Abstract
Introduction
The phosphodiesterase-3-inhibitor cilostazol induces migraine-like attacks in patients with migraine without aura, and may be used as a pharmacological trigger in human experimental models of migraine. However, the reproducibility of cilostazol-induced migraine-like attacks has never been investigated.
Methods
We performed a post-hoc analysis of clinical data from two brain-imaging studies including subjects who had received cilostazol 200 mg orally. Only subjects who developed migraine-like attacks on study day 1 were included on study day 2. After cilostazol ingestion, subjects and the investigator recorded headache intensity and characteristics once every hour on a purpose-developed questionnaire. Primary end-points included incidence and time to onset of migraine-like attacks between two separate study days.
Results
Thirty-four subjects completed both experimental days and were included in this study. Thirty-four out of 34 subjects (100%) developed migraine-like attacks after cilostazol ingestion on both study days 1 and 2. Time to onset of migraine was five hours (range 1–8 hours) on study day 1 and four hours (range 1–8 hours) on study day 2, p = 0.16. We found no difference in median peak headache score, median time to peak headache score, or median time to intake of rescue medication between study days 1 and 2.
Conclusion
A second-time administration of cilostazol reproduces migraine-like attacks in all subjects who report an attack after their first cilostazol induction. There was no difference in time to migraine onset between separate inductions. Experimental migraine provocation using cilostazol is a highly efficient and useful approach for studying the ictal phase of migraine without aura.
Introduction
Much research effort is being invested in developing optimized preventive treatment for the debilitating and prevalent condition of migraine (1). To invent tailored treatment options that specifically target the molecular pathways of migraine, the underlying pathophysiological mechanisms need to be better understood. Human experimental models are crucial in this context, as they add to our fundamental understanding of migraine-specific mechanisms while providing a suitable model for drug testing. Throughout the last decades, a variety of pharmacological models have been applied to investigate the molecular pathways involved in the initiation of a migraine attack (2), more recently also with the phosphodiesterase-3-inhibitor, cilostazol (3). Cilostazol has been shown to induce headache in healthy volunteers (4,5) with a high degree of reproducibility between separate study days (5). In subjects with migraine without aura, cilostazol induces migraine-like attacks (3) with an aptitude superior to other pharmacological triggers including calcitonin gene-related peptide (CGRP), pituitary adenylate cyclase-activating peptide-38 (PACAP38), and glyceryl trinitrate (6,7). However, whether cilostazol-induced migraine is reproducible within subjects across time has thus far never been investigated. Knowledge of retest reproducibility would solidify the role of cilostazol as a potent migraine inducer for experimental models, paving the way to delineate biomarkers of the migraine condition as well as to test anti-migraine medication. In this study, we examined the ability of cilostazol to induce and reproduce migraine in subjects with migraine without aura on two separate study days.
Methods
Participants and design
Subject data collected in this study derived from two brain-imaging studies (studies A and B, Figure 1) where subjects with migraine without aura received cilostazol on two separate days as part of experimental models of migraine (clinicaltrials.gov ID NCT02549898 (MRI study, results not yet published) and NCT02309606 (PET study, results not yet published)). The aim of study A was to investigate perivascular inflammation of the cerebral arteries with MRI in migraine without aura. The aim of study B was to investigate serotonin receptor binding in migraine without aura using PET scans during the ictal and interictal phase of migraine. For both studies A and B, cilostazol was solely used to induce migraine-like attacks in order to perform imaging during the ictal phase of migraine.
Study enrolment.
Inclusion criteria for study A were: a) women aged 18–50 years, b) diagnosis of migraine without aura (8), c) between one and eight migraine days per month, and d) unilateral migraine in ≥ 70% of attacks. The triptan response was not recorded for subjects in study A.
Inclusion criteria for study B were: a) age 18–65 years, b) diagnosis of migraine without aura (8), c) at least one migraine attack every other month but less than five migraine days per month, and d) reported successful treatment of migraine attacks with sumatriptan.
