Non-invasive methods for measuring vascular changes in neurovascular headaches

Abstract

Vascular changes during spontaneous headache attacks have been studied over the last 30 years. The interest in cerebral vessels in headache research was initially due to the hypothesis of cerebral vessels as the pain source. Here, we review the knowledge gained by measuring the cerebral vasculature during spontaneous primary headache attacks with the use of single photon emission tomography (SPECT), positron emission tomography (PET), magnetic resonance imaging (MRA) and transcranial Doppler (TCD). Furthermore, the use of near-infrared spectroscopy in headache research is reviewed. Existing TCD studies of migraine and other headache disorders do not provide solid evidence for cerebral blood flow velocity changes during spontaneous attacks of migraine headache. SPECT studies have clearly shown cortical vascular changes following migraine aura and the differences between migraine with aura compared to migraine without aura. PET studies have shown focal activation in brain structures related to headache, but whether the changes are specific to different primary headaches have yet to be demonstrated. MR angiography has shown precise changes in large cerebral vessels during spontaneous migraine without aura attacks. Future development in more precise imaging methods may further elucidate the pathophysiological mechanisms in primary headaches.

Introduction

The interest in vascular mechanisms in primary headaches started with the pivotal study by Ray and Wolff¹ showing that mechanical, chemical and electrical stimulation of cranial vessels resulted in migraine-like headache.¹ This led to the concept that there is dilatation of intracranial vessels during migraine attacks² and subsequently result in activation of trigeminal nociceptors leading to the perception of pain.³ Although this concept has been greatly challenged,⁴ there have been extensive research focusing on the cranial vessels in migraine and other primary headaches.^3,5 Cranial vessels can be studied in experimental animal models⁶ and by the use of headache provocation models using vasoactive neuropeptides covered elsewhere in this special issue (Cerebral hemodynamics in the different phases of migraine and cluster headache by JM Hansen and CJ Schankin). Investigation of vascular changes during headache attacks are obviously of great interest and with the development of non-invasive imaging methods several decades after the pivotal studies by Ray and Wolff,¹ it became possible to study the intracranial vasculature in vivo during spontaneous headache attacks. This evolved into studying several different aspects of headache pathophysiology: (1) cortical blood flow studies, primarily during migraine aura, with single photon emission tomography (SPECT),⁷ positron emission tomography (PET)⁸ and later magnetic resonance imaging (MRI);⁹ (2) measuring intracranial vessel diameter changes indirectly with transcranial Doppler (TCD)¹⁰ and later directly with MRI;¹¹ (3) indirectly measuring brain activity through local blood flow changes using PET¹² and later MRI¹³ (Table 1). This review attempts to give an overview of non-invasive imaging techniques that have been used when assessing vascular changes in neurovascular headaches. We will only focus on describing vascular changes during spontaneous attacks, as neurovascular mechanisms of primary headaches, imaging findings in induced attacks with signaling molecules, resting state and functional imaging will be covered elsewhere in this journal edition. In addition, we will review the use of near-infrared spectroscopy (NIRS) in headache imaging as this is a relatively new non-invasive technique, which is promising as a tool to measure cortical changes during spontaneous attacks. We will discuss the existing findings in relation to headache pathophysiology and highlight the importance and challenges in measuring vascular changes during spontaneous attacks and point to future studies.

Table 1.

Comparison of practical issues for TCD, SPECT, PET and MRI.

	Doppler	SPECT	PET	MRI
Price	< 20,000 USD	∼ 1 million USD	∼ 2 million USD	∼ 1 million USD
Availability	High	Moderate	Limited	Moderate
Temporal resolution	Continuously	5–10 min (i.a adminstration) 45 min (inhalation)	Every 90 s	Every 1 s
Spatial	Low, only the insonated vessel	8–10 mm	5 mm	< 1 mm
Blood flow	Indirectlya	Yes	Yes	Yes
Radiation exposure	No	Yes	Yes	No

^ablood flow assessment assumes constant diameter in the insonated vessel.

i.a: intra-arterial; PET: positron emission tomography; SPECT: single photon emission computer tomography; TCD: transcranial Doppler; MRI: magnetic resonance imaging.

Methods

TCD

TCD was introduced in 1982 by Aaslid et al.¹⁴ and has over the last 30 years been used to measure blood flow velocity (BFV) in the middle cerebral artery (MCA), the anterior cerebral artery (ACA) and the posterior cerebral artery (PCA) in headache and other neurological disorders such as stroke.¹⁵ This was achieved by changing the Doppler frequency from the previous conventional 5–10 MHz to 1–2 MHz for which the attenuation in bone is much less. Furthermore, to achieve in-depth recordings, an acoustic lens with a focal length of 5 cm was applied.¹⁴ It is necessary to assume that the angle between the ultrasound beam and the intracranial artery is sharp, between 0 and 30°, which results in a maximum error of 15%.¹⁴ The blood velocity V can then be calculated by measuring the Doppler shift via the equation: V = 0.039 × f, where f is the Doppler frequency shift. When measuring changes in velocity and assuming that the cerebral blood flow is constant, it is also possible according to Dahl et al.¹⁶ to retrieve the diameter change Δd of the MCA by the equation:

$$\Delta \text{d}=((\sqrt{\frac{\text{VMCAbaseline}}{\text{VMCAchange}}}-1)\times 100)$$

