Abstract
We have previously shown that nitric oxide (NO) and cyclic guanosine monophosphate (GMP) may cause headache and migraine. However, not all findings in previous studies can be explained by an activation of the NO-cGMP pathway. Calcitonin gene-related peptide (CGRP) causes headache and migraine in migraine patients, but CGRP receptor activation causes an increase in cyclic adenosine monophosphate (cAMP). In order to investigate the role of cAMP in vascular headache pathogenesis, we studied the effect of cilostazol, an inhibitor of cAMP degradation, in our human experimental headache model. Twelve healthy volunteers were included in a double-blind, randomized, crossover study. Placebo or cilostazol (200 mg p.o.) was administered on two separate study days. Headache was scored on a verbal rating scale (0-10) and mechanical pain thresholds were measured with von Frey hairs. The median peak headache score 0-16 h postdose was 0 (range 0-2) after placebo and 3.5 (range 0-7) after cilostazol (P = 0.003). The median headache curve peaked at 6-9 h postdose. The headaches induced were usually bilateral and pulsating. Nausea occurred in two volunteers, photo- and phonophobia were not seen. Two volunteers had a headache that fulfilled International Headache Society criteria for migraine without aura after cilostazol. No change in mechanical pain thresholds in the forehead was seen (P = 0.25). The headache after cilostazol was equal to or more severe than headache induced by glyceryl trinitrate in previous experiments. The present study thus indicates that increased levels of cAMP may play a role in headache and migraine pathogenesis.
Introduction
Human experimental migraine models have proven useful in suggesting novel antimigraine pharmacological targets (1, 2) and in the analysis of signalling pathways involved in the induction of the headache and migraine response (3, 4). The nitric oxide (NO) hypothesis of migraine is well established (3) and studies in healthy volunteers and migraine patients with the highly selective inhibitor of phosphodiesterase type 5 (PDE5) sildenafil (4, 5), have demonstrated that cyclic guanosine monophosphate (cGMP) is a likely mediator of headache responses elicited by NO.
Calcitonin gene-related peptide (CGRP) is an endogenous neuropeptide that appears to play a causative role in the migraine attack (2) and the CGRP receptor antagonist BIBN4096BS has been proven efficient in aborting spontaneous migraine attacks (6). Vascular and other effects of CGRP are mediated by CGRP1 or CGRP2 receptors, although controversy remains. Most actions of thesereceptors, but not all, appear to be mediated by an increase in cyclic adenosine monophosphate (cAMP) (7, 8).
We hypothesized that cAMP, like cGMP, may be a mediator of experimental headache. Therefore, we studied the headache-generating effect of cilostazol, an inhibitor of PDE3 (9), one of the two quantitatively most important cAMP degrading enzymes in cerebral arteries (10). The cerebrovascular data from this study have been reported previously (11).
Materials and methods
Design and participants
We used a double-blind, randomized, balanced crossover trial in which opaque gelatine capsules containing placebo (lactose, potato starch, magnesium stearate and talc) or cilostazol tablets 2 × 100 mg were administered by mouth on two study days separated by at least 1 week.
Sixteen healthy participants were included in the study after a full medical history and physical examination. The participants had no history of migraine or other primary headache, except episodic tension-type headache occurring less than once a week. None of the participants reported having first-degree relatives with migraine. The participants had no other neurological or cardiovascular disorders and had no daily intake of any medication except oral contraceptives. Four volunteers did not complete the study: one withdrew after experiencing severe migraine-like headache after receiving cilostazol on day 1 and another due to circumstances unrelated to the study. One patient was excluded because of vomiting a few minutes after the intake of placebo and one after cilostazol due to the development of a migraine-like headache and junctional escape nodal beats on the ECG with no clinical manifestations. Twelve participants completed the study, six males and six females aged 26 ± 2 years and weighing 54–100 kg (mean 73 kg). All participants gave their informed consent before inclusion. The study was approved by the ethics committee of Copenhagen County (KA 00111s) and the Danish Medicines Agency (2612-1435) and was conducted according to the Helsinki II declaration.
Experimental protocol
The participants arrived at the clinic at 08.00 h, free from headache. The participants fasted for at least 8 h before the start of the study, except for water (12), and no medication, except oral contraceptives, were allowed during the same period. The participants received either cilostazol 200 mg or placebo by mouth. After the intake of cilostazol or placebo, the participants were allowed the intake of one bread roll with no spread (12).
