Abstract
Treatment for chronic tension-type headache (CTTH) is unsatisfactory. Our aim was to investigate the efficacy of the N-methyl D-aspartate (NMDA) antagonist memantine in the prophylactic treatment of CTTH. We included 40 patients in a randomized, double-blind, placebo-controlled, crossover trial. Memantine 20–40 mg/day or placebo was each given for 10 weeks separated by a 2-week wash-out period; 29 patients completed the study. The primary efficacy variable, area-under-the-headache curve (duration X intensity), did not differ between memantine (1352 ± 927) and placebo (1449 ± 976; P = 0.10). Headache intensity in both sexes was significantly lower on a 0–10 verbal rating scale with memantine (3.8) than with placebo (4.1; P = 0.03). In women, area-under-the-headache curve was significantly lower with memantine (1343 ± 919) than with placebo (1555 ± 1019; P = 0.01). The most common side-effects were dizziness and nausea. In conclusion, although no statistically significant effect was seen in the primary end-point, some beneficial effects of memantine were observed in women. Memantine was shown to reduce pain intensity in CTTH patients, albeit to a limited extent. Future NMDA antagonists with higher efficacy could be of major interest as regards the pathophysiology and future treatment of CTTH and other chronic pain disorders.
Keywords
Introduction
Tension-type headache is the most common type of headache. To most people with the episodic form it poses no health problem, but to the 3% of the adult population (1) who experience daily or nearly daily headache, the disorder leads to high socioeconomic costs and a considerable loss of quality of life. Tension-type headache on ≥ 15 days/month is known as chronic tension-type headache (CTTH) (2); often it is resistant to treatment. To date, the only prophylactic treatment to show efficacy in several studies is the tricyclic antidepressant amitriptyline. However, the effect is often inadequate and/or hampered by side-effects, primarily fatigue, weight gain and dry mouth (3). A more efficient treatment with fewer side-effects is therefore needed.
Central sensitization is pivotal in the pathophysiology of CTTH (4–6). Drugs that counteract central sensitization are therefore the most obvious candidates for new prophylactics for CTTH. The N-methyl D-aspartate (NMDA) receptor is involved in both the initiation and maintenance of central sensitization, and it has been shown that NMDA antagonists can prevent the induction of central sensitization and reverse existing central sensitization in both animal and human models of pain (7–9). Although memantine has been shown effective in several animal pain models, to our knowledge no human, blinded, randomized studies have shown that memantine has an analgesic effect. Memantine is a non-competitive NMDA receptor antagonist approved for the treatment of Alzheimer's disease; it is reported as being safe and has limited side-effects (10, 11). It is therefore logical to test its effect in CTTH. Such a study has been suggested (12), but memantine has not previously been tested against placebo in the treatment of headache.
Our aim was to test if memantine could be effective as a prophylactic treatment of CTTH. We used a randomized, double-blind, placebo-controlled, crossover trial.
Subjects and methods
Patients
Ninety patients were screened for inclusion (Fig. 1). Patients were recruited from advertisements in five different local newspapers and from the out-patient headache clinic at a tertiary headache centre (Danish Headache Centre, Glostrup, Denmark). Patients were enrolled by the authors and recruited between April 2006 and February 2007; the trial was completed in August 2007. Patients underwent a general examination, including a 12-channel ECG and a neurological examination, and they completed a diagnostic headache diary (13) for a 4-week run-in period. Blood pressure and pulse were measured at first and last visit. A blood sample was taken at the first visit and haemoglobin, leucocytes, platelets, liver and kidney parameters were checked.
Flow of participants through the study.
The inclusion criteria were a diagnosis of CTTH (2) and age between 18 and 65 years. Women of childbearing age were requested to use adequate contraceptive measures throughout the study. Exclusion criteria were a history of more than one migraine attack per month during the previous year; serious somatic or psychiatric diseases, including depression [Hamilton Depression Score > 17 (14); overuse of analgesics (according to the International Classification of Headache Disorders, 2nd edn (2))]; and intake of prophylactic headache medication. All patients gave written informed consent to participate in the study, which was approved by the regional ethics committee and National Board of Health and conducted in accordance with the Declaration of Helsinki. The study was monitored by the Copenhagen University Hospital Good Clinical Practice (GCP) unit and met the standards of the GCP guidelines (15). The study was registered at the public database at http://www.clinicaltrials.gov and at the Danish Data Protection Agency. Patients were informed that the study included placebo tablets, but received no further information about the study design.
