Abstract
We report the result of a double-blind placebo controlled study of sodium valproate (SV) (1000-2000 mg/day) in the prophylaxis of cluster headache CH. Episodic and chronic CH were defined according to the International Headache Society classification. Ninety- six patients were included, 50 in the SV group and 46 in the placebo group. After a 7-day run-in period, patients were treated for 2 weeks. The primary efficacy criterion was the percentage of patients successfully improved, i.e having an at least 50% reduction in the average number of attacks per week between the run-in period and the last week of treatment. Whatever the type of CH, there was no difference between the two groups: 50% of subjects in the SV group and 62% in the placebo group were successfully improved (P = 0.23). This high success rate observed in the placebo group, which is likely to be due to the spontaneous remission of the episode, does not allow us to draw any valid conclusion with regard to the true efficacy of SV in the prophylaxis of CH.
Introduction
Cluster headache (CH) remains a difficult to treat condition despite the efficacy of oxygen inhalation and of subcutaneous sumatriptan for the acute treatment of attacks and that of verapamil and lithium prophylactically (1). In 1989, Hering and Kuritzky (2) performed an open study of sodium valproate (SV) (600–2000 mg daily) in the prevention of CH attacks. Out of 15 patients, 11 were remarkably improved, with a complete disappearance of attacks in nine and a marked reduction in two. We report herein the results of a double-blind placebo controlled study of SV in the prophylaxis of CH attacks.
Methods
Sixteen European centres participated in this double blind placebo-controlled parallel group study (Appendix 1).
Patients suffered episodic or chronic CH as defined in the International Headache Society classification (3), with one to three attacks per day. Males aged 18–70 years and post-menopausal women were included. All patients gave their informed consent and the study was approved for all the French centres by the ethics committee of Saint Antoine hospital in Paris and by their local ethics committee for their other centres.
Exclusion criteria included drug or alcohol abuse, liver or kidney disease, psychiatric disorders, intake of antidepressant or neuroleptics, and contraindications to SV, including abnormal hepatic transaminases. No prophylactic treatment should have been used in the 2 weeks prior to the first visit or in the 4 preceding weeks in the case of lithium prophylaxis.
Treatment consisted of SV 500 mg release tablets or placebo tablets identical in shape and colour. Patients were blindly assigned to treatment according to a randomization list balanced by blocks of four that had been predefined by the biometric Sanofi research department.
After the first visit there was a run-in period of 7 days during which the patient documented his/her attacks in a diary. If the number of attacks was between 7 and 21 the patient was randomized into SV or placebo groups and was treated for 2 weeks (15 days) with assessments at the end of each week. The dose of SV was 1–2 g per day. From day 1 to day 3, patients received two tablets (1 g of SV or placebo) in the evening. From day 4 to day 8, according to the clinical status, one tablet could be added in the morning and, from day 9 on, a fourth tablet was eventually added so that the dose remained unchanged from day 9 to day 15.
The only authorized treatments of attacks included subcutaneous sumatriptan at a maximum of 6 mg b-i-d, with at least 1 h between the two injections and oxygen inhalation at a flow rate of 7 L/min.
The primary efficacy criterion was the percentage of patients successfully improved. Success was defined as at least a 50% reduction in the average number of attacks per week between the run-in week and the last week of treatment (i.e. the second week). Secondary efficacy parameters included: (i) the number of patients with over 75% reduction in the mean number of attacks between run-in and last week; (ii) percentage of attack-free days; (iii) the average pain intensity (assessed by a 100-mm visual analogue scale); (iv) the mean duration of attacks; (v) the consumption of acute treatment by sumatriptan or oxygen; and (vi) the investigator's global assessment of improvement (based on a seven-step scale going from very much improved to very much worse).
Safety was assessed by recording all spontaneously reported adverse events, by clinical examination, standard ECG, and standard laboratory parameters including transaminases.
A sample size of 120 was calculated based on an expected success rate ≤25% in the placebo group and ≥60% in the SV group, with α = 5%, β = 10%, with a two-sided test and taking into account a 15% rate of non-evaluable patients.
Results were analysed on an intention to treat (ITT) basis. For the primary efficacy criterion, the percentage of subjects with a success was compared between treatments using the Mantel-Haenszel chi-square test. A P < 0.05 was taken as significant.
For the secondary efficacy criteria: (i) the average pain intensity per week and maximum duration of attacks per week were compared between treatments using the Wilcoxon rank sum test; and (ii) the pain in the previous week, as measured by the visual analogue scale, was summarized and the change from baseline was compared between treatments using a two-sample (unpaired) t-test.
Patients
Ninety-six patients were included, 50 in the SV group and 46 in the placebo group. (Inclusion of patients was stopped prematurely because of a too-slow recruitment, mostly due to the fact that a prophylactic treatment had already been used, which was an exclusion criteria.) One patient from the placebo group was not assessed for efficacy while under treatment. He withdrew from the study because of adverse events after 3 days on study drug. He was excluded from the ITT analysis but he was maintained in the safety analysis.
Six patients discontinued the study prematurely (SV: 4, P: 2). Two, one from each group, had adverse events; one had severe somnolence (SV) and one was hospitalized for an unrelated condition (placebo). Three patients from the SV group stopped the study treatment due to lack of efficacy and one from the placebo group did not attend visit 4 but came back 1 month later and returned his diary.
