Effect of hypoxia on BOLD fMRI response and total cerebral blood flow in migraine with aura patients

Abstract

Experimentally induced hypoxia triggers migraine and aura attacks in patients suffering from migraine with aura (MA). We investigated the blood oxygenation level-dependent (BOLD) signal response to visual stimulation during hypoxia in MA patients and healthy volunteers. In a randomized double-blind crossover study design, 15 MA patients were allocated to 180 min of normobaric poikilocapnic hypoxia (capillary oxygen saturation 70–75%) or sham (normoxia) on two separate days and 14 healthy volunteers were exposed to hypoxia. The BOLD functional MRI (fMRI) signal response to visual stimulation was measured in the visual cortex ROIs V1–V5. Total cerebral blood flow (CBF) was calculated by measuring the blood velocity in the internal carotid arteries and the basilar artery using phase-contrast mapping (PCM) MRI. Hypoxia induced a greater decrease in BOLD response to visual stimulation in V1–V4 in MA patients compared to controls. There was no group difference in hypoxia-induced total CBF increase. In conclusion, the study demonstrated a greater hypoxia-induced decrease in BOLD response to visual stimulation in MA patients. We suggest this may represent a hypoxia-induced change in neuronal excitability or abnormal vascular response to visual stimulation, which may explain the increased sensibility to hypoxia in these patients leading to migraine attacks.

Introduction

Migraine with aura is a complex neurovascular disorder characterized by recurrent episodes of visual and sensory disturbances.¹ Cortical spreading depression (CSD) is a slowly propagating wave of neuronal and glial depolarization and remains the most likely putative mechanism of migraine aura.² Susceptibility to CSD and to migraine aura involves genetic, vascular and neural mechanisms.^3–5 It has been suggested that abnormal cortical excitability may influence susceptibility to migraine with aura.^6,7 Blood oxygenation level-dependent (BOLD) functional MRI (fMRI) signal provides an indirect measure of brain activity, based on neural activity-dependent blood flow changes.⁸ BOLD studies have provided insights into cortical excitability in migraine with aura.^9–12 Repetitive visual sensory stimulation led to a higher increase in BOLD response in visual cortex of patients during the attack-free period compared to healthy volunteers.^9–12 The authors suggested that the higher increase in BOLD response in migraine with aura patients (MA patients) might reflect increased cortical excitability and thereby susceptibility to CSD.^9–12 A study of patients during visual aura reported dynamic BOLD signal changes with initial increase followed by diminished BOLD signal in response to visual activation.¹³

Recently, we demonstrated that normobaric poikilocapnic hypoxia provoked migraine attacks with aura.¹⁴ Interestingly, hypoxia can induce CSD,^15,16 decrease CSD threshold and increase duration of CSD induced by potassium chloride in mice.¹⁶ In the present study, we investigated the dynamics of the BOLD signal in response to normobaric poikilocapnic hypoxia. We hypothesized that the BOLD response to visual stimulation of visual areas V1, V2, V3, V4, and V5 during hypoxia is different in migraine patients compared to healthy volunteers. To test this hypothesis, we performed a randomized double-blind sham-controlled study in migraine with aura patients and healthy volunteers.

Material and methods

We recruited 15 patients who suffered exclusively from migraine with typical visual aura fulfilling the International Classification of Headache Disorders 3 beta version criteria¹ with a minimum of one attack per month. In addition, we included gender and age (±3 years) matched healthy controls with no history of migraine and no first-degree relatives with migraine. We used the following exclusion criteria for all participants: any other type of headache (except episodic tension type headache less than five days per month), any somatic or psychiatric disease, any daily medication apart from oral contraceptives, smoking and a history of mountaineering training. We recruited participants via the outpatient clinic at the Danish Headache Center (Rigshospitalet Glostrup, Copenhagen, Denmark) and via a Danish website for recruitment of volunteers to health research (www.forsoegsperson.dk).