The Dahl et al. study proposed this equation following infusion of the vasoactive drug nitroglycerin, but assuming or controlling for no changes in CBF during a spontaneous migraine, this equation can also be applied for analysis of spontaneous attacks. One important issue is that changes in respiration frequency may often lead to either hypocapnia or hypercapnia and thereby to a compensational change in CBF, whereas the diameter of the large arteries such as the MCA remains constant, which results in a proportional change in MCA blood velocity. This effect can be corrected by measuring end-tidal carbon dioxide partial pressure (P_CO2), and computing a corrected velocity normalized to a standard P_CO2 value of 40 mmHg as detailed by Markwalder et al.¹⁷

$$V_{\text{C}}=V_{\text{I}}·{\text{ e}}^{0.0\text{34}\cdot \left({\text{P}}_{{\text{CO}}_{\text{2}}}^{40\text{\hspace{0.17em}mmHg-PI}}{\text{CO}}_{\text{2}}\right)}$$

This correction factor may, however, underestimate the effect of CO₂ on the velocity change in some conditions. In a human migraine model testing pituitary adenylate cyclase activating peptide 38 (PACAP38), which causes hyperventilation,¹⁸ the MCA diameter dilation measured with TCD and calculated with the above equation was 9.5% at the end of a 20-min infusion of 10 ρ m/kg/min PACAP38.¹⁸ However, the exact same dosage in a MR angiography study only induced an insignificant 0.8% change in MCA circumference at 20 min.¹⁹ Thus, it is possible that under some circumstances, such as infusion of vasoactive substances, there is an underestimation of the CO₂ effect on the velocity change in intracranial arteries as assessed by TCD.

The fact that some primary headaches, such as migraine and cluster headache (CH), are often or always unilateral in presentation makes TCD an ideal tool. Vessels from both hemispheres can be assessed simultaneously, with the MCA velocities being nearly equal on the two sides in the healthy adult,¹⁴ and with day-to-day variations of only 16% being reported.²⁰ Today, the TCD technique can be used at the bedside with a swift application of the ultrasound probes, which makes the system ideal for experimental monitoring during headache attacks.

SPECT

SPECT is a nuclear imaging technique, which measures and quantifies cerebral blood flow. In SPECT, a gamma ray is generated by the decay of a radioisotope, and directly or indirectly recorded by a gamma camera rotating around the head, which can then, via reconstruction algorithms, create two- or three-dimensional images. SPECT has a lower reconstructed spatial resolution from 8 to 10 mm,²¹ is less expensive, and is more widely available than PET (Table 1), and it utilizes radioisotopes with longer half-lives from hours to days, which makes it possible to have the radioisotopes delivered from an off-site supplier. Corrected CBF values, as with TCD, have to be corrected for significant changes in P_etCO₂ by 2% for each mmHg change in P_etCO₂²² in order to measure changes related to changes occurring as a result of the headache attack and not due to a change in respiratory rate. Regional and global CBF measurements with ¹³³Xe inhalation SPECT have been shown to be a reproducible method for assessing absolute changes¹⁷ with a day-to-day coefficient of variation of 8%,²³ which makes it slightly better than MRI and approximately equal to PET.²⁴ SPECT ictal recordings have been performed in headache research since the late 1970s to measure CBF alterations during migraine aura and migraine without aura attacks. The first studies used up to 16 collimated sodium iodide (NaI) crystal scintillation detectors placed over both hemispheres and thereby having a very limited spatial resolution.²⁵ Later, a 254-channel instrument was developed leading to an improved spatial resolution.⁷ The temporal resolution can be brought down to 5–10 min using intra-arterial administration, but this is a relatively invasive procedure and therefore the use of SPECT in spontaneous headache attacks in larger studies has ceased over the last two decades.

PET

PET can be used to study cerebral blood flow changes and spontaneous activation of brain areas. PET uses radiotracers that emit positrons when they decay. The emitted positron travels a few millimetres and then interacts with an electron, leading to annihilation of both particles, and thereby creation of two gamma rays going in opposite directions, which can then be detected simultaneously. Through this detection, a line of coincidence can be algorithmically constructed and the location of the original emitted proton can be topographically mapped from the detection of thousands of gamma ray pairs. The radiotracers used for PET can be physiologically relevant tracers such as [¹⁵O]H₂O for CBF measurements²⁶ or 2-[¹⁸F]flouro-2-deoxy-D-Glucose (FDG) for metabolism measurements.²⁷ Specifically designed PET radiotracers can also be used to study pharmacological effects in the brain. As an example, Hostetler et al.²⁸ showed in vivo quantification of calcitonin gene-related peptide (CGRP) receptor occupancy by telcagepant in rhesus monkey and human brain using the PET radiotracer [11C]MK-4232, and showed a low receptor occupancy in the brain at an efficacious dose. Thus, it is unlikely that antagonism of central CGRP receptors are required for migraine efficacy, which point to extracerebral mechanisms being responsible for the generation of pain in migraine Similarly, Schankin et al.²⁹ recently demonstrated in a PET study with the radiotracer (11)C-dihydroergotamine (DHE), which is chemically identical to pharmacologically active DHE, that DHE does not cross the blood–brain barrier (BBB) during an induced migraine attack by a nitric oxide donor. Future studies may show the exact location of a CGRP or other signalling molecule antibodies used for migraine treatment.

