In order to test the authors intriguing hypothesis that topical beta blockers improve acute migraine we will need a well-designed, statistically valid, large sample size, randomized, double masked, placebo controlled, clinical trial that would have defined primary outcome measures as well as strict inclusion and exclusion criteria.
Although a few prior anecdotal case reports have described successful topical beta blocker use for migraine, a recent larger series of 7 patients was published by Migliazzo and Hagan in the August 2014 issue of Missouri Medicine. In this report, an initial patient reported improvement of her migraine headache pain after topical beta blocker use for glaucoma. Seven similar patients were then described all of whom reported improved headache within minutes after topical treatment. Included patients completed questionnaire data and their case histories were described in more detail by the authors. The authors emphasized the superior cost, convenience, tolerability, and overall safety profile of topical beta blocker compared with other migraine therapies.
Dr. LeeCompanion editorials appeared in the same issue of the journal discussing the strengths and weaknesses of the report. Oral beta blockers have long been used for prophylaxis of migraine but have not been shown to be effective for acute abortive treatment in migraine. Dexter and Cady noted several weaknesses in the methodology of the study including: 1) small sample size; 2) an unknown non-participation rate for the questionnaires; 3) an older aged cohort (> 60 years) compared with a typically younger migraine population; and 4) the concomitant use of other migraine treatments.
Chung described the multiple potential mechanisms for beta blocker action in migraine including vasodilation, inhibition of sodium release, reduction or neuronal firing in the locus ceruleus, serotonin modulation, thalamic nuclei activation, and reduction of cortical spreading depression and highlighted some of the same limitations of the paper.
From my perspective, the work of Migliazzo and Hagan is interesting and exciting but as with all anecdotal clinical observations it is useful for hypothesis generating but not necessarily for hypothesis testing. One way of considering the strength of any observation is the potential use of specific causation criteria. Although initially developed not for causation per se but for epidemiologic study, the Hill criteria are a potentially useful framework for this discussion. The list of the Hill criteria are as follows:
According to the Hill criteria, in order to support the hypothesis that topical beta blockers can be used to treat acute migraine we would like to see a very strong effect in a larger sample size; a consistent and reproducible pain reduction with a close temporal relationship between drug and effect that is coherent both with the known pharmacokinetics of the drug and the time to onset for treatment; a biologic dose response curve for treatment effect; a biologically plausible mechanism of action; coherence of effect for drugs within the same class of agent; and an experimental design that can account for the effects of chance alone and confounding variables including the placebo effect. Ideally there would be a statistically appropriate sample size to answer the question given a set of specific assumptions about error tolerance, masking (of patients and observers), standardization of the primary outcome measure, appropriate randomization and allocation of study participants prospectively, and of course most importantly a placebo control. Despite the limitations of their work I commend the authors for making a potentially important clinical observation and many a therapy in medicine has been discovered in a similar serendipitous fashion.
The bottom line is that in order to test the authors intriguing hypothesis that topical beta blockers improve acute migraine we will need a well-designed, statistically valid, large sample size, randomized, double masked, placebo controlled, clinical trial that would have defined primary outcome measures as well as strict inclusion and exclusion criteria.
Andrew G. Lee, MD, is Professor of Ophthalmology, Neurology, and Neurosurgery, Weill Cornell Medical College and Chair of Ophthalmology at Houston Methodist Hospital, Houston, Texas. Clinical professor of ophthalmology at the University of Texas Medical Branch (UTMB), Galveston, TX and the UT MD Anderson Cancer Center, Houston, Texas and Adjunct Professor of Ophthalmology at Baylor College of Medicine, Houston, TX.