Exclusion criteria for both studies included any other type of headache (apart from episodic tension-type headache ≤ 5 days per month), any previous serious somatic or psychiatric condition, or daily intake of medication (apart from oral contraceptives), including migraine prophylactic medication. Study A had additional exclusion criteria (e.g. MRI contraindications) not relevant for the present study.
The regional committee on Health Research Ethics, Capital Region, approved the two original studies (protocol numbers H-1500-5669 and H-6-2014-057), which were conducted in accordance with the Declaration of Helsinki of 1964, with later revisions. All participants provided written informed consent prior to any study-specific procedures. All participants underwent a full medical examination including ECG. Pregnancy tests were performed on all female participants.
Experimental protocol
Subjects ingested 200 mg cilostazol (Pletal® Otsuka Pharmaceutical Europe Ltd.) orally with 200 ml of milk (9) on two separate study days. All subjects had been migraine-free for at least 48 hours prior to cilostazol ingestion. The subjects were informed that cilostazol might induce headache and possibly migraine in some individuals.
Headache recording
Headache intensity was recorded at baseline and then once every hour on a purpose-developed questionnaire. In study A, headache characteristics were recorded by the subject on study day 1 and by the physician on study day 2, apart from the last few hours where some subjects had been discharged and continued the recording on their own. In study B, headache characteristics were recorded by the subject on study day 1 and for the first few hours on study day 2 prior to undergoing PET scan, and then by the physician for the remaining hours.
Headache was rated on a numerical rating scale (NRS) of 0–10, where 0 represented no headache, 1 represented a very mild headache (including a sensation of pressing or throbbing), 5 a moderate headache, and 10 the worst possible headache (10). Intensity, headache localization, characteristics, accompanying symptoms, and premonitory symptoms (yawning, unusual fatigue, neck stiffness, mood swings) were also recorded.
On study day 1, participants were discharged immediately after cilostazol ingestion. After discharge, participants were carefully instructed to perform recordings of headache and adverse events on a purpose-developed questionnaire until 16 hours post-administration or until they had gone to bed.
Only subjects who had developed migraine-like attacks on study day 1 were included on study day 2. All subjects ingested cilostazol a second time and then underwent MRI respective PET scans after migraine onset. Headache and adverse events were recorded on a purpose-developed questionnaire for time points similar to day 1.
Rescue medication
For subjects participating in experimental provocation studies, the option of acute treatment of the induced attack is mandatory (11). For this reason, subjects in studies A and B were offered rescue medication after cilostazol ingestion, in adherence with protocol-specific time points.
On study day 1, subjects from study A were advised that they could take their usual abortive medication when they reached a headache intensity of 4 on the NRS, or at any desired time if they were not able to wait until NRS 4. Subjects from study B were instructed to take a sumatriptan 100 mg tablet when they reached a headache intensity of 4 on the NRS.
On study day 2, all subjects received either subcutaneous injection of 6 mg sumatriptan or a tablet of 25 mg promethazine and/or 10 mg metoclopramide. Subjects who received promethazine and metoclopramide were in an experimental arm of study A where traditional analgesics and triptans could not be administered due to protocol specifications.
Headache classification
Any experimentally-induced migraine attack is, by definition, a secondary headache, and thus cannot strictly fulfill IHS criteria for migraine without aura (8). A few considerations are important when defining criteria for pharmacologically-induced migraine attacks in human provocation models (3,12). Firstly, the majority of patients report that the induced attacks mimic their spontaneous migraine attacks (3,6). Secondly, it is well known that many spontaneous migraine attacks develop in a matter of hours, and in the early stage of the attack phenomenologically only fulfill the criteria for tension-type headache before the headache worsens, becomes unilateral, and exhibits the associated symptoms required for migraine. Thirdly, most patients are able to predict the development of migraine in the early stage of the attack, and cannot be denied rescue medication in a research study. Hence, induced migraine attacks are often treated before all criteria are fulfilled.