Magnetic resonance angiography

Three-dimensional magnetic resonance angiography (MRA) using the time-of-flight (3D-TOF-MRA) method has been one of the most important non-invasive approaches for in vivo examination of the human cephalic and cerebral arteries in headache studies during the last ten years.^19,30–34 The 3D-TOF-MRA method provides high resolution (< 1 mm voxel size) images of the arteries in and the scanning requires no contrast agent,³⁵ making it the method of choice when repeated measurements are required. The technique is based on flow-related enhancement of magnetically unsaturated blood entering stationary regions which have been previously magnetically saturated by multiple radio frequency (RF) pulses. Thus, the ‘fresh’ blood appears brighter compared to the background, enabling measurement of the luminal diameter and circumference of a given artery. This method was first introduced in the field of migraine research in 2008³⁴ and has since been used in several headache and migraine studies.^{11,19,32,33,36,37} The intra- and inter-observer variations using this method are less than 5%, while day-to-day and side-to-side variations in the luminal circumference of the MCA and the middle meningeal artery (MMA) are below 10%.³⁷ Apart from non-contrast and radiation-free scans allowing repeated recordings, the advantages of this method are that it can measure several arteries simultaneously. Disadvantages of this method include the long image acquisition time compared to conventional computerized axial tomography (CAT-scan) angiography (i.e. acquisition time < 1 min), increasing the risk of movement artefacts and the audible noise produced by the scanner machine, which can aggravate pain in patients suffering from phonophobia during the migraine attack.

NIRS

NIRS is a non-invasive imaging modality that uses near-infrared light to measure changes in cerebral oxy- (HbO) and deoxy-hemoglobin (HbR) concentrations. It has become an effective tool to study normal and pathological brain physiology.^38,39 Following the first proof-of-principle experiment in a human subject in 1977,⁴⁰ the 1980s saw the development and clinical implementation of cerebral oximeters that use the NIRS technology to monitor trends in cerebral oxygenation, mainly for neuromonitoring during cardiovascular surgeries.^41,42 In 1993, came the first demonstrations of the capability of NIRS to measure the functional hemodynamic response to cortical activation. ^43–45

Functional NIRS (fNIRS) uses near-infrared light emitters – laser diodes or light-emitting diodes (LEDs) – at typically two wavelengths to illuminate the scalp. The light propagates diffusively through the scalp, skull and brain, and is partly absorbed by different chromophores, mainly the hemoglobin of the microvasculature blood. A photodetector – generally a photodiode or avalanche photodiode (APD) – is placed a few centimeters away from the source (3–5 cm) and detects part of the diffused light (Figure 1). As cerebral blood volume and oxygenation vary, the local absorption of light and therefore the detected light intensity change accordingly. Through models of light propagation in the head,^46–48 and taking advantage of the differential absorption spectra of HbO and HbR, one can retrieve the concentration changes in these two hemoglobin species, and consequently total hemoglobin (Hbt). The normal hemodynamic response to the increased metabolic demand arising from neuronal activity is a focal increase in blood flow largely exceeding the increased oxygen consumption,⁴⁹ resulting in the typical increase in HbO and decrease in HbR measured by fNIRS (Figure 1).

Figure 1.

(a) Array of NIRS sources and detectors (dots), and corresponding channels (lines). (b) Sensitivity profile of an fNIRS channel (source-detector pair) from Monte Carlo simulations. (c) Typical hemodynamic response to cortical activation measured by fNIRS.

Generally, an array of light sources and detectors is placed on the head, with each source-detector pair creating a measurement channel, and the combination of all channels allowing the retrieval of a two-dimensional map of the hemodynamic response to brain activation. Most traditional fNIRS systems employ optical fibers to deliver light to and from the head, but recent wearable devices instead place sources and detectors directly on the scalp.^50–54 A variety of fNIRS devices are available commercially and in research settings,⁵⁵ varying in flexibility and complexity, from simple forehead-only monitors to high-density whole-head systems.⁵⁶

fNIRS offers many attractive features for a wide range of functional studies: relatively low-cost, non-invasiveness, portability, tolerance to motion, high temporal resolution, and functional contrast. Despite these assets, the modality also has limitations that need to be considered. First, its spatial resolution is roughly a couple of centimeters (roughly the source detector separation), although advances in instrumentation (e.g. high-density arrays) and image reconstruction algorithms (e.g. anatomical guidance^57–59) can improve resolution potentially by a factor of 3.⁶⁰ Second, as the light must travel back and forth through the scalp to reach the brain, measurements are only sensitive to the superficial cortex and are contaminated by the extracerebral vasculature that reflects systemic physiology. Multiple approaches have been proposed and are still being actively developed to minimize this contamination.^61–63 Finally, because fNIRS is particularly suited to populations where subjects are moving or talking, it is prone to motion artifacts. Hardware and software methods can reduce this motion contamination,^50,64 or partially correct for them during post-processing.^65–67

The range of fNIRS applications and users has grown exponentially since the first demonstration of the technology,^38,39 encompassing notably the fields of developmental neuroscience,^68,69 psychiatric conditions,⁷⁰ and neurological disorders and injuries such as migraine, epilepsy, movement disorders and stroke.⁷¹ It is particularly adapted to challenging populations that may not be amenable to other imaging modalities, including infants and children, neurocritical care patients, and freely-moving, behaving participants (testing of cognitive or executive functions, gait studies, speech studies). In addition to traditional evoked-response functional studies, some research has focused on the spontaneous oscillations in the fNIRS signals, driven either from neuronal or vascular fluctuations.⁷² In this context, fNIRS has been applied to study resting-state functional connectivity,^73–76 and cerebral autoregulation and vascular reactivity.^15,66,77

