Botulinum toxin in headache treatment: Finally a promising path?

Botulinum toxin as a treatment for headaches has created many controversies during the past decade. Widely adopted in most North American headache centres, there is little enthusiasm for its use in Europe. Claimed to be efficacious in many open label studies, it failed to reach superiority over placebo in randomised, controlled trials (RCTs) in chronic tension-type headache, episodic migraine or chronic daily headache which led, in 2006, to the provocative ‘end of the road’ editorial by Evers and Olesen (1). In a review for a national medical journal, we suggested that efficacy data from subgroups of patients with so-called chronic daily headache might indicate the ‘beginning of a new path’ (2). The two PREEMPT trials (3,4) provide evidence that such a ‘path’ may actually exist and leads towards patients suffering from chronic migraine, a subgroup for which a diagnostic consensus is available in the ICHD-II appendix (CM-R A 1.5.1) (5). Because of their positive outcome, these trials have already raised great interest in the headache community, and they will be used as evidence for the efficacy and, thus utilisation, of Onabotulinum toxin A, the toxin studied, in chronic migraine patients. Is this justified? The following considerations may help you to make up your mind.

Do the studies provide scientific evidence of efficacy?

Both studies undoubtedly do support efficacy in the patients studied and have demonstrated undisputable strengths. They were well-designed, performed by experienced investigators and included large numbers of patients with a long blinded follow-up of 24 weeks and a subsequent open label period of 32 weeks. Onabotulinum toxin A was statistically superior to placebo in a range of outcome measures, including quality of life and disability scales. ‘Headache episodes’, the notable exception to this in PREEMPT 1, is clearly not an adequate outcome measure, nor clinically relevant in patients who have almost daily headaches and use symptomatic pain-relieving therapies.

Is the study population representative of our patients?

One major concern is that both trials recruited a majority of patients (up to 65%) fulfilling criteria for medication overuse headache (MOH-R A 8.2) (5). Although this reflects real-world life, it means that many patients may have been suffering from a secondary headache disorder and not, as stated in titles and methods section, from CM per se. Detoxification suffices in many patients to convert the chronic headache pattern into an episodic one and/or ineffective preventives to effective drugs. This may explain part of the beneficial effect, admittedly both in study and placebo groups. Both groups had indeed an average decrease of ±10 acute medication intakes at week 24, but there was a significantly greater decrease of triptan intake in Onabotulinum toxin A-treated patients. It would have been informative to know whether the treatment response differed between patients with and without medication overuse.

Along the same line of patient selection, it is surprising that up to 40% of enrolled patients never received a preventive treatment before, the more so as mean duration of CM in both trials was 20 years and mean age of participants ±40 years! Although the reliability of the historical data on migraine pattern and previous treatments may be questioned, knowing if prophylaxis-naive patients respond differently, information which was not provided, is of practical interest.

Are there methodological weaknesses?

As in other botulinum toxin studies, one may be concerned about the effective blinding of treatments. The authors reject this possibility because of the absence of a nocebo effect in the placebo group. However, such a nocebo effect can only occur if patients are aware or informed about the motor/autonomic effects of the botulinum toxin injections, which is not mentioned in the articles.

To evaluate the level of unblinding, subjects and investigators could have been questioned as to what treatment group (active or placebo) the subject was assigned to during the study. This was apparently not done in the PREEMPT trials.

The primary end-point in PREEMPT 2 was changed from headache episodes to headache days, based on the negative results for the former in PREEMPT 1. From a strict methodological point of view, this change, clearly done after most if not all, patients were included in PREEMPT 2, is open to criticism, but it is unlikely to be crucial for the overall treatment effect.

Are the observed effects clinically relevant?

A frequent question in RCTs concerns the clinical significance of statistical differences. There is no consensus on the minimal change in an outcome measure required to be considered clinically relevant. Because of the well-known high placebo response in headache trials using injections of drugs, in particular, of botulinum toxin, the therapeutic gain of Onabotulinum toxin A over placebo is not impressive in numerical terms: absolute gain of 6.7% and 11% for headache days and 7.9% and 11.3% for migraine days, respectively in PREEMPT 1 and 2. Looking at the absolute mean change in headache days, there is little doubt that a substantial number of migraine patients switched from a chronic to an episodic pattern. According to data from the American Migraine Prevalence and Prevention Study, conversion from CM to EM occurs over a 3-year period in 26% of patients, whereas CM persists unchanged in only 22% (6). Unfortunately, the percentage of patients who converted and the difference between treatment arms are not given in the PREEMPT articles, nor are the respective percentages of responders with a 50% or 25% reduction in headache days, so that a comparison with the natural history of the disorder is not possible. These missing data, as those on the proportion of patients who would choose to continue the treatment, would have been useful to estimate further the clinical relevance of the results.

To what can these data be compared?

The only other RCT performed in CM has studied topiramate (7) and it is somewhat surprising that the PREEMPT authors, several of whom also participated in the topiramate trial, do not compare the results in the discussion. Although preventive treatment-naive patients were excluded in the topiramate trial contrary to PREEMPT 1 and 2, the effect size in the three studies seems to be similar, e.g. 10.2% absolute gain over placebo for migraine days in the topiramate trial compared to 11.3% in PREEMPT 2.

Onabotulinum toxin A was directly compared to topiramate in a small randomized double-blind single-centre trial (8). No statistical difference was found between the two treatment groups, but the results are not based on an intent-to-treat analysis. Adverse events, however, were more numerous and severe for topiramate.

Are the PREEMPT 1 and 2 results convincing enough to justify our present or change our future medical practice?

This is the only question that really matters to patients and doctors. My answer is ‘Yes’ with some reservations. Onabotulinum toxin A was proven in these trials to be moderately superior to placebo, and probably to natural outcome, in a population of chronic migraine patients of whom the majority was overusing acute medication and a substantial proportion had never received preventive drugs. This is good news as CM is a most disabling disorder with a 1–2% prevalence in the general population in which even a small effect or an effect in a small percentage of patients can be of medical value. A major advantage of Onabotulinum toxin A compared to other drug treatments is its excellent tolerability. Because the PREEMPT trials leave unsolved a number of clinically relevant questions, Onabotulinum toxin A cannot become yet the pre-emptor of CM treatment. It seems wise at this stage to restrict its use to specialised headache centres where it can become part of the multidisciplinary armamentarium recommended in chronic headache patients. Phase IV studies should be able to provide answers to crucial open questions like: ‘How to identify responders?’ or ‘Is the treatment cost-effective?’ For translational benefits, they should be paralleled by more research on the mode of action of Onabotulinum toxin A in chronic migraine, which still remains a mystery. Thus this ‘path’ is promising, but needs to be more clearly delineated and clarified in clinical practice, before anyone can take it on a routine basis.