The study was approved by the Ethics Committee of the Capital Region of Denmark (H-4-2012-182) and the Danish Data Protection Agency. The study was registered at Clinicaltrials.gov (NCT01896167). All participants provided their written informed consent to participate in the study after detailed oral and written information about the study in accordance with the Declaration of Helsinki 1964, with later revisions.

Data on hypoxia induced migraine, MR spectroscopy, MRI angiography, vital variables, and venous blood samples content in this study have previously been published.¹⁴

Experimental design

Using a double-blind, crossover design, the patients were randomly allocated to 180 min of inhaling normobaric poikilocapnic hypoxia resulting in capillary oxygen saturation (SpO₂) of 70–75% or atmospheric air (sham) on two separate days with a minimum of seven days between. The healthy volunteers were exposed to a similar hypoxic challenge on one study day (Figure 1). Hypoxia and sham were induced by an AltiTrainer system (SMTEC, Nyon, Switzerland) through a 7-m tube, a one-way valve and a tight fitting full-face mask (Hans Rudolph, Shawnee, Kansas, USA).¹⁴ One investigator (JB) and the participants were blinded to the content of the inhaled air. The participants continuously breathed through the mask during 0–180 min including during the MRI scans by the use of extended tubes. SpO₂ was measured continuously using a fingertip pulse oximeter (Veris Monitor, Medrad, USA).

Figure 1.

MRI scan flow chart.

We obtained 3.0T MRI imaging (BOLD fMRI, Phase-Contrast Mapping (PCM) and anatomical T1 images) at baseline, during and after inhalation of test gas (Figure 1).

Participants were offered a standard small meal (bread with soft cream cheese, grapes, banana and water) after the baseline scan and after 180 min. Participants could choose between supine or sitting position during the first 80 min of inhalation of the test gas. Participants were discharged after the last scan and instructed to complete a validated headache diary for the next 8 h.

All participants arrived headache free at the laboratory at the same time on each study day (±2 h). Arrival time for patients and healthy volunteers was individually matched (±2 h). Participants were not allowed to consume coffee, tea, cocoa, alcohol, or other methyl xanthine-containing food or beverage for at least 12 h prior to the start of the study days. If the participant had had a migraine attack within five days or any headache 48 h prior to the study days, the study was postponed. Female participants did not menstruate within two days before and after the study days.

BOLD fMRI

MR scans were performed on a 3.0T Philips Intera Achieva MRI scanner (Philips Medical Systems, Best, The Netherlands) using a 32-element phase array receive head coil. Functional imaging was performed with a gradient-echo planar imaging sequence (32 slices of 4.0 mm thickness; slice gap 0.1 mm; field of view 230 mm × 230 mm; in-plane acquired resolution 2.9 mm × 2.9 mm; repetition time 3.0 s; echo time 35 ms; flip angle 90°; and SENSE (SENSitivity Encoding factor 2). To ensure steady-state longitudinal magnetization, dummy scans (two volumes) were performed. Participants were presented to visual stimulation using OLED video goggles (Figure 2(a)) (NordicNeuroLab, Bergen, Norway; SVGA, 800 × 600 pixels, refresh rate 85 Hz, field of view 30°horizontal, 23°vertical, stimulus luminance: 70–110 cd/m²). The system was connected to a control computer outside the scanner room by a fiber optic cable. The visual stimulation paradigm was a block-design with alternation of stimulation and rest blocks each comprising 18 s. To drive large expanses of visual cortex, a salient high contrast motion stimulus was used, i.e. a moving black and white dartboard pattern (diameter: 22° [circular aperture]; ring width: 0.6°; spoke width: 15°; patterns in each spoke moved in opposite directions, alternately inward and outward, with random changes of the motion direction approx. every 2–3 s).^17,18 Freely available Matlab-based software (Mathworks, Natick, MA, USA) was used to generate the stimulus. A complete scan comprised 32 18 s blocks and lasted 576 s. The participants were instructed to focus on a central fixation point during the entire scan. The onset of visual stimulation was triggered by the scan acquisition. The lighting conditions in the scanner room were identical during each scan.