The two treatment groups were similar as far as baseline characteristics are concerned (Table 1). The male to female ratio was 7.6 : 1. There was a majority (77%) of patients with episodic CH. There was no difference in the mean inclusion time between the two groups of episodic CH: 46.8 days±35.4 with SV and 48.4 ± 38.8 with placebo. The mean duration of previous episodes was, however, shorter in the placebo group (62.4), than in the SV group (78.3). This difference is not statistically significant.
Clinical characteristics of the 95 cluster headache patients included in the study. Values are mean ±
Results
As regards the primary efficacy, no difference was found between the two treatment groups on the primary efficacy criterion: a success (more than 50% reduction in the average number of attacks between the run-in week and the last, i.e. second, week of treatment) was found in 50% of subjects in the SV group and in 62% of the placebo group (P = 0.23).
There was no difference either in any of the secondary parameters. The number of patients with a more than 75% reduction in the mean number of attacks between run-in and last week was 14 (28%) in the SV group and 12 (26.7%) in the placebo group. The percentage of patients much or very much improved according to the investigator's global assessment at the end of the last week was similar in the two treatment groups (SV: 65%, placebo: 64%). Results concerning other secondary efficacy criteria are indicated in Table 2: no difference is seen as regards the mean percentage of attack-free days, the mean pain intensity, the mean duration of attacks and the percentage of patients using rescue medication.
Secondary efficacy criteria
The success rate was lower (but not significantly so) in chronic than in episodic cluster: the mean number of attacks in the SV group was 10.3/week during the run-in period and 7.8 during the last week and, for placebo, respectively, 13 and 8/week. There was no significant difference between the two treatment groups but numbers are very small (SV: 11, placebo: 6).
Side-effects were more frequent with SV (40%) than with placebo (28%). The most frequent were nausea and/or vomiting (SV: 12%, placebo: 4%) and somnolence (SV: 12% and placebo: 2%).
Discussion
The most striking finding of the present study is the high success rate observed in the placebo group: 62% of patients in the placebo group had a more than 50% decrease in the number of attacks between the run-in week and the last treatment week and 26.7% had a more than 75% decrease. Rather than a true placebo effect, the most likely explanation is the spontaneous decrease in the frequency of attacks towards the end of the cluster period. Indeed patients with episodic cluster were entered late in their episode: 46.8 days in the SV group and 48.4 in the placebo group. Unfortunately there was an imbalance between the two groups as regards the mean duration of the previous cluster period, which was higher in the SV group (78.3 days) than in placebo group (62.4). It is therefore likely that a spontaneous remission in the placebo group has obscured a true treatment effect of SV. It is thus impossible to draw any conclusion as regards the efficacy of SV in episodic CH. No obvious efficacy was found either in the chronic group, but numbers are too small to draw any firm conclusion.
An unexpected placebo effect was also found by Steiner et al. (4) in a study of CH attacks prevention by lithium. They used a far more stringent efficacy criterion: the percentage of patients whose attacks ceased within 1 week. They expected a zero placebo effect and a 50% lithium effect, but they observed attacks cessation in 14% of patients in the placebo group and in 15% in the lithium group. As the mean duration of the cluster period before randomization was short (respectively 20.7 and 23.6 days), this probably corresponds, in contrast to our study, to a true placebo effect and not to the spontaneous remission of the cluster period.
These two negative studies thus raise important methodological issues as regards drug trials in the prophylaxis of CH attacks. First, episodic and chronic cluster require different trials; secondly, there is a placebo effect that is more important than previously thought even in chronic cluster; thirdly, in trials devoted to episodic cluster, a run-in period should be best avoided because of the short duration of episodes, which is not compatible with both a run-in period and a sufficiently long period of treatment; fourthly, attention should be paid to the duration of previous episodes in order to avoid the imbalance between the two groups that occurred in the present study and that possibly obscured a treatment effect.
Footnotes
Acknowledgements
The authors are grateful to all the investigators involved in the present trial and to Sanofi research department for meeting the costs of this trial and agreeing to publish the results of this trial.
Appendix 1
Investigators: (number of patients included)
(1) Doctor MAF Bomhof, Breda, The Netherlands (25)
(2) Doctor Valade, Paris, France (16)
(3) Professor Bousser, Paris, France (13)
(4) Professor Warter, Strasbourg, France (8)
(5) Doctor Ten Houten, Alkmaar, The Netherlands (6)
(6) Doctor Koehler, Heerlen, The Netherlands (5)
(7) Doctor Kok, Den Helder, The Netherlands (4)
(8) Doctor Chedru, Meaux, France (4)
(9) Doctor Caekebeke, Aalst, Belgium (3)
(10) Doctor Siebenga, Gouda, The Netherlands (3)
(11) Doctor Verhagen, Nijmegen, The Netherlands (3)
(12) Professor Weber, Nancy, France (2)
(13) Doctor Denef, Tienen, Belgium (1)
(14) Doctor Maertens De Noordhout, Liege, Belgium (1)
(15) Doctor Van Der Zwan, Amersfoort, The Netherlands (1)
(16) Doctor Breuer, GelDoctorop, The Netherlands (1)