Studies of spontaneous attacks in migraine

TCD in migraine

A total of 17 studies of spontaneous migraine attacks have been published between 1990 and 2010.⁷⁸ The studies have examined either migraine with or without aura or mixed populations of both migraine subtypes. Most of the studies reported no change in the BFV during spontaneous attacks versus outside of the attack. Measurement of the BFV has traditionally been used as a surrogate marker of dilation of the MCA in spontaneous migraine studies.¹⁷ This is possible if the change in cerebral blood flow (CBF) is known at the same time. SPECT studies have demonstrated that CBF changes during attacks of migraine with aura,⁷ but not without aura.⁷⁹ It is therefore necessary to assess each migraine subgroup (i.e. with and without aura) separately to interpret possible MCA dilation based on BFV measurements. In a recent systematic review of TCD measurements of the MCA during spontaneous migraine attacks,⁷⁸ data from all published studies were divided into three subgroups: (1) migraine with aura, (2) migraine without aura, and (3) mixed migraine (Table 2). In the migraine with aura group, four out of nine studies reported decreased BFV during migraine attacks, and four studies reported no change. In the migraine without aura group, only 2 out of 11 studies reported decreased BFV during attacks, whereas all others showed no change. Four studies reported data on mixed (with and without aura) migraine. Two studies reported decreased BFV, while two found no change. Overall, most studies showed unchanged BFV during spontaneous migraine attacks, but interestingly decreased BFV was more prevalent in migraine with aura studies.⁷⁸ This trend was more pronounced in the earlier patients who were assessed with TCD during the attack.⁷⁸ Six studies reported data on the comparison of pain side BFV versus the non-pain side BFV.⁷⁸ Five of these studies reported no change, whereas one study reported significantly more decreased BFV in the MCA on the pain side in patients suffering from migraine without aura.⁷⁸ The higher prevalence of BFV changes in migraine with aura in the earliest phase of the attack may suggest that TCD measurements in spontaneous migraine may be affected by changes in CBF. We cannot fully exclude that subtle CBF changes are also present in migraine without aura that could not be detected by conventional SPECT methods. Thus, inferences about vasodilation based on TCD measurement should be very cautious. In conclusion, existing TCD studies of migraine do not provide solid evidence for vasodilation (using BFV changes as a surrogate measure) during spontaneous attacks of migraine headache.

Table 2.

Studies investigating vascular changes during spontaneous migraine attacks using SPECT.

Authors	Method	Patients	Recordings	Main findings
Andersen et al.80	SPECT 133xenon inhalation	7 MA patients	MA attacks admission, 3–8 h, 20 to 24 h and 1 week later after the onset of symptoms	19% reduced rCBF 3 h followed by 19% hyperperfusion in the affected hemisphere
Lauritzen et al.81	SPECT 133xenon intraarterial injection	9 FHM patients	MA attacks, series of rCBF studies, spaced by intervals of 5 to 10 min	Reduced rCBF. The hypoperfusion spread 2 mm/min over the hemisphere
Olesen et al.7	SPECT 133xenon intraarterial injection	9 MA patients + 1 MO patient.	MA attacks in 6 MA patients, series of rCBF studies, spaced by intervals of 5 to 10 min	The attacks were initiated by focal hyperemia in three patients followed by hypoperfusion gradually spread anteriorly in the course of 15 to 45 min. In four patients, a global hypoperfusion. was observed
Lauritzen et al.81	SPECT 133xenon inhalation	12 MO and 11 MA patients	Spontaneous attacks, MA patients investigated at a mean of 2 h into attacks, while MO patients were scanned at a mean of 7 h into the attack	Normal CBF during MO attacks in 12 patients. 8/11 MA patients displayed a unilateral region of hypoperfusion during attacks, while 3 had a normal flow pattern
Friberg et al.83	SPECT 133xenon intraarterial injection	3 FHM patients	FHM attacks	Focal hypoperfusion in the frontal lobe.In association with the rCBF changes the patients developed transient motor and/or sensory deficits
Olesen et al.7	SPECT 133xenon inhalation	12 MO patients	Food induced attacks. Up to four measurements	No change in CBF during MO attacks
Sakai and Meyer25	SPECT 133xenon inhalation	2 MA patients  3 MO patients  3 CH patients  7 TTH patients	Measurements during headache and 5–14 days later	The five MO/MA patients had a mean increase of 31% in total CBF compared to the interictal phase. The three CH patients had a mean increase of 45% in total CBF compared to the interictal phase.  No increase in TTH patients
Ferrari et al.85	SPECT using technetium Tc99m hexemethyl-propyleneamineoxime	20 MO patients	Measurements outside an attack, during an attack and after attack treatment of with 6 mg of subcutaneous sumatriptan	No detectable focal changes of the rCBF during the different conditions
Kobari et al.86	133xenon inhalation	18 MO patients  12 MA patients	During attacks (n = 9 MO and n = 7 MA) and without headache (n = 9 MO and n = 5 MA)	The rCBF increased in the fronto-temporal cortex, thalamus and basal ganglia. No significant diffences between MO and MA patients

SPECT: single photon emission computer tomography; MA: migraine with aura; FHM: familial hemiplegic migraine; MO: migraine without aura; MA: migraine with aura; TTH: tension-type headache.