Headache recordings
Headache was rated verbally on a 0–10 scale every 15 min throughout the study. A score of 1 represented a mild headache (including a sensation of pressing or throbbing in the head not associated with pain), 5 a headache of moderate intensity and 10 the worst headache imaginable by the subject (13). Headache intensity, localization and characteristics, and other adverse events were recorded during the in-hospital period (Table 1). The participants were discharged 4 h postdose and continued recording adverse events every hour until 16 h postdose by a self-administered questionnaire.
Headache response and adverse events 0–4 and 4–16 h postdose
McNemar's test.
Mechanical pain threshold
The pain threshold for mechanical stimulation was determined using a set of calibrated von Frey hairs (VFH) (Somedic, Hoerby, Sweeden) at time points t 10 (baseline), t 90, t 150 and t 210. Each VFH was assigned a number representing a logarithmic increase in pressure. Two regions, both approximately 1 cm2, were selected and marked, one just above the right supercilium in the frontal region innervated by the first branch of the trigeminal nerve, and the other at the volar side of the left forearm (centrally from side to side and from the elbow to the wrist). Since subject cooperation is pivotal, a test run was performed before obtaining baseline. To allow for interindividual variation in skin sensitivity, a suitable range of VFH was selected during the test round, usually hair nos. 11–17. The participants were blinded for the VFH number and were stimulated four times with a frequency of 1–2 Hz in each region, with each VFH applied in random order. They were instructed to report how many of the stimulations felt painful like a pin-prick (0–4). This series was repeated three times at each time point; the summary measure was the median of the lowest VFH number reported to cause pain in at least two of four stimulations (14). There were no significant changes in room air humidity or temperature on the two study days.
Statistical methods
Pressure pain thresholds and headache scores are presented as medians (range).
The difference between placebo and cilostazol in headache response (peak and area under the headache score curve) between treatments was analysed using the Wilcoxon signed rank sum test. Changes in pressure pain thresholds over time were analysed for each treatment separately using a non-parametric analysis of variance (Friedman). Dichotomous data were analysed with McNemar's test. All analyses were performed with SPSS for Windows 11.5 (Chicago, IL, USA), and P < 0.05 was considered significant.
Results
During the total registration period, cilostazol induced headache in 11 volunteers compared with three after placebo (P = 0.008, McNemar's test). The median peak headache score after placebo was 0 (range 0–2) from 0 to 16 h postdose. After cilostazol, the median peak headache was 1 (0–4) during the in-hospital period (0–4 h) and 3.5 (0–7) from 4 to 16 h postdose (Fig. 1a,b). The difference was statistically significant in both periods (P = 0.03 and P = 0.003, respectively). The area under the headache score curve was similarly larger after cilostazol than after placebo during the in-hospital period (P = 0.02) and after discharge (P = 0.003). Headache characteristics and other adverse events are listed in Table 1. In the laboratory, the headache was frontal in two participants, frontal and retro-orbital in one, frontal, parietal and located at the vertex in one, exclusively at the vertex in one and occipital in the last subject. Some participants reported unilateral pain at some time point, but all of these experienced bilateral pain at other time points. The headache was pulsating and aggravated by physical activity in the majority of patients but was not accompanied by photo- or phonophobia. Two participants had nausea and thus experienced symptoms of migraine without aura according to ICHD-2 criteria after cilostazol, as did two withdrawn participants (15). Over-the-counter rescue medication according to individual preference was taken at home by eight of 12 participants after cilostazol and none after placebo (P = 0.008).
Individual (thin lines) and median (thick dotted line) headache scores after cilostazol 200 mg p.o. (a) and placebo (b).
There were no changes in pressure pain thresholds after cilostazol compared with placebo in the frontal area (P = 0.25) or on the forearm (P = 0.24) (Fig. 2a,b).
Box-and-whisker plot representing medians and quartiles of mechanical pressure pain expressed as lowest number of calibrated von Frey hairs experienced as painful in at least two of four stimulations in random order on the forehead (a) or in a control area (b). No changes were seen compared with baseline or placebo after cilostazol 200 mg.
Discussion
In the present study, cilostazol induced headache in 11 of 12 participants with no medical or family history of migraine. The headache experienced was often moderate and even severe in a few patients. Frequently it had migraine-like features such as pulsating pain quality and aggravation by physical activity. In two participants, the symptoms fulfilled criteria for one attack of migraine without aura (15). Headache has been reported as a dose-related adverse event in the clinical use of cilostazol. At a dose of 100 mg b.i.d., 33.4% of patients reported headache compared with 13.1% after placebo (16). The peak plasma concentration of cilostazol during steady-state conditions was similar to that after a single dose of 200 mg cilostazol as used in the present study (17). There are several possible explanations why we recorded considerably more headache at the same plasma levels. Young volunteers may be more susceptible to headache than the age groups normally being treated with cilostazol and acute dosing may cause more headache than chronic dosing. Further, mild or transient headaches are more likely to be reported in a study of headache mechanisms than in a clinical trial.