Study design and medication
The study was designed as a randomized, double-blind, placebo-controlled, crossover trial. After a 4-week run-in period, patients were randomly allocated in blocks of 5–15 to start in one of two arms (Fig. 2). The first arm started with a 10-week period with active treatment, followed by a 2-week wash-out period and then a 10-week placebo period. The second arm started with a 10-week placebo period, followed by 2 weeks’ wash-out and then 10 weeks with active treatment. In the first 4 weeks of the active period the daily dose of memantine increased by 10 mg each week, reaching a maximum intake of 40 mg/day for the remaining time of the active treatment period. Each treatment week had three containers; every day the patients would take one tablet from each of the first two containers and two tablets from the last container. If side-effects occurred, patients were asked to reduce the tablet intake to three or two tablets per day from a specified container. Patients received four tablets daily throughout the 22 weeks of treatment, provided the dose was not reduced. Patients were requested to take the tablets at bedtime. The allocation sequence was generated by the Central Pharmacy in Copenhagen County where the tablets were packed in sealed, numbered containers. The packed containers were numbered 1–40 and patients received the corresponding number in the order in which they were enrolled. Memantine (Ebixa®; Lundbeck, Copenhagen, Denmark) and matching placebo tablets were provided by Lundbeck. Throughout the study while filling out the headache diary, patients were blinded to the memantine/placebo content of the tablets. The study was not unblinded until completion of the typing in of the completed headache diary in a database.
Study design of the 22-week, double-blind, placebo-controlled, crossover trial. Each patient was randomly allocated to one of the two possible treatment sequences. Clinical visits are indicated with ▴ and telephone contacts indicated with ∧.
Recording of efficacy variables
Throughout the study the patients kept a headache diary with recordings of headache intensity, headache duration, intake of analgesics and side-effects. Intensity was recorded on an 11-point verbal rating scale (VRS) (0–10), where 0 indicated the headache-free condition, 5 a moderate headache and 10 the worst headache imaginable. In addition, patients recorded side-effects; localization and quality of the headache; whether the headache was aggravated by physical activity; and the presence or absence of nausea, photophobia and phonophobia.
Clinical visits
The patients were seen after run-in, before they were included, and total tenderness score (TTS) (16), Hamilton Depression Score, weight and intake of medication were recorded. Follow-up visits were performed in weeks 10–12 and in week 23, i.e. after termination of the treatment and placebo periods. At these visits the headache diary was checked, side-effects reported by the patients were recorded, and compliance was tested by checking the returned packages and by counting any returned tablets. In addition, medication supplies were issued at the visit in weeks 10–12. Before the run-in and between clinical visits, patients were telephoned by one of the investigators to go through the headache diary, to record side-effects and to assure compliance (Fig. 2). The completed headache diary was sent to the authors before each telephone conversation.
Evaluation of efficacy and side-effects
The efficacy variables were determined in advance. Area-under-the-headache curve (AUC) was calculated as the sum of the daily recordings of headache duration multiplied by headache intensity. The primary efficacy variable was AUC in the last 4 weeks in the treatment period compared with the last 4 weeks in the placebo period. To evaluate the time course of headache, the improvement in AUC was calculated for each treatment week as mean AUC for the 4 weeks of run-in (baseline) minus AUC for each of the 10 weeks of treatment. Secondary efficacy variables were comparison between treatment and placebo in the last 4 weeks of treatment regarding: (i) headache frequency; (ii) headache duration; (iii) mean headache intensity per headache day; and (iv) days with analgesic intake.
Patients were instructed to record side-effects in the headache diary and to telephone the investigator if they were moderate or severe. During every telephone conversation and clinical visit, patients were asked if side-effects had occurred. Their answer was checked to see if it was in accordance with the received headache diary. If an event occurred, the intensity and duration of the potential side effect were recorded and it was also noted if the event was of probable relation to memantine. Furthermore, it was recorded if the event was unexpected and/or serious, and any action, e.g. dose reduction, was noted.
Statistics
Sample size calculations were based on an expected standard deviation (
The Kolmogorov–Smirnov test was used to determine whether the data were normally distributed. Data that were normally distributed were presented as means with
Results
Study population
Forty patients were enrolled in the study (Fig. 1). An additional three patients were recruited but withdrew their consent before starting treatment; nine of the 40 patients reported coexisting infrequent migraine (< 1 day/month), two of whom dropped out of the study. Most patients had previously tried numerous treatments for tension-type headache and had had frequent headaches for many years. Baseline characteristics are presented in Table 1. In the final 4 weeks of the active treatment period there were a total of two migraine attacks and in the placebo period there were four attacks, i.e. no difference between the groups (P = 0.50, McNemar test). There was no change in Hamilton (P = 0.38) or TTS (P = 0.66) between the two periods (Wilcoxon).