SPECT in migraine

In the late 1970s, SPECT studies began to emerge, which recorded regional CBF (rCBF) changes during spontaneous migraine attacks. Sakai and Meyer²⁵ investigated migraine and CH patients during attacks compared to healthy controls as well as the interictal headache using the Xenon inhalation method with 16 detectors placed over both hemispheres and thereby having a very limited spatial resolution. The study compared three migraine without aura (MO) and two migraine with aura (MA) patients with serial images during the headache phase and when headache free.²⁵ The authors found a mean increase of 31% in total CBF during attacks in these five patients with no apparent differences between migraine subtypes. Interestingly, one MA patient was described in detail, in which there was a local CBF reduction in the post-central gyrus corresponding to the contralateral loss of sensation.²⁵ Three years later, the groundbreaking study by Olesen et al.⁷ was published, which used the intra-arterial Xenon 133 method and 254 channel equipment with repeated measurement by 5–10 min, which was able to detect that in migraine with aura, attacks were initiated by focal hyperemia and then an oligemia which gradually spread anteriorly in the course of 15 to 45 min.⁷ Thus, the study was the first to clearly show what may be the vascular footprint of cortical spreading depression in migraine aura. Then followed a series of studies, which replicated the findings.^80–83 This was achieved using Xenon intraarterial injection and Xenon inhalation (Table 2). The studies from the Copenhagen group also showed that during familial hemiplegic migraine (FHM) attacks, a pattern similar to that of MA is observed.⁸³ Furthermore, they observed no increase in CBF during MO attacks.^82,84 This was also confirmed using the technetium Tc99m hexemethyl-propyleneamineoxime method.⁸⁵ In contrast, Kobari et al.⁸⁶ investigated MO and MA patients during attacks and observed that CBF increased by 20–40% in the fronto-temporal cortex, thalamus and basal ganglia, which may be due to cortical activation as a result of pain.⁸⁷ In addition, they observed no differences between MO and MA patients, which is in contrast to the majority of studies, but similar to the Sakai study.²⁵ The SPECT studies in migraine laid the foundation to describing the pathophysiological background behind migraine with aura and showed, in most cases, a clear difference in vascular alterations between spontaneous MO and MA attacks.

PET in migraine

To our knowledge, there have been a total of seven studies investigating spontaneous attacks with the PET technique (Table 3), and most of these studies have been recognised as pivotal studies, which moved the field of knowledge in migraine pathophysiology. The first study, which was in sharp contrast to the previous SPECT findings, was a result of an incident. A migraine without aura patient experienced a spontaneous migraine, while being in a PET scanner as a healthy subject undergoing 12 repeated measurements every 15 min during a visual test with a series of line drawings displayed at a rate of 2/s.⁸⁸ Before the seventh scan, the patients experienced a headache fulfilling the criteria for MO, and without a clear aura prior or during the attack, except for the retrospective description by the patient 30 min after the headache that she had been “unable to focus her vision clearly on the screen”.⁸⁸ This study has laid ground for an unresolved debate on whether this single finding is demonstrating so-called silent aura hemodynamic findings due to CSD or if the results are due to atypical migraine aura or methodological pitfalls.^89–91 Another group demonstrated in nine MO patients scanned within 24 h from headache debut and compared to the headache-free interval a small but significant decrease in global CBF (10%) and CBV (10%).⁹² Later, Denuelle et al.⁹³ also showed a 10% bilateral posterior cortical hypoperfusion during spontaneous MO attacks.⁹⁴ The study contrasts with the original study by Weiller et al.,¹² who demonstrated in nine MO patients with unilateral attack patients no ictal changes in rCBF between the hemispheres ipsilateral and contralateral to the headache side. However, the study showed a 11% rCBF increase during the acute attack compared to the headache-free interval in median brain stem slightly contralateral to the headache side. The PET resolution could not identify activation at specific brainstem nuclei, but the activation was in an area which harbours the dorsal raphe nucleus (DRN) and the locus coeruleus (LC), which have been suggested to provide autonomic control of the cerebral and dural blood flow.⁹⁵ Other PET studies have shown focal activation in the brainstem, hypothalamus, anterior cingulate, posterior cingulate, cerebellum, thalamus, insula, prefrontal cortex, and temporal lobes.^96,97 In conclusion, some PET studies have shown a small decrease in global and occipital lobe CBF during spontaneous attacks, but the studies are hampered by the lack of serial ictal measurements and new more precise studies are highly needed. There have been findings of focal activation in brain structures related to migraine, but whether the changes are specific to migraine has yet to be demonstrated.

Table 3.

Studies investigating vascular changes during spontaneous migraine attacks using PET.

Authors	Method	Patients	Recordings	Findings
Weiller et al.12	PET with C15O2 inhalation	9 MO patients with only unilateral attacks of migraine	rCBF during spontaneous attacks (< 6 h from debut), after the relief of headache by 6 mg s.c. sumatriptan and during headache-free interval > 3 days after the headache	No differences in rCBF were found between the hemispheres ipsilateral and contralateral to the headache side.  11% rCBF increase during the acute attack compared to the headache-free interval in median brain stem  Structures contralateral to the headache side
Bednarczyk et al.92	PET with C15O2 inhalation	9 MO patients	Spontaneous attacks (< 24 h from debut) and during headache-free interval > 48 h	Global CBF decreased 10% and CBV 5% during migraine attacks compared to the headachefree state
Afridi et al. 200597	PET with C15O2 iv infusion	3 MO and 2 MA patients	Spontaneous attacks (< 24 h from debut) and during headache-free interval > 72 h. Three right-sided and two left-sided attacks	Activation seen in the dorsal pons, lateralized to the left, right anterior cingulate, posterior cingulate, cerebellum, thalamus, insula, prefrontal cortex, and temporal lobes. An area of deactivation, located in the pons, lateralized to the right
Denuelle et al.96	PET with C15O2 iv infusion	7 MO patients	Spontaneous attacks (< 4 h), after headache relief following 6 mg sumatriptan s.c. and during headache-free interval (15 to 60 days later). Three right-sided, three left-sided and one bilateral	Activations in the midbrain bilateral, pons and hypothalamus bilateral, all persisting after sumatriptan
Denuelle et al.93	PET with C15O2 iv infusion	7 MO patients	Spontaneous attacks (< 4 h), after headache relief following 6 mg sumatriptan s.c. and during headache-free interval (15 to 60 days later). Three right-sided, three left-sided and one bilateral	10% bilateral posterior cortical hypoperfusion, which persisted (12%) after headache relief following sumatriptan injection
Woods et al.88	PET and oxygen-15 labelled water	1 apparent MO patient	1 unexpected attack. 12 measurements at 15 min interval	Bilateral hypoperfusion in the occipital lobes at headache onset. rCBF decreased 40% and spread forward
Andersson et al.94	PET and oxygen-15 labelled water	3 MO and 8 MA patients	Wine induced attacks. Scans at aura phase, headache phase, after sumatriptan and later a baseline scan	Reduced flow (23%) and metabolism (23%) in the occipital lobe during head ache phase

PET: positron emission tomography; MA: migraine with aura; MO: migraine without aura; MA: migraine with aura; rCBF: regional cerebral blood flow.