For these reasons, we applied the following criteria to define a headache as migraine-like (12):
Headache fulfilling criteria C and D for migraine without aura according to the IHS criteria. C, Headache has at least two of the following characteristics: Unilateral location, pulsating quality, moderate to severe pain intensity, or aggravation by physical activity. D, At least one of the following accompanying symptoms: Nausea and/or vomiting, or photophobia and phonophobia. Headache described as mimicking the participant's usual migraine attack and treated with acute migraine rescue medication.
Statistical methods
Frequencies and percentages were calculated for categorical variables (incidence of migraine-like attack) and medians with range were calculated for continuous variables (time to headache onset, time to migraine-like attack onset, peak headache intensity score, time to peak headache intensity score, and time to rescue medication).
The primary endpoints were: a) incidence of migraine-like attack, b) median time to onset of migraine-like attacks, and c) median time to onset of headache on study days 1 and 2. Secondary endpoints were: a) median peak headache intensity score, b) median time to peak headache intensity score, and c) median time to rescue medication on study days 1 and 2. Due to differences in study design, all secondary endpoints were calculated separately for the two studies.
Differences in time to headache and migraine onset, peak headache intensity and time to medication intake were tested using Wilcoxon signed-rank test. A non-parametric test was applied, since data were not normally distributed as assessed by visual inspection of histograms.
All p-values were two-sided and considered statistically significant if < 0.05. All analyses were performed using GraphPad Prism 7 for Mac OS X.
Results
Demographic and clinical characteristics of 34 migraine patients originating from two imaging studies, study A (n = 26) and study B (n = 8).
Migraine-like attack induction
On study day 1, 48 out of 56 subjects developed a migraine-like attack after ingestion of cilostazol, corresponding to a migraine-induction effect of 86%. On study day 2, 34 of the total 34 participating subjects developed a migraine-like attack, corresponding to 100% reproducibility of migraine-induction after cilostazol on the two separate study days.
Timing of migraine-like attack onset
Time from cilostazol ingestion to onset of migraine was five hours (range 1–8 hours) on study day 1 and four hours (range 1–8 hours) on study day 2, p = 0.16. Time to onset of headache was two hours (range 1–8 hours) on study day 1 and two hours (range 1–6 hours) on study day 2, p = 1.000. Due to missing hourly data points for a subset of subjects, the above two analyses were performed on the 20 subjects with hourly headache registrations.
Temporal characteristics of migraine-like attacks induced by cilostazol on two separate study days (n = 20*).
Due to missing data of hourly headache registrations, n total was reduced to 20 for these analyses.
Median headache score. In study A (top panel), median time (range) to intake of rescue medication was 7 h (4–9 h) on day 1. On day 2, it was 5 h (4–6 h) for the sumatriptan group and 5 h (4–15 h) for the other* treatment group. In study B (bottom panel), median time (range) to intake of rescue medication was 6 h (3–9 h) on day 1 and 6 h (4–9 h) on day 2.
Characteristics of cilostazol-induced migraine-like attacks versus spontaneous attacks
Clinical characteristics of headache and associated symptoms in 34 migraine patients after cilostazol on two study days and during spontaneous migraine attacks.
Spon: Spontaneous attack.
Lateralization/intensity/quality (throb = throbbing, pres = pressing)/aggravation. For the two study days, these are reported at the timepoint where the subject first fulfilled the migraine-like headache criteria described in “methods”.
Nausea/photophobia/phonophobia. For the two study days, these are reported for the timepoint where the subject first fulfilled the migraine-like headache criteria described in “methods”.
Migraine-like attacks are classified according to criteria described in “methods”.
Exact time of migraine onset is not registered, VRS is recorded at 4 hours after cilostazol ingestion when it was first registered that the subject fulfilled migraine-like attack criteria.