Figure 2.

MRI methods. (a) The BOLD response to visual stimulation was measured by fMRI. Screen shot of the applied visual stimulus and example of BOLD response to visual stimulation in one patient. (b) Cerebral blood flow was measured by phase-contrast measurements. Anterior-posterior and lateral view of the placement of the measurement slices on a 3D TOF angiography in one patient. Three scans were carried out to obtain velocity maps perpendicular to the carotids and the basilar artery. Regions of interest (white contours) of the left and right carotid and the basilar artery are shown.

Regions exhibiting significant stimulus-correlated changes in blood BOLD signal were identified using FMRIB's Software Library (FSL) version 4.1.6 for Mac OS.¹⁹ FSL FLIRT linear registration was used to register functional data to brain extracted T1-weighted high-resolution scans and to the MNI152 standard space. The registration from high resolution structural to standard space was further refined using FSL FNIRT nonlinear registration.

First-level analysis was performed using FSL FEAT (FMRI EXPERT Analysis Tool). Data pre-processing included slice time correction, spatial smoothing (FWHM 5 mm), high pass filtering (cut-off 36 s), head motion correction by FSL MCFLIRT and brain extraction of functional and anatomical images by FSL BET. Voxel-wise analysis was performed using a general linear modeling approach, with local autocorrelation correction, of eight regressors (main stimulus (block design convolved with a canonical single gamma hemodynamic response function), a temporal derivative and six motion regressors). Followingly, t-statistic maps were calculated for the parameter estimates (beta values) of the main stimulus regressor. Z (Gaussianized t) statistic images were thresholded using clusters determined by Z > 2.3 and a corrected cluster significance threshold of P < 0.05 (using a distribution based on Gaussian Random Field Theory).

Subsequently, values (see below) from the early visual areas (V1-V5) ROIs were extracted from the previously calculated FSL-FEAT results using featquery (a part of the FSL software package). V1–V5 ROIs were derived from the Juelich histological (cyto- and myelo-architectonic) atlas.¹⁷ The values used for calculations were the mean main stimulus parameter estimates (i.e. beta values) expressed as percentual changes from baseline periods to visual stimulation periods (hereafter simply referred to as “BOLD response”).

Total cerebral blood flow

Changes in brain hemodynamics can alter the BOLD response.²⁰ To control that any group differences could not be explained by baseline or hypoxic-induced cerebral blood flow differences, we measured total cerebral blood flow (tCBF).

Total CBF was calculated by measuring the blood velocity in the internal carotid arteries and the basilar artery using phase-encoding technique.²¹ Initially, a 3D TOF angiography was acquired and used to place the measurement slices orthogonal to the carotid arteries and the basilar artery in three consecutive scans (Figure 2(b)). The imaging parameters for the measurement were: 1 slice, field of view = 240 × 240 mm², voxel size = 0.75 × 0.75 × 8 mm³, echo time = 7.3 ms, repetition time = 12.5 ms, flip angle = 10 degrees, seven repeated measures, total scan time = 1 min 17 s, non-triggered, velocity encoding = 100 cm/s). To reduce scan time, cardiac triggering was not applied. tCBF was calculated by measuring the mean velocity and cross-sectional area of each of the three feeding arteries by drawing regions of interest (ROI) corresponding to each artery (Figure 2(b)). For each scan, ROIs were drawn on the first measurement and subsequently copied to the following measurements and inspected for misalignment from possible motion and corrected by moving the ROI when necessary. To attain quantitative physiological global CBF values in mL/100 g/min, the total flow of the three arteries was normalized to whole-brain tissue weight. Processing was done using in-house Matlab scripts. Data were analyzed by two investigators (NA and MV), who were blinded to the experimental day and scan session.