MRA in migraine

Although several headache and migraine studies have been conducted using this method, there is only a single study reporting data on spontaneous migraine attacks.¹¹ In this study, the long-standing theory of extracranial vasodilation as the cause of the pulsating migraine pain was tested using a 3D-TOF-MRA scan during spontaneous attacks of unilateral migraineous pain in 19 patients suffering from migraine without aura.¹¹ These patients were subsequently scanned 30 min after subcutaneous injection of 6 mg sumatriptan on the migraine attack day, and again on another migraine- and headache-free day. The circumference of several intra- and extracranial arterial segments was measured and compared between the three conditions.¹¹ Data revealed a 13.0% dilation of the MCA, 11.5% dilation of the cerebral ICA and 11.4% dilation of the cavernous ICA on the pain-side during attack compared to outside of attack. A modest dilation was also reported on the non-pain side MCA (6.1%) and cavernous ICA (4.9%).¹¹ However, the side-to-side comparison of these arteries showed 9.1–14.4% more dilation on the pain-side versus non-pain side during the attack day. No day-to-day or side-to-side differences of the extracranial arteries were demonstrated in the study.¹¹ After effective treatment with sumatriptan, all extracranial arteries and the cavernous ICA were constricted by 10–20%, whereas the MCA and cerebral ICA remained dilated.¹¹ These segments are thought to be protected by the blood–brain barrier, which may explain the lack of constrictor effect of sumatriptan. The cavernous ICA was the only arterial segment that was dilated during attack and constricted after sumatriptan. This part of ICA is in the bone and is covered by nerve plexus, which may be compressed by the surrounding bone when the artery dilates. May et al.⁹⁸ reported dilation of the cavernous ICA after activation of the trigeminovascular system by ipsilateral injection of capsaicin in the forehead. Thus, the conclusion of this study was that migraine pain may be associated with intra- but not extracranial vasodilation.⁹⁸ However, dilation of the MCA and ICA was not regarded as the cause of the pain, as effective treatment with sumatriptan did not reverse the dilation of these arteries. The strengths of this study are the investigation of spontaneous migraine attacks and the fact that the attacks were unilateral. Thus, the non-pain side can be used as control for the pain side. The limitation of the study was that only 19 patients were included, so that subgroup analyses were not possible. Examination of untreated spontaneous migraine attacks in a MRI scanner is a difficult task that requires an effective and demanding collaboration between many different departments. Although, no statistical significant extracranial dilation was found, data revealed that circumferences of all extracranial arteries were numerically greater on the pain side versus non-pain side. Thus, subtle circumference changes (less than 10%) cannot be rejected based on these findings. A small (5%) dilation of MMA on the pain side has previously been reported in a case-report of one patient.⁹⁹ Subtle dilation of the MMA per se may not be painful as drug-induced 30% dilation of this artery has been reported in migraine without aura patients who had no pain.³⁶ The ipsilateral lateralisation of small circumference changes could indicate perivascular inflammation causing slight relaxation of the vessel walls.

NIRS in migraine

NIRS has been used in migraine research to investigate cortical vasomotor reactivity during metabolic,^100–102 physiological,^103,104 pharmacological^105,106 and functional tests.^107,108 NIRS has also been used in one study investigating neurostimulation in CH.¹⁰⁹

Using a breath hold (BH) test to investigate vasoreactivity as a result of changes in arterial CO₂ concentration,¹¹⁰ Akin et al.¹⁰⁰ showed that interictal HbR increase following BH was smaller in MO than in healthy controls, while HbO increase was similar in the two groups. In a different study, Liboni et al.¹⁰¹ showed smaller HbO increases and lower HbR decreases in 30 MO patients than in healthy controls during BH, which may have been biased by a shorter breath holding duration in MO patients than in healthy subjects. Another method of assessing vasomotor reactivity is via head-down maneuver inducing an increase in frontal lobe pressure, which usually leads to elevated CBF. This maneuver produced a significantly smaller increase in right-sided HbT in migraineurs when compared to volunteers.¹⁰³ Migraine patients showed a smaller increase in right-sided HbT than in left-sided HbT following the head down maneuver. In addition, migraine patients had a greater increase in right-sided HbT than healthy volunteers in left-sided HbT.¹⁰³ It is difficult to draw clear conclusions from this study as the patients were not coded as having MO or MA. The lateralized vasoreactivity is difficult to understand, but has been attributed to a sympathetic lateralization in the right hemisphere,^111,112 even though it is not known if this may lead to asymmetric vasomotor responses. Vernieri et al.¹⁰² investigated side differences in MA patients in the interictal period, showing an increased cerebral vasoreactivity to CO₂ inhalation measured by NIRS on the side of most frequent attacks. A recent study investigated 10 MA and 10 MO patients during the interictal period of migraine in comparison to healthy controls.¹¹³ The time-delay between the R-wave of an electrocardiogram and the arterial pulse wave of cerebral microcirculation, detected by NIRS on the frontal cortex of both sides, was used to evaluate the presence of cerebral arteriolar vasoconstriction. Both migraine groups had a significantly longer time delay than the control subjects, but there were no differences between sides. The same group measured the amplitude of the arterial pulse wave in patients with prolonged aura (1 h to 7 days) and found during aura a significant decrease of the arterial pulse wave and an increase of cerebral tissue oxygen saturation ipsilateral to the pain side and contralateral to aura symptoms compared with the headache-free periods.¹⁰⁴ Overall, the studies on vasomotor reactivity are few and do not translate into one concept of altered vasomotor reactivity following metabolic or physiological changes in migraine patients.