Exact time of migraine onset is not registered, however the subject did fulfill migraine-like attack criteria at 4 hours after cilostazol ingestion.
Other treatment was promethazine and/or metoclopramide.
N/A: Not applicable due to missing data.
Out of the total 68 induced attacks (34 subjects with two attacks on two different study days), 60 subjects (88%) expressed that the migraine induced by cilostazol mimicked their spontaneous attacks. Twenty-five out of 34 subjects (74%) reported the same laterality of their induced attack as of their spontaneous attacks on at least one of the two study days. Twenty-eight out of 34 subjects (82%) reported unilateral pain localization of their induced attacks on at least one of the two study days.
Adverse events were recorded for both studies A and B and for both induction days, with unusual tiredness, flushing and stiff neck reported as the most common adverse events.
Discussion
The main finding of the present study is that cilostazol-induced migraine-like attacks had 100% retest reproducibility between two separate study days. There was no difference in time to migraine onset between the two days of induction.
Reported adverse events after cilostazol administration on study day 1 and 2, for study A and B.
Our results also confirm the pronounced migraine-inducing effect of cilostazol (3), as 86% of subjects developed migraine-like attacks in our study also. Consequently, we can concur that cilostazol has a superior success rate in inducing migraine-like attacks compared to other compounds, with reported induction potentials of 58–73% for PACAP38 (11,14), 67% for CGRP (6), and 80% for glyceryl trinitrate (7). Based on this great induction potential and the superb reproducibility of attacks between separate study days, the place for cilostazol in experimental human headache models has been further solidified.
Cilostazol is a selective inhibitor of phosphodiesterase 3, a degrading enzyme of the second messenger cyclic AMP. Thus, cilostazol ingestion leads to accumulation of the cyclic AMP within smooth muscle cells of cerebral arteries (14). Previous studies have shown that cyclic AMP plays a role in experimentally-induced headache (4), suggestively by sensitizing trigeminal neurons (15) and modulating the nociceptive input in pain processing brain regions. Namely, CGRP, a neuropeptide known to induce migraine, exerts its effects by binding to CGRP 1 and 2 transmembrane receptors, leading to increased intracellular cAMP (16,17).
The disparity between cilostazol's migraine-inducing effect compared to other inducing compounds could indicate that cilostazol activates a cascade further downstream than PACAP38 and CGRP (3), as the latter two activate transmembrane receptors in cerebral vascular cells, leading to the formation of cAMP (16,17), whereas cilostazol causes the accumulation of cAMP by inhibiting phosphodiesterase 3.
Optimal timing and execution of experimental research is crucial, perhaps particularly so in studies with experimentally-induced migraine, a painful state that invariably burdens the participating subjects. In studies A and B, we designed the experimental setup so that all subjects were screened with cilostazol prior to actually undergoing the imaging protocols, to exclude any subjects who did not develop migraine-like attacks. As our results have confirmed the high success rate of cilostazol in inducing migraine-like attacks and demonstrated solid reproducibility between separate inductions, one could argue that screening for migraine induction potential of cilostazol may be superfluous in future studies. However, screening may still be a valid precursor, particularly in studies that include imaging modalities, to optimize the experimental procedures and the use of temporal and financial resources. For those subjects who did not respond to cilostazol on the screening day and who were thus excluded from studies A and B, it would have been highly interesting to attempt a rechallenge of migraine induction with cilostazol. Knowledge of the cilostazol effect not only in first-time responders but also in first-time non-responders would further elucidate the efficacy of cilostazol in human experimental headache models, and would clarify whether the reproducibility of a negative response is as solid as that of a positive response.