Anatomical scan

High-resolution anatomical scans were obtained with a 3D T1-weigthed turbo field echo sequence (field of view = 256 × 256 × 170 mm³; voxel size = 1.1 × 1.1 × 1.0 mm³; echo time = 4.6 ms; repetition time = 9.0 ms; echo train length = 200; flip angle = 8°). A whole-brain tissue mask including cerebral hemispheres (excluding the ventricles), cerebellum, and the brainstem was made using the FSL BET and FAST tools (FMRIB Software Library, Oxford University, Oxford, UK).²² A brain tissue density of 1.05 g/mL was assumed for calculation of brain weight.²³ To exclude any pathology, all anatomical scans was examined by a radiologist (FW).

Statistical analysis

Data are presented as mean with 95% confidence interval (CI). Baseline was defined as t₀ before the start of inhalation.

Sample size was based on previous migraine-provocation studies carried out by our group reporting migraine incidence of 60–80% after pituitary adenylate cyclase activating peptide-38, calcitonin gene-related peptide, and glyceryl-trinitrate.²⁴

The primary endpoints were differences between patients and healthy volunteers on hypoxia day and secondary endpoints were differences between hypoxia and sham in MA patients for the following data:

• Changes from baseline in mean BOLD response in the visual cortex area ROIs V1-V5 at 180 min (during hypoxia/sham) and 240 min (60 min after end of hypoxia/sham) in the visual cortex area ROIs V1-V5.

• Changes from baseline in tCBF between hypoxia and sham at 120 min (during hypoxia/sham) and 220 min (40 min following hypoxia/sham)

Changes from baseline and baseline differences of all variables between hypoxia and sham were compared using the non-parametric Wilcoxon signed rank test. Difference between patients and healthy volunteers were compared by the Mann–Whitney test.

Analyses were performed with SPSS version 23.0 (Chicago, IL, USA). We did not adjust for multiple analyses. Level of significance was accepted as 5% (P < 0.05).

Results

Fifteen migraine patients with aura (12 females, mean age 28 (range 18–46), mean body mass index 22 kg/m² (range 19–26)) and 14 healthy volunteers (11 females, mean age 28 (range 20–45), mean body mass index 25 kg/m² (range 20–34)) completed both study days (Figure 1).

ROI-analysis showed no side-to-side (right vs. left) difference in the BOLD response change from baseline to hypoxia. We therefore used an average of the BOLD response from right and left-sided ROIs for the further analyses to increase signal-to-noise ratio.

At baseline and 60 min after the end of the hypoxia/sham inhalation, we found no differences in BOLD response or tCBF between the two experimental days in MA patients (baseline to post hypoxia/sham) (Table 1 and Supplementary Table 1) or between patients and healthy volunteers (baseline to post hypoxia) (Table 1). In MA patients, the BOLD response decreased significantly and tCBF increased significantly during hypoxia compared to sham (Table 1 and Supplementary Table 1).

Table 1.

fMRI BOLD response and total cerebral blood flow on hypoxia day in patients and healthy controls.

		Baseline	P	Hypoxia	P	Post hypoxia	P
V1	Healthy	2.17 (1.68–2.65)	0.880	0.95 (0.63–1.28)	0.006	2.34 (1.81–2.86)	0.936
	Patients	2.20 (1.70–2.70)		0.70 (0.44–0.97)		2.71 (2.14–3.28)
V2	Healthy	2.04 (1.55–2.53)	0.914	0.90 (0.59–1.21)	0.023	2.20 (1.71–2.69)	0.979
	Patients	2.07 (1.60–2.54)		0.67 (0.43–0.91)		2.54 (2.01–3.07)
V3	Healthy	2.37 (1.85–2.88)	0.780	1.06 (0.75–1.36)	0.044	2.60 (2.14–3.06)	0.501
	Patients	2.38 (1.91–2.86)		0.85 (0.59–1.11)		2.84 (2.30–3.38)
V4	Healthy	2.04 (1.58–2.51)	0.914	0.98 (0.62–1.33)	0.037	2.24 (1.83–2.64)	0.609
	Patients	2.03 (1.64–2.43)		0.78 (0.57–1.00)		2.37 (1.91–2.83)
V5	Healthy	0.89 (0.60–1.18)	0.683	0.49 (0.09–0.88)	0.132	0.98 (0.70–1.26)	0.893
	Patients	1.01 (0.78–1.23)		0.38 (0.30–0.47)		1.19 (0.91–1.47)
Total CBF	Healthy	58.7 (53.2–64.2)	0.533	76.9 (69.3–84.4)	0.705	55.4 (49.9–61.8)	0.851
	Patients	55.6 (49.7–61.4)		70.2 (62.3–78.1)		51.0 (45.5–56.5)

