Meningeal Contribution to Migraine Pain

A Magnetic Resonance Angiography Study

Sabrina Khan; Faisal Mohammad Amin; Casper Emil Christensen; Hashmat Ghanizada; Samaira Younis; Anne Christine Rye Olinger; Patrick J. H. de Koning; Henrik B. W. Larsson; Messoud Ashina


Brain. 2019;142(1):93-102. 

In This Article


Here, we present the earliest MRA recordings during attacks of migraine without aura. The main finding of this study was a slight dilation of MMA ipsilateral but not contralateral to the reported head pain during the onset phase of migraine. Subsequent administration of sumatriptan constricted MMA on both the pain and non-pain sides. Moreover, the first-ever exploratory investigation of intra- and extracerebral arteries in the late phase of migraine in patients not treated with migraine-specific rescue medication, suggests that MMA is subject to prolonged, bilateral dilation.

The exact role of vascular involvement in migraine has been the topic of much debate, recently fuelled by the findings of Amin and colleagues (2013), who suggested that vasodilation itself is not a principal component in migraine pain. This notion has contributed to a shift of emphasis from mechanical distension of vascular smooth muscle cells towards mechanisms related to activation and sensitization of perivascular sensory afferents in mediating migraine pain (Ashina et al., 2017).

As MMA receives trigeminal innervation along its entire course (Schueler et al., 2014), it is a tentative assumption that circumference changes pertaining to the extracranial MMA are a direct reflection of changes occurring around intracranial MMA. An important novel finding of this study is that MMA is the only cranial artery to increase in circumference exclusively on the pain side during the early phase of migraine, possibly reflecting an activation of dural trigeminal nociceptors by cytochemical changes occurring in the perivascular space. This activation involves plasma protein extravasation (Markowitz et al., 1987) and release of various neuropeptides including the potent vasodilator calcitonin gene-related peptide (Edvinsson and Uddman, 1981; Messlinger et al., 1993).

The vasodilation observed in this study cannot be attributed to a systemic drug effect of cilostazol, as the non-pain side of MMA as well as both sides of STA did not increase in circumference. A post hoc power analysis revealed that with a significance level of 5%, a standard deviation of 0.24970 and sample size of 24, we would be able to detect a difference in circumference change between MMA pain and non-pain side of 0.15 mm with 80% power, and 0.17 mm with 90% power. The actual detected difference in circumference change of MMA in this study was 0.178 mm. Interestingly, STA and MMA, both extracranial arteries, receive comparable trigeminal innervation (Liu et al., 2003, 2008), yet they exhibit different vascular responses in the onset phase of migraine. Considering this finding, it would further appear that MMA dilation occurring ipsilateral to the headache is suggestive of a meningeal site of migraine pain.

Sumatriptan constricted all extracerebral arteries but had no effect on the cerebral arteries (i.e. MCA or ICAcerebral), corroborating earlier findings (Asghar et al., 2011; Amin et al., 2013). Apart from its vasoconstrictive properties, sumatriptan also attenuates plasma levels of CGRP during electrical stimulation of the trigeminal ganglion (Buzzi et al., 1991), blocks plasma protein extravasation in dura mater (Buzzi and Moskowitz, 1990) as well as synaptic transmission between first and second order trigeminovascular neurons (Levy et al., 2004), contributing to its migraine-aborting properties. The anti-nociceptive effect of sumatriptan treatment is reflected in patient-reported verbal rating scale scores, which dropped from 7 before sumatriptan to 3 after treatment.

Cilostazol crosses the cell membrane and acts intracellularly by inhibiting the phosphodiesterase-3-enzyme, resulting in an accumulation of cyclic AMP (Birk et al., 2004). Interestingly, our findings show that sumatriptan blocks this intracellular effect of cilostazol in extracerebral but not intracerebral arteries.

Previous MRA Studies in Migraine

Earlier MRA studies have found varying changes in MMA calibre during migraine attacks, alternating between no dilation (Schoonman et al., 2008), moderate dilation on only the pain side (Asghar et al., 2011), and substantial bilateral dilation (Amin et al., 2014). These discrepancies in vascular response could be attributed to difference in time from migraine onset to time of scan, as vascular changes during the migraine attack depend on time from onset to examination (Shayestagul et al., 2017). Difference in time to scan could also explain our finding of slight MMA dilation on the pain side, while Asghar and colleagues (2011) reported moderate dilation of the same artery. The authors did not report time from onset of migraine to MRA; however, as our early attack scan captures the onset phase of migraine, it is possible that the full potential of MMA dilation may not have been reached at this time point.

It could also be considered that different migraine triggers induce differential vascular responses over time. CGRP provocation may induce immediate migraine attacks in 40% of patients (Guo et al., 2016), whereas cilostazol only induces delayed migraine attacks (Guo et al., 2014). This distinction in induction pattern could explain our finding of pain side dilation of only MMA in cilostazol-induced migraine, whereas CGRP-induced migraine leads to pain side dilation of both MMA and MCA, suggesting that the migraine cascade has evolved further along its course at a given time point after CGRP compared to cilostazol, exhibiting dilatory responses from more arteries. Another difference between CGRP and cilostazol is that the latter dilates intracerebral arteries whereas CGRP has no direct effect on the MCA (Asghar et al., 2010, 2011). Therefore, a possible slight side-to-side difference of MCA calibre in our study may be hidden under a relatively large bilateral MCA dilation. Notably, pharmacological migraine triggers may vary in biochemical properties and direct comparisons of agents should consider these differences.