As we evaluate the efficacy and reproducibility potential of cilostazol, it is intriguing to consider whether some subjects may be predisposed to migraine induction over others. In this context, it has been suggested that migraine patients exhibit innate variations in migraine threshold (18) so that the likelihood of developing a migraine attack varies over time. A recent functional neuroimaging study reported that the hypothalamus, with its intricate brainstem connections, is the migraine generator, oscillating in activity depending on whether the patient is in the postictal, interictal or ictal phase (19). If migraine attacks indeed demonstrate cyclic periodicity, the question is whether pharmacologically-induced migraine-like attacks can only be triggered when the subject is naturally moving towards their next spontaneous attack, thus being more susceptible to inducing substances. The optimal way to answer this question would be to collect headache diary registrations from all subjects, covering the last month prior to and the first month after cilostazol induction. This would allow us to comprehend each subject's individual attack pattern across time, and possibly grasp whether any subjects may have been in a pre-ictal state on study days 1 and 2. However, with our sample size, the dual induction in all subjects with medians of 33.5 and 315 days between the two study days, and our finding of 100% reproducibility between the two inductions, it would be highly unlikely that all subjects were naturally moving towards their forthcoming spontaneous migraine attack and were only susceptible to attack provocation in and of this fact. Also interesting to note is that the subjects in our analyses ranged in monthly migraine attack frequency from median two to median four days.
It is unclear whether the induction potential and reproducibility of cilostazol is independent of baseline headache frequency. Further investigation of this aspect could be achieved by including subjects with only a few attacks per year. Planning cilostazol-induction in the aftermath of any such sporadic attack would allow us to consider the effect of cilostazol as truly unrelated to innate cyclic variations.
To the best of our knowledge, no previous studies have tested the reproducibility of migraine inducing substances in subjects with migraine without aura. Considering the importance of experimental human headache models, the retest quality of headache inducing agents is clearly underexposed. For future optimization of a similar study, the study design could include a three-arm crossover set-up, with subjects randomized to cilostazol:cilostazol:placebo. This would allow a blinded evaluation of cilostazol reproducibility in the presence of placebo. Also, we do recognize that an absolute comparison of cilostazol to previous pharmacological migraine inducers based solely on the present study is tentative, as our study has no placebo group and the investigators were unblinded.
Also important to consider when scrutinizing the efficacy of migraine-inducing substances is whether differences in dosage and formulation within the same drug may impact the degree of headache induction. It would be interesting to tailor future studies to test different dosages in parallel, to see whether this would affect the induction of migraine-like attacks.
With this study, we have demonstrated that a second-time administration of cilostazol reproduces migraine-like attacks in all subjects who report an attack after their first cilostazol induction. We found no difference in median time to migraine onset between the two inductions, which facilitates the planning of experimental studies across time. We have also solidified the role of cilostazol as a powerful migraine inducer, and anticipate that this new knowledge will optimize the planning and execution of future human experimental studies of migraine.
Clinical implications
A second-time administration of cilostazol reproduces migraine-like attacks in all subjects who report an attack after their first cilostazol induction. There is no difference in median time to migraine onset between the separate cilostazol inductions, which facilitates the design and scheduling of experimental migraine studies. Experimental migraine-provocation using cilostazol is a highly efficient and useful approach for studying the ictal phase of migraine without aura.
Footnotes
Acknowledgements
The authors would like to thank lab technicians Winnie Grønkjær and Lene Elkjær for their assistance.
Declaration of conflicting interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: S Khan has acted as invited speaker for Novartis during the conduct of this study. M Ashina is a consultant or scientific advisor for Allergan, Amgen, Alder, ATI, Eli Lilly, Novartis and Teva, primary investigator for Amgen 20120178 (Phase 2), 20120295 (Phase 2), 20130255 (OLE), 20120297 (Phase 3) and GM-11 gamma-Core-R trials, and reports grants from Lundbeck Foundation (R155-2014-171) and Novo Nordisk Foundation (NNF11OC101433) during the conduct of the study.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Lundbeck Foundation (R155-2014-171 and R180-2014-3398), Novo Nordisk Foundation (NNF11OC101433 and NNF15OC0017132), and the European Union’s Seventh Framework programme (FP/-EUROHEADPAIN-no.602633).