Characteristics of patients in study
Data for the 40 patients enrolled and the 29 patients who completed the study are presented, taken from history and from headache diaries completed before enrolment in the trial. Mean and standard deviation or median values with quartiles in parentheses are given depending on whether data are normal distributed or not. VRS, verbal rating scale (0–10).
Treatment efficacy: primary variable
The mean AUC in the active treatment was 1352 (± 927) and in the placebo period 1449 (± 976) (P = 0.10, CI −444, 39; independent t-test). There was no period effect (P = 0.68, CI −192, 291) or treatment–period interaction (P = 0.54, CI −512, 958), both tested with independent t-test. A post hoc analysis was done in women, and AUC was significantly lower during the active treatment period (1343 ± 919) than during placebo (1555 ± 1019) (P = 0.01, CI −751, −106; independent t-test). In women, curves representing weekly AUC separated after 2 weeks on full dose and were significantly different in weeks 6, 8 and 10. In men there was no difference (Fig. 3).
Improvement in area-under-the-headache curve (AUC) values in men and women with chronic tension-type headache during 10 weeks of treatment with memantine and placebo. Values are calculated as mean [AUC for the 4 weeks of run-in (baseline)]—(AUC for each of the 10 weeks of treatment). ∗P < 0.05 (paired sample t-test). ▴, memantine; ▪, placebo.
Treatment efficacy: secondary variables
Headache intensity was significantly lower during treatment with memantine 3.8 (2.9–5.6) than during placebo 4.1 (3.3–6.2) (P = 0.03, Mann–Whitney). There were no differences in headache duration, headache frequency or analgesic intake between active and placebo treatments (Table 2). Post hoc analysis showed diminished headache intensity in women during treatment with memantine 3.9 (2.3–5.7) compared with placebo 4.5 (3.3–6.5) (P = 0.005, Mann–Whitney), as well as a reduction of intake of analgesics in the memantine period compared with the placebo period (5.1 ± 3.9 vs. 6.4 ± 4.4, P = 0.006; independent sample t-test). There were no other significant differences in secondary variables in women and none in men.
Treatment efficacy: secondary variables
P-values are given from Mann–Whitney tests. Data for analgesics are normally distributed and therefore analysed with an independent sample t-test. n = 29. Median values with quartiles in parentheses or mean and standard deviation are presented. VRS, verbal rating scale (0–10).
Side-effects and drop-outs
Of the 29 patients who completed the study, 13 were reduced in memantine dose from 40 to 20 mg. One patient increased the dose again to 40 mg. Of the 13 patients who were reduced in dose, four were men and nine were women. One of the patients was reduced in dose in the placebo period, the rest in the memantine period. Five patients dropped out because of dizziness, all in the active period. Three of these patients dropped out in the second of the 10 treatment weeks, one patient in the fifth and one in the ninth treatment week. Three patients dropped out because of lack of effect, two in the active and one in the placebo period. Two patients were excluded because they were non-compliant in taking the medication. One patient was excluded because of suspicion of depression in the first half of the study, in the placebo period (Fig. 1).
There was significantly more dizziness (P < 0.01) and nausea/vomiting (P = 0.04) (McNemar) during active treatment (Table 3). Mild infections were common in both periods; none was assessed to be related to memantine. In Table 3, ‘Various’ covers events maximally reported by two patients. In both periods the following symptoms were reported: palpitations, restlessness, muscle tenderness/muscle pain. In the active treatment period the following events were reported by no more than one patient: paraesthesia in an arm, black-out, dry mouth, muscle cramp in a leg, increased appetite, poor sleep and mood swings. In the placebo period the following events were reported by no more than one patient: debut of gout, depression, distended abdomen, abdominal pain, car crash, nocturnal cough, discovery of a benign lump in breast.
Side-effects
All side-effects/events recorded during the study (n = 40). The numbers in parentheses specify how many of the events were noted to be of probable relation to memantine.
Discussion
The primary overall finding of this study was that memantine did not have a substantial effect in the prophylactic treatment of CTTH. There was a trend towards efficacy (P = 0.10, CI −444, 39) and the CI shows that we cannot exclude a 31% reduction (−444/1449 × 100%) in headache AUC compared with placebo. However, a 3% worsening of headache cannot be excluded either (39/1449 × 100%). Most of the effect seen in the primary variable was due to the reduction in intensity. This reduction in intensity is statistically significant both for all patients and for women, but it is a modest reduction of < 1 on the VRS 0–10 scale. This modest reduction in intensity makes it questionable whether the reduction is of clinical significance. This study largely conformed to the International Headache Society guidelines (18), except for treatment periods being shorter and the primary efficacy variable being AUC. Shortening of the study was chosen to reduce the drop-out rate and increase compliance. We believe compliance was fairly good, except for the two patients who were excluded for lack of compliance. AUC was chosen as a primary end-point as it seems to be more sensitive than headache frequency for trials in CTTH (19–21).