NIRS has been used to investigate sumatriptan treatment investigated MO patients during attacks followed by subcutaneous sumatriptan or sham injections.¹⁰⁵ After sumatriptan injection HbO decreased,¹⁰⁵ which was correlated to changes in skin blood flow measured by laser Doppler flowmetry. Human studies have not shown that sumatriptan constricts extracerebral arteries more than cerebral arteries,^11,30,114 and the decrease found by may therefore very likely be a systemic extracerebral effect. Thus, NIRS monitoring using scalp probes are probably not entirely brain specific, and the clinical value of this type of investigation, when systemic changes are expected, is questionable. Studies attempting to separate superficial and cerebral signals would be useful to further explore these types of pharmacological effects. In patients with familial hemiplegic migraine (FHM), Schytz et al.¹⁰⁶ investigated the changes in low-frequency oscillations following glyceryl trinitrate (GTN), a nitric oxide donor. Interestingly, only FHM patients with common migraine subtypes had a lower increase in LFO amplitude compared to healthy controls and pure FHM patients, which suggests that the vascular differences between FHM patients and healthy controls were due to the common migraine subtypes.¹⁰⁶

fNIRS studies have been performed in a few studies. Schytz et al.¹⁰⁷ investigated frontal activation using a Stroop test interictally in 12 MO patients compared to 12 healthy controls and observed a clear vascular response in both groups, but no differences in amplitude or time response between groups, pointing to a normal neurovascular coupling in the frontal cortex. In contrast, Coutts et al.¹⁰⁸ investigated hemodynamic response to a range of visual stimuli in 15 MA and 5 MO patients. There were no differences in the amplitude of the response between migraine and control groups, but migraine patients had a faster hemodynamic response to visual stimuli, which, however, normalized after applying active lenses that reduced distortions of text and improved comfort. The underlying mechanism for the time differences in hemodynamic response and their normalization is unknown and needs further investigation. Thus, there may be differences in neuronal, vascular or neurovascular coupling in migraine patients following stimuli, but so far this has only been shown in the occipital cortical vasculature interictally. NIRS may also be used to investigate responses to neurostimulation, which was shown in a small study with five CH patients with sphenopalatine ganglion (SPG) stimulators, in which high frequency SPG stimulation induced a relative increase in HbO in the frontal region primarily on the stimulated ipsilateral side.¹⁰⁹ In conclusion, NIRS has been used as a non-invasive tool to measure cortical changes following different types of experimental stimulation paradigms. So far, there are no studies investigating vascular changes during spontaneous attacks. The development of new portable NIRS devices,¹¹⁵ which can even be controlled by smartphones¹¹⁶ may result in the opportunity to investigate cortical changes during spontaneous attacks. It is possible that patients could apply the equipment during spontaneous attacks at home and measure continuously for hours. The challenges of NIRS recording during spontaneous attacks would be the lower spatial resolution and motion artifacts.

Studies of spontaneous attacks in trigeminal automatic cephalalgias

TCD in TACs

TCD studies in trigeminal autonomic cephalalgias (TACs) are very rare, most likely due to the nature of the disorders. Most patients are agitated during the attacks making it difficult to lie still for TCD measurements. Afra et al.¹¹⁷ reported decreased BFV in the MCA on the pain side during CH compared to an attack-free period. In an earlier study,¹¹⁸ bilateral decreased BFV in the MCA was found during CH attacks in patients. However, the attacks were induced by nitroglycerin and the decrease was most probably an effect of the drug¹¹⁹ rather than attack-related changes.

SPECT in TACS

There are relatively few SPECT studies investigating CBF during spontaneous CH or other TACs and to our knowledge none have been performed over the last 20 years (Table 4). The advantage of these headache types is their unilateral nature, which makes it possible to study hemispheric differences, whereas the challenges in performing imaging during attacks are the relative short attack duration (CH:15–180 min, paroxysmal hemicrania (PH): 2–30 min, short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT): 1–600 s) and the fact that TACs patients can be restless or agitated during attacks, resulting in movement artifacts. There are three studies investigating CBF changes during spontaneous attacks compared to the interictal state (Table 4). Sakai et al.²⁵ showed a total CBF increase in three CH patients during attacks compared to the interictal phase and measured extracerebral blood flow, which increased most markedly on the pain side. Kobari et al.⁸⁶ showed that rCBF increased 36–66% bilaterally during CH attacks in al cortical and subcortical regions except the occipital cortex. However, Nelson et al.¹²⁰ showed ictally an 11–28% decrease in right or left hemisphere CBF in 3/4 CH patients, but a 15% increase in 1/4 CH patient, in which the decrease was greater in the dominant hemisphere irrespective of the side of the headache.¹²⁰ There was no apparent correlation between CBF change and intensity of the pain.¹²⁰ The same study also studied the CBF changes following nitroglycerin provocation and found three patients with an increase of 15 to 34%, while two had a 5 to 11% decrease. In comparison, Krabbe et al.¹²¹ performed the largest study so far of induced attacks with 133-Xenon inhalation high-resolution SPECT (200 regions) in eight CH patients and showed that during the headache phase, no significant changes of mean CBF from baseline occurred.¹²¹ The Nelson et al. and Sakai et al. studies used inhalation or intravenous injection of 133-Xenon with external stationary detection, which has a poor spatial resolution, and cannot clearly discriminate extra- from intracerebral blood flow changes. Thus, there is no strong support for changes in rCBF in cortical regions to play a large pathophysiological role in the development of CH attack. In SUNCT, one study has been performed using SPECT and Tc99m hexamethylpropyleneamine oxime (HMPAO) injected intravenously during a bout and between attacks lasting 30–60 s in one patient in which the radiocompound was injected 5 to 10 s after the start of an attack and showed normal tracer uptake and symmetric perfusion during headache periods.¹²²

Table 4.