Mechanisms involved in the headache response
After cilostazol 200 mg, velocity in the middle cerebral artery (VMCA) measured with transcranial Doppler sonography decreased 21.5 ± 5.7% (P = 0.02 vs. placebo). Because the regional cerebral blood flow remained unchanged, this indicates an increase in MCA diameter (11). By comparison, glyceryl trinitrate (GTN) at a dose of 2 µg kg−1 min−1 decreased VMCA by 28 ± 2.6% (P < 0.001) (18). The same dose rate of GTN induced a median headache of 1.5–2 on the headache rating scale in healthy volunteers (13, 19), which appears to be less than the mean peak response to cilostazol 200 mg (3, 5). Our results concur with several previous reports on the role of cerebral artery dilation in migraine and experimental vascular headache (20–22). Intraluminal balloon inflation of the MCA causes pain in the ipsilateral temple, orbit and forehead (23), which is the commonest site of migraine, and of headache in the present study. During unilateral, spontaneous migraine attacks, a small but significant MCA dilation has been observed on the painful side and a dilation of the MCA is also seen after administration of most migraine and headache provoking substances, such as GTN and histamine (24). However, the moderate degree of dilation after cilostazol makes a simple mechanical dilation an unlikely source of pain (24). Furthermore, the PDE5 inhibitor sildenafil causes headache in healthy volunteers and migraine in migraine patients without dilating the MCA (4). Peripheral or central sensitization of the trigeminovascular system has previously been proposed (25), possibly via cAMP-dependent mechanisms (26). Unless this proposed sensitization occurs exclusively in the subpopulation of trigeminal neurons receiving input from the meninges or cerebral arteries, increased sensitivity or allodynia should also be present in the facial skin. In fact, cutaneous allodynia has already been demonstrated with VFH during spontaneous migraine attacks (27). In the present study, we found no trigeminal sensitization in the first 4 h, but since the most severe headache occurred later, we cannot rule out sensitization at a later stage of cilostazol-induced headache.
Selectivity of cilostazol
A pivotal point in the interpretation of the present study is whether cilostazol increases cAMP in cerebrovascular tissue. cAMP in vascular smooth muscle cells is predominantly degraded by PDE3 and PDE4 (10). Two different isoforms of PDE3 are described in humans, PDA3A and PDE3B (28). Cilostazol is a potent inhibitor of both isoforms of PDE3; the inhibitory concentration (IC50) is 0.20 µmol/l for the predominantly vascular isoform PDE3A, and 0.38 µmol/l for PDE3B, predominantly located in fat tissue. The IC50 for PDE5 is 4.4 µmol/l (29). Peak plasma concentrations of cilostazol after a single oral administration of cilostazol 200 mg is 2.2 ± 0.6 µmol/l occurring at 2.8 h (range 1.5–4.0 h) (17). Thus, a minor inhibitory effect on PDE5 during peak concentrations in the present study cannot be excluded, while there appears to be no relevant effect on PDE1, PDE2, PDE4 and PDE7 (29). However, the NO–cGMP pathway and hence PDE5 does not appear to play an important role in the vasodilator response to cilostazol in cerebral arteries in vitro in clinically relevant concentrations (10).
Cilostazol, like dipyridamole, is an inhibitor of adenosine re-uptake into cardiac myocytes, coronary smooth muscle cells and endothelium (30). Whether cilostazol elevates plasma adenosine in man is not known. However, adenosine and dipyridamole cause hyperventilation (31, 32), but cilostazol did not (11). Therefore, it is unlikely that plasma adenosine was much elevated in the present study. In contrast to cilostazol, intravenous adenosine causes only very mild headache in normal volunteers (33) and may even counteract headache by activation of the A1 receptor (34).
In conclusion, the present study supports the hypothesis that not only cGMP- but also cAMP-dependent pathways may be involved in the pathogenesis of vascular headache, but further investigations are needed to analyse its significance in migraine and explore the mechanisms and signalling pathways involved.
Footnotes
Acknowledgements
The authors thank the Lundbeck Foundation, the Wendelboe-Joergensen Foundation, the County of Copenhagen and the Danish Medical Research Council for generous financial support.