Note: Mean (CI) BOLD response (%) in V1–V5 and total cerebral blood flow (ml min⁻¹ 100g⁻¹). P-values are for comparison of patients and healthy volunteers at baseline and absolute changes during and after hypoxia (Mann–Whitney test). The P-values marked in bold indicate statistical significance. Missing values are shown in Figure 1.

fMRI BOLD: functional MRI blood oxygenation level-dependent.

Eight (52%) MA patients reported migraine-like attacks during hypoxia compared to one (7%) during sham (P = 0.039). Three patients reported aura during hypoxia and four patients reported possible aura (three during hypoxia, one after hypoxia). No patients reported aura during sham (normoxia), one patient reported aura after sham, as previously published.¹⁴

BOLD response during hypoxia in visual cortex and tCBF

We found a larger decrease in BOLD response in V1 (P = 0.006), V2 (P = 0.023), V3 (P = 0.044) and V4 (P = 0.037) in MA patients compared to healthy volunteers from baseline to hypoxia, but no difference in V5 (P = 0.132) (Table 1, Figure 3). There were no differences between patients and healthy volunteers in the changes of total CBF during hypoxia (P = 0.705) (Table 1).

Figure 3.

Mean absolute changes (95% CI) from baseline to hypoxia in BOLD response in V1-V5 and tCBF. *P-values are for comparison of absolute changes from baseline to hypoxia in: patients and healthy volunteers (P-values shown in Table 1); patients who developed migraine-like attack and healthy volunteers (V1: P = 0.048, V2: P = 0.130, V3: P = 0.328, V4: P = 0.328, V5: P = 0.510, tCBF: P = 0.718); patients with no attack and healthy volunteers (V1: P = 0.018, V2: P = 0.045, V3: P = 0.032, V4: P = 0.024, V5: P = 0.090, tCBF: P = 0.841); patients who developed migraine-like attack and patients with no attack (V1: P = 0.937, V2: P = 0.329, V3: P = 0.247, V4: P = 0.126, V5: P = 0.082, tCBF: P = 0.095).

In explorative analyses, we found larger decrease in BOLD response in V1 (P = 0.048), but no difference in V2–V5 (V2: P = 0.130; V3: P = 0.328; V4: P = 0.328; V5: P = 0.510) in MA patients who developed attacks compared to healthy volunteers (Figure 3). Patients who did not develop migraine attack also had a larger decrease in BOLD response during hypoxia in V1 (P = 0.018), V2 (P = 0.041), V3 (P = 0.032), V4 (P = 0.024) (except for V5, P = 0.090) compared to healthy volunteers (Figure 3). We found no difference in BOLD response in V1–V5 during hypoxia in MA patients who developed migraine-like attack and patients who did not develop an attack (Figure 3). Explorative post hoc comparison analysis of tCBF showed no statistical differences between these subgroups (Figure 3).

Vital variables

Changes in vital variables have previously been published.¹⁴ Briefly, in both patients and controls, we found a similar hypoxia-induced increase in heart rate, decrease in SpO2 and end-tidal CO2 tension, and unchanged mean arterial pressure. In addition, we have now conducted explorative analysis that reveals no difference between patients and controls in hypoxia-induced changes of vital variables during the fMRI BOLD scan (Supplementary Table 2).