Late Attack Phase of Migraine

This MRA study is part of a larger parent MRI protocol, with a study design that included a late attack scan performed 1 day after onset of migraine. Data from this late phase of migraine allow us the unique assessment of vascular changes of the cranial circulation across time. The purpose of this exploratory analysis was to investigate whether dilation from the onset phase of migraine persisted into the late phase of the attack, and whether unilaterality of dilation was maintained. The three arteries of interest were chosen to represent the intracerebral (MCA), extra-cerebral (ICAcavernous), and extracranial (MMA) circulations. The sequential circumference changes in patients who did not treat their attacks with sumatriptan suggest that MMA further dilates in the late phase of migraine, both on the pain and non-pain side that was first reported by the patients at migraine onset. ICAcavernousmaintains its circumference increase from the early attack phase, while MCA circumference decreases towards baseline.

It could be contemplated whether the persistent dilation of MMA and ICAcavernouscan be attributed to the drug effect of cilostazol. We find this highly improbable, as any drug effect would also affect MCA, which is not the case. Also, 29 h after migraine onset, the drug effect of cilostazol would have diminished greatly, considering that cilostazol reaches its maximum plasma concentration after 3 h and has a half-life of 13 h (Bramer et al., 1999). Second, one could consider whether the persistent bilateral circumference increase is related to head pain, especially since the majority of investigated patients reported ongoing headache at the time of the late attack scan (Figure 4). Interestingly, >50% of patients reported a diffuse headache at some point between the early and late attack scans, despite the attack beginning as unilateral. These findings suggest that unilateral activation of perivascular nociceptors is a key characteristic of specifically the onset phase of migraine in our patients, associated with initiation of head pain. In later stages of the attack, intra- and extracranial arteries exhibit bilateral activation and sensitization of trigeminal perivascular nociceptors. This phenomenon corresponds to clinical reports from patients who experience unilateral head pain in the initial phase of their attack, which later spreads to a diffuse, bilateral head pain. Taken together, we suggest that the temporal changes of MMA circumference may be indicative of the ditto temporal evolution in migraine pain location, starting as unilateral and spreading to diffusely bilateral. With the present data, we cannot explain the observed sustained increase in ICAcavernous circumference in the late phase of the migraine attack.

Because of the technicalities of performing late attack MRA analyses, only three MMA segments were eligible for analysis in patients who received sumatriptan treatment. Therefore, we cannot draw conclusions on the temporal pattern of MMA circumference changes for this patient group.

The software program applied uses the partial volume effect to achieve sub-voxel accuracy. The ratio of the signal intensity inside and outside the vessel is used to place the boundary using a variation of the full-width half-maximum (FWHM) method, as previously described by de Koning et al. (2003). In this work, the authors evaluated the detection method against X-ray angiography and ground truth measurements of phantoms. The method was able to accurately detect vessels that were three voxels wide (<5% error compared to ground truth). Applying this logic to our data and using the median acquired voxel size of 0.49 mm, this would correspond to a least detectable vessel circumference of 4.7 mm. The smallest MMA circumference measured in our study was 5.00 mm. The detected difference of 0.23 in MMA circumference on the pain side is under a voxel size; however, it is a mean difference measured across 24 eligible subjects for MMA analysis, where the value for each subject is a mean of up to 26 circumference values measured every 0.2 mm on a maximally 5-mm long vessel segment. Hence, we consider this detected difference a reliable measure.

Our study has a number of limitations. First, angiography analyses of the late phase migraine attack were associated with technical challenges, as blooming artefacts due to the application of iron oxide nanoparticles as contrast agent could in principle influence the visualization of arterial luminal boundaries, leading to overestimating of circumference. However, this does not explain why only MMA and ICAcavernous, but not MCA, display dilation in the late phase of the migraine attack.

Second, as we mainly investigated the larger arteries of the brain, we cannot exclude that circumference changes also may have affected other and possibly smaller cerebral and dural arteries. However, circumference changes of the smaller cerebral arteries may alter cerebral blood flow, which does not seem to be the case in migraine without aura (Lauritzen and Olesen, 1984; Ferrari et al., 1995). With regards to smaller dural arteries, we have recently examined the relationship between the large extracranial dural MMA and the smaller intracranial portion of the artery, and found similar dilation between the two segments (Christensen et al., 2018). We are therefore confident that the MMA dilation found in our study may be considered representative of changes in the smaller branches of MMA in the first part of the cranial convexity.

Third, with an MRA study we can only glean circumference changes at a given time point along the continuum of pathophysiological events that occur before, during, and after a migraine attack. This fosters the question whether migraine without aura could be associated with short-lived changes in circumference, which could go unrecognized. While we cannot exclude having missed minute- or second-long calibre changes, we find this improbable considering the temporal aspect of dilation induced by vasoactive neuropeptides. An experimental migraine study demonstrated that pituitary adenylate cyclase-activating peptide-38 (PACAP-38) infusion induced prolonged vasodilation of the superficial temporal and middle meningeal arteries lasting at least 2 h (Amin et al., 2014). Also, CGRP infusion was shown to cause vasodilation of the superficial temporal artery after pretreatment with placebo compared to pretreatment with the CGRP receptor antagonist olcegepant; a dilatory response that lasted from 30 to 150 min after infusion (Petersen et al., 2005). Furthermore, Birk and colleagues (2004) demonstrated that cilostazol induced a reduction in MCA blood flow velocity that did not reach a plateau within the 4-h observation period. Based on the assumption that activated perivascular dural nociceptors would result in release of vasoactive peptides, including CGRP and PACAP-38, and corroborating this assumption with the presented experimental findings, we find it improbable that migraine would be associated with fleeting vasodilation.

The major novel finding of the present study is that initiation of a migraine attack is associated with increase in MMA circumference on the head pain side, suggesting activation of dural perivascular nociceptors. We propose that this headache-specific dilation of MMA precedes unilateral dilation of intracerebral arteries in the initiating cascade of migraine, suggesting a meningeal site of migraine headache.