Abstract
Key design issues in migraine with aura trials, as identified by Hauge et al.
Experienced headache specialists have always known that for some migraine sufferers it is aura, rather than headache, that is disabling. Over time, scientific knowledge of the mechanisms underlying aura has increased. Recent epidemiological research provides reason to suspect that repeated attacks of aura – even in the absence of headache – may be undesirable rather than harmless. Animal models now allow the identification of candidate drugs that block cortical spreading depression and might be expected to work against aura in humans. All of these things have increased interest in treatments that target aura, not just headache, and the tonabersat trials highlight aspects of migraine treatment trial design that may need modification.
Subject selection, treatment timing and administration, and data collection
Until recently, acute and preventive trials of migraine have principally been conducted in mixed populations of migraine subjects; only about a quarter of participants in most migraine trials have migraine with aura. With a few exceptions, the possible differential effects of treatment in this subgroup of participants have not routinely been evaluated. Hauge et al. (1) point out that, in treatment trials designed to evaluate drug effects on migraine with aura, it is desirable to select a study population that experiences a large number of auras, so that treatment effects can be more easily detected. They suggest that subjects should report having had at least one aura a month over the 3 months prior to study entry; because auras may not occur particularly frequently, they do not think a baseline data collection period is practical. They see no need to exclude subjects who also have attacks of migraine without aura, or even chronic migraine; however, for acute treatment trials, only attacks with migraine should be treated.
For trials of acute aura treatment, they sensibly suggest subjects should have auras that last long enough so that treatment effects can be detected; this, they say, means they should have auras lasting at least 20 min. The researchers also discuss the tricky matter of medication timing, which among other things may depend upon whether it is the aura or subsequent headache that is of interest. Timing is everything when the treatment window is very narrow, leading to the reasonable proposal that rapid-acting drug formulations should be used when the outcome of interest is pre-emption or shortening of aura.
Perhaps the single most useful aspect of the Hauge et al. (1) paper is the example they provide of the diary used to diagnose and classify episodes of aura (and subsequent headache). The way in which study subjects are queried about possible episodes of aura is of critical importance. Imprecise questions, such as asking whether subjects have experienced ‘visual disturbances’, for example, lead to overestimates of aura. This is because vague disturbances such as visual blurring are common in migraine but do not constitute the focal neurological deficit that is a requirement for the diagnosis of aura (3). Yet visual blurring and other minor symptoms will be captured as aura by imprecisely worded questions. Although this seems a minor point, such overestimation of aura has turned out to be a serious problem in many of the large cohort studies used to examine the possible association between migraine types and outcomes such as ischaemic stroke (4,5). Most such studies used inexact questions to classify subjects as having aura: this led to subjects without true aura being included in the migraine with aura groups. If, as is now strongly suspected, the increased risk of ischaemic stroke is confined to migraine with aura, this dilution of the aura group with subjects who do not have aura has probably biased results of many studies towards the null. This, in turn, may have led to systematic underestimation of the magnitude of the association between migraine with aura and ischaemic stroke.
Outcome measures
Hauge and colleagues suggest that researchers must be clear in migraine with aura treatment trials whether the aim is to affect aura or headache. If the former, then they suggest that the main outcome parameter should be aura duration in acute treatment trials and number of auras in prophylactic trials. If the latter, then traditional outcomes remain appropriate (such as pain relief or freedom from pain at 2 h after treatment). Cross-over designs should be used when feasible because they can improve study power.
Based on their experience, the authors recommend that episodes of neurological dysfunction that meet all but one of the International Classification of Headache Disorders (ICHD) (6) criteria for migraine should, nonetheless, be counted as ‘probable’ attacks of aura. Similarly, they have come to doubt the wisdom of attempting to distinguish among the various types of headache that may follow aura. They suggest that any headache following an aura event might be considered a migraine, since the ‘pathophysiological mechanism triggering headache is the same’.
Rethinking the matter of drug approvals
One suggestion made by the authors is likely to be controversial. They contend that treatment trials in mixed populations cannot provide evidence that treatments are effective in those who have migraine with aura. Thus, they recommend that regulatory authorities, such as the United States Food and Drug Administration, should ‘not allow a drug to be listed for migraine with aura without studies done in that population’ since it is ‘impossible in mixed trials to conclude a drug works for both types’.
The way forward
In well-understood diseases with an established tradition of therapeutic research, there is usually considerable agreement about the way in which randomised clinical trials of treatment should be designed and conducted. Consensus develops over time about the outcomes and end-points that should be measured and that matter to clinicians and patients. When new treatments emerge, they are tested in trials of similar design. For disorders whose pathophysiology is less well understood, however, novel treatment targets may emerge that require refinement of conventional trial methods. This is the case for migraine. With periodic modifications and additions, the International Headache Society clinical trial guidelines for migraine have performed well for almost two decades (7). With adjustments similar to those suggested by Hauge and colleagues, they will continue to serve us admirably in the future.