Discussion

The major outcome of the present study is that hypoxia induced smaller BOLD responses in the visual cortex in migraine with aura patients compared to healthy volunteers. We found no difference in hypoxia-induced total CBF increase between patients and healthy volunteers. In the following, we summarize previous major findings on cerebral hemodynamics during hypoxia and discuss possible mechanisms behind altered BOLD responses in MA patients during hypoxia.

BOLD fMRI responses and cerebral hemodynamics during hypoxia

Similar to our findings, previous studies showed a decrease of the BOLD response to visual stimulation during severe hypoxia in healthy volunteers.^25–27 The BOLD response in V1 decreased by 72% (CI 66–78%) during hypoxia in patients and by 51% (CI 35–67%) in controls. Rostrup et al.²⁶ (2005) found a similar average BOLD response reduction of 63% during severe hypoxia (FiO2 10%). The arterial oxygen saturation responses to hypoxia with FiO2 10% are very heterogenous varying from 60 to 90%.^28,29 Instead of a fixed FiO2, we regulated FiO2 to obtain and maintain a capillary saturation of 70–75% (mean 72% SD 4.76%).¹⁴

The BOLD signal originates from the paramagnetic properties of deoxyhemoglobin in the blood. Under normoxic conditions, during activation (e.g. visual stimulation), the regional blood flow increases disproportionally more than the extraction of oxygen causing decreased deoxyhemoglobin and increased BOLD signal. During hypoxia, the relative increase of deoxyhemoglobin concentration in the vascular compartment likely affects both baseline and stimulus-induced BOLD signal changes due to the inherent paramagnetic property of deoxyhemoglobin.^20,26,30 Furthermore, the concentration of deoxyhemoglobin depends on a range of physiological parameters, including CBF, cerebral blood volume (CBV) and cerebral metabolic rate of O2 (CMRO2).³¹ During severe hypoxia, these parameters are altered and affect the BOLD signal. The hypoxia-induced increase in CBF (33%) in our study is comparable to earlier studies with the same severity of hypoxia using Kety–Schmidt method (FiO2 10%)^32,33 and PET (SpO2 74%)³⁴ in which CBF increase ranges from 35% to 36%.³⁵ The cause for the decreased BOLD response during severe hypoxia is likely due to the combined effects of arterial deoxygenation and hypoxia-induced global hyperperfusion diminishing the reserve flow capacity and causing the increase in perfusion in response to activation to be lessened.^25–27 Hyperperfusion from administration of acetazolamide or hypercapnia has shown similarly diminished BOLD response to visual stimulation.^36–38 The effect of severe hypoxia on CBV or CMRO2 is more scarcely investigated than CBF, but one study has reported on CBV during severe hypoxia (10 min exposure, FiO2 10%) and found a small increase of 5.2%.³⁹ Experimental studies reported unchanged CMRO2 during severe hypoxia (FiO2 range: 7–10%; exposure range 15–40 min)^32,40,41 However, a recent MRI study (2016)²⁹ reported an increase in CMRO2 of 8.5% during 40 min of hypoxia with FiO2 10% (n = 23). Overall, the effect from severe hypoxia on CBV and CMRO2 is minor and the effect of global hyperperfusion and arterial deoxygenation is likely the most important causes for the diminished BOLD response. It should be mentioned that several studies have reported reduced BOLD response during mild hypoxia (SPO2 > 80%) without any change CBF.^42–44 Thus, we cannot compare studies using mild and severe hypoxia due to different evoked physiological compensatory response and likely different mechanism behind BOLD response changes.

