ, Boeckxstaens G. Failure of reux inhibitors in clinical trials: bad drugs or wrong patients? Gut 2012; 61: 15019. 9. Zerbib F, Duriez A, Roman S, Capdepont M, Mion F. Determinants of gastrooesophageal reux perception in patients with persistent symptoms despite proton pump inhibitors. Gut 2008; 57: 15660. 10. Lidums I, Lehmann A, Checklin H, Dent J, Holloway RH. Control of transient lower esophageal sphincter relaxations and reux by the GABAb agonist baclofen in normal subjects. Gastroenterology 2000; 118: 713. 11. Beaumont H, Boeckxstaens GE. Does the presence of a hiatal hernia affect the efcacy of the reux inhibitor baclofen during add-on therapy? Am J Gastroenterol 2009; 104: 176471. 12. Vela M, Tutuian R, Katz PO, Castell DO. Baclofen decreases acid and non-acid post-prandial gastro-oesophageal reux measured by combined multichannel intraluminal impedance and pH. Aliment Pharmacol Ther 2003; 17: 24351. 13. Cossentino MJ, Mann K, Armbruster SP, Lake JM, Maydonovitch C, Wong RK. Randomised clinical trial: the effect of baclofen in patients with gastrooesophageal reux - a randomised prospective study. Aliment Pharmacol Ther 2012; 35: 103644. Lal R, Sukbuntherng J, Tai EH, et al. Arbaclofen placarbil, a novel R-baclofen prodrug: improved absorption, distribution, metabolism, and elimination properties compared with R-baclofen. J Pharmacol Exp Ther 2009; 330: 91121. Zerbib F, Bruley des Varannes S, Roman S, et al.Randomised clinical trial: effects of monotherapy with ADX10059, a mGluR5 inhibitor, on symptoms and reux events in patients with gastrooesophageal reux disease. Aliment Pharmacol Ther 2011; 33: 91121. Development of ADX10059 ended for long-term use [press release]. Geneva, Switzerland: Addex Therapeutics; December 15, 2009. Available at: http:// www.addextherapeutics.com/investors/ press-releases/news-details/?tx_ttnews [tt_news]%20 = 88&cHash=d075 a6fa63acce925582f8f481fb8dc3. Accessed October 27, 2012. ClinicalTrials.gov. Selective 5-HT4 receptor agonist and proton pump inhibitor (PPI) in subjects with gastroesophageal reux disease (GERD). Available at: http://www.clinicaltrials. gov/ct2/show/NCT01472939?term= shire+and+GERD&rank=1. Accessed October 27, 2012.

14.

15.

16.

17.

Aliment Pharmacol Ther 2013; 38: 107-117 2013 John Wiley & Sons Ltd

117