Studies investigating vascular changes during spontaneous TAC attacks using SPECT and PET.

Authors	Method	Patients	Recordings	Findings
Poughias and Aasly122	SPECT using Tc99m HMPAO i.v.	1 SUNCT patient	During bout and three days without pain	Normal tracer uptake and no differences between sides
Sakai and Meyer25	SPECT 133xenon inhalation	3CH patients	During CH attacks and headache free intervals	A 45% increase in rCBF during attacks
Nelson et al.120	SPECT 133xenon intravenous	4 CH patients	During and after attacks	In 3/4 patients rCBF decreased (range 11 to 28%). 1 patient had an 15% increase in CBF during attack
Kobari et al.86	SPECT 133xenon inhalation	5 CH patients	During attacks (n = 2) and without headache (n = 3)	rCBF increased bilateral during attacks in al cortical and subcortical regions except the occipital cortex (range compared 36–66%)
Matharu et al.125	H215O PET	7 paroxysmal hemicrania patients	4–6 scans in pain and off indomethacin.  3–5 scans pain-free and off indomethacin. Pain-free after indomethacin i.m. randomized	No differences in rCBF during headache compared with the pain-free state when off indomethacin. But, increased rCBF in the contralateral posterior hypothalamus and contralateral ventral midbrain when comparing to patients receiving indomethacin
Sprenger et al.123	H215O PET	1 chronic CH patient with stereotactic electrode implantation in left posterior hypothalamus two months earlier.	During an experimental study the stimulator was turned off and then the patient experienced a spontaneous attack. Three PET scans of 90 s each were obtained during this period and compared to the no pain-phase	The CH attack induced activation increases in the ipsilateral hypothalamus, medial thalamus, and contralateral perigenual ACC compared to no pain-phase

SPECT: single photon emission computer tomography; PET: positron emission tomography; SUNCT: short-lasting unilateral neuralgiform headache attacks; CH: cluster headache; rCBF: regional cerebral blood flow. ACC: anterior cingulate cortex; TAC: trigeminal automatic cephalalgia.

PET in TACs

Two PET studies exist in TACs patients during spontaneous attacks (Table 4). Sprenger et al.¹²³ published a case report of a spontaneous CH attack in a CH patient who two months earlier had underwent stereotactic electrode implantation for deep brain stimulation of the left posterior hypothalamus.¹²³ During a planned experimental study of brain activation, the stimulator was turned off, and the patient had a spontaneous attack in which three PET scans of 90 s each were obtained over 30 min and compared to the no-pain phase. The study found activation increases in the ipsilateral hypothalamus, medial thalamus and contralateral perigenual ACC, which are similar to activation found during nitroglycerin-induced CH attacks.¹²⁴ In paroxysmal hemicranias, Matharu et al.¹²⁵ showed no differences in rCBF during headache compared with the pain-free state when patients were off indomethacin, whereas an increased rCBF in the contralateral posterior hypothalamus and contralateral ventral midbrain was demonstrated when comparing to patients receiving indomethacin treatment. Thus, no pronounced cortical alterations are found in TAC patients interictally using PET.

Studies of spontaneous attacks in tension-type headache

Andersen et al.¹²⁶ investigated rCBF in 41 chronic tension-type headache (TTH) patients during habitual headache using Xenon inhalation or Tc99m HMPAO i.v. SPECT and found normal regional and global CBF when comparing to 33 healthy controls.¹²⁶ In 21 headache-free episodic TTH patients, increased blood flow velocities were measured in the anterior, middle and posterior cerebral, with no significant asymmetries of blood flow velocities in corresponding arteries.¹²⁷ The same group did not observe increased BFV in 20 chronic TTH patients.¹²⁸ Thus, there is no clear evidence of hemodynamic changes in TTH patients.

Conclusion and future directions

This review has gathered the available studies of vascular changes during spontaneous attacks in primary headaches. In migraine, aura symptoms clearly have a pattern of gradually spreading cortical hyperperfusion followed by gradual spreading hypoperfusion as shown by SPECT studies. There is not a pattern in migraine without aura of uniform vascular changes during attacks, but local changes may occur due to neuronal activation. MRA and TCD studies point to a small increase in cerebral vessel diameter during the headache phase, which may be the result of vasoactive peptides being released from perivascular nociceptors or parasympathetic nerve fibers. There is no proof that the vasodilation per se leads to pain. In TACs, there is no evidence of a uniform vascular change during attacks, but focal changes likely reflect neuronal activity in pain pathways. NIRS may be a new tool to further investigate cortical changes, especially during migraine with aura, which may elucidate underlying pathophysiological mechanisms. Moreover, the MR-based arterial spin labeling (ASL) sequence may also provide important information of the cortical blood flow changes in future studies. MRI images may be improved and patients may be scanned during attacks with motion correction.¹²⁹ The MMA has been suspected as the source of pain in migraine using primarily animal models and from surgical experience. To date, no studies have investigated the intracranial segment of the MMA during spontaneous attacks, which would be extremely valuable in future studies.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.