Possible mechanisms behind decreased BOLD responses in patients during hypoxia

The reason for an altered effect of hypoxia on the BOLD response in migraine patients with aura is not clear. While the effects of migraine triggers on the BOLD response in migraine patients has not been investigated, the migraine-inducing agents sildenafil, glyceryl trinitrate, and CGRP did not affect the BOLD response to visual stimulation in healthy volunteers.^38,45,46

The diminished BOLD response in patients during hypoxia is unlikely to be explained by differences in arterial deoxyhemoglobin or global perfusion since a similar SpO2 and tCBF was observed in patients and healthy volunteers. However, we cannot rule out regional CBF differences in the occipital cortex in MA patients and healthy volunteers since only global CBF was measured. Recently, theoretical discussion of previously reported data suggested that migraine with aura patients may have an impaired ability to homogenize capillary flow pattern during a metabolic challenge causing increased capillary transit time heterogeneity (CTH).⁴⁷ Hypoxia may theoretically cause an increase in CTH in MA patients in the visual cortex, leading to a higher increase in flow to compensate for a reduced oxygen extraction efficacy. This would further decrease the flow reserve capacity and thus decrease the BOLD response. This could also explain the increased BOLD response during different types of visual stimulation in migraine aura in the interictal phase.^9–12,47,48 These data are, however, heterogeneous and may be stimulus-dependent. One study found increased responses in a visual-driven functional network outside the occipital cortex using the same visual stimulus as used in the present study.⁴⁸ In present study, we found no differences at baseline (before hypoxia) in the BOLD response to visual stimulation between MA patients and healthy volunteers.

The level of hypoxia used in present study caused hypocapnia,¹⁴ which has been shown to increase neural excitability.^49,50 There were no difference in hypoxia-induced decrease of end-tidal CO₂ in MA patients and healthy volunteers.¹⁴ However, an increased sensitivity to hypocapnia in migraine with aura patients has been suggested based on studies showing a larger decrease in blood velocity in the middle cerebral artery during hyperventilation-induced hypocapnia in aura patients compared to healthy volunteers⁵¹ and migraine without aura patients.⁵² Thus, hypoxia-induced hypocapnia may cause a further increased neural excitability in aura patients and thus increased neuronal response to stimulation. An increased neuronal response increases the extraction of oxygen from the blood, but likely not the CBF due to the diminished reserve flow capacity during hypoxia, thus causing an increase in deoxyhemoglobin and diminished BOLD response. An increased neuronal reactivity to transient hypoxia in MA patients has been suggested in a study where transient cerebral hypoxia was induced by intravenous injections of air microbubbles in MA patients with patent foramen ovale (PFO) and healthy controls with PFO.⁵³ During passage of microbubbles in the cerebral arteries, spectral electroencephalogram (EEG) analysis revealed a significant increase in electroencephalographic power in MA patients but not in controls. The effect of general hypoxia on EEG is not clear. One study found no EEG changes during poikilocapnic hypoxia (SPO2 75%) in healthy volunteers.⁵⁴ It would be interesting to compare the EEG response during general hypoxia in MA patients and healthy volunteers.

Explorative analysis in the present study revealed no differences in BOLD or total CBF response to hypoxia in those who developed attack and those who did not. We found significant difference in BOLD response changes between patient sub-groups and controls (Figure 3). This indicates that the effect of hypoxia on the BOLD response in aura patients is robust, and not only affecting a subgroup of patients. If this represents preictal changes in cortical excitability, it is possible that longer duration of hypoxia could have provoked aura attacks in all patients.

Strength and limitations of the study

The study included a well-defined and relatively large group of patients suffering from migraine with visual aura in all attacks, taking no preventive medication and suffering from no other illness, and a well-matched control group. Furthermore, the study was performed using high field strength MRI (3 T).

PCM is a validated method to measure global CBF.²¹ However, it is a limitation of the study that we did not measure regional CBF. We cannot, therefore, exclude hypoxia-induced regional differences in CBF, particularly in the occipital cortex. In addition, we did not measure CBV and CMRO2. As discussed, the effects of hypoxia on CBV and CMRO2 in healthy volunteers are minor. However, whether the impact of hypoxia on these parameters in migraine with aura patients is different needs to be further investigated.

Conclusion

In conclusion, hypoxia induced a greater decrease in BOLD response to visual stimulation in migraine with aura patients compared to healthy volunteers, which may explain the increased