Memantine was used in doses of 20–40 mg/day. The approved dose in treating Alzheimer's disease is 20 mg/day with an escalating dose of 5 mg/week. The higher dose was chosen so that an effect would not be missed due to insufficient dose. The reason for a more rapid up-titration was to shorten the study. The side-effects constitute a risk of unblinding and thereby bias, especially in a crossover study such as this. However, the patients were not informed of the exact design of the study, and we do not believe that unblinding was a problem. Despite the side-effects, memantine was safe and there were no serious adverse events. A clinically relevant effect may have been overlooked in this study due to inadequate dose, but it is highly unlikely considering the substantial proportion of the participants who experienced side-effects. It is possible that a statistically significant effect on the primary variable could have been shown if fewer patients had dropped out.
Although there was no clinically relevant effect, there was a clear trend towards an effect in the primary efficacy parameter, and there was a significant effect on headache intensity in the whole group and a significant effect on the primary efficacy parameter in women. Together, this strongly indicates that memantine does have an effect on the headache pain perceived in CTTH, which is of major interest with regard to the pathophysiology and possibilities for future treatment modalities in CTTH. In a new study of memantine, it would be advisable to test the effect only in women with CTTH with a lower dose and slower up-titration.
Gender plays an important role in chronic pain as women are more prone to chronic pain than men (22). We found an effect in the primary effect parameter in women when each sex was analysed separately, but no effect was found in men. In agreement with this effect of a NMDA antagonist primarily in women, it has recently been shown that the activity of NMDA-receptors in the dorsal root ganglia neurons of female rats was 2.8-fold higher than in male rats. Furthermore, the activity of the NMDA-receptor was found to increase more in females rats when adding 17-β-estradiol than in male rats (23). However, any connection between these findings in animals and our findings in this study is speculative.
The crossover design in a clinical trial can be valuable, especially in chronic non-fluctuating conditions because of increased power. However, the potential disadvantage of a crossover trial is increased sensitivity to drop-outs and an increased risk of drop-outs due to longer duration of the trial (24). In this study 11 patients dropped out despite our efforts to reduce the number, e.g. with the option of dose reduction and frequent contacts. Similar drop-out ratios have been seen in related studies (25, 26).
To our knowledge, this is the first randomized, placebo-controlled, double-blind trial of memantine in headache prophylaxis; furthermore, memantine has not been previously tested in chronic myofascial pain disorders. Recently, an open study has suggested an effect of a low dose (10 mg) of memantine in the prophylactic treatment of migraine (27). Few of our patients had migraine attacks in the study period; therefore, this study does not add to the evidence for or against efficacy of memantine in the prophylaxis of migraine.
Studies in animals have shown an antinociceptive effect of memantine. Neuropathic pain models in rats (28) and monkeys (29, 30) have indicated this, as have behavioural pain models in rats (31, 32) and mice (33). In open human studies memantine has had an effect on pain, but to date no long-term effect has been shown in blinded, placebo-controlled, randomized studies. An open study found that memantine reduced pain and improved motor functions in complex regional pain syndrome (34). In randomized controlled trials (RCTs) memantine has been found to be ineffective in treating chronic phantom limb pain (35, 36); postherpetic neuralgia; diabetic, postamputation and HIV-associated neuropathy (37–40). In a RCT, memantine in combination with brachial plexus blockade reduced phantom limb pain at 4 weeks and 6 months but not after 12 months in patients with traumatic amputation (41).
A continued search for better treatment of CTTH is needed. Our results on headache intensity suggest that NMDA-receptors are involved in the pain of patients with CTTH, especially in women. Thus, future improvement in the treatment of CTTH could involve the NMDA-receptor system. In this search a better understanding of the pathophysiology is warranted, and a big step towards this would be an animal model. Important progress is being made in this field (42).
In conclusion, although treatment did not reach the primary efficacy end-point for the study population as a whole, post hoc subgroup analysis indicates that the primary end-point was met for women. We have shown that memantine shows a modest reduction of headache intensity, and future NMDA antagonists with higher efficacy could be of major interest regarding the pathophysiology and future treatment of CTTH and other chronic pain disorders.
Footnotes
Acknowledgements
We thank Mrs Hanne Andresen for skilful technical assistance and Lundbeck, Denmark for supporting the study with a grant. Lundbeck did not take the initiative in this study, and had no say in its design or carrying out, calculation of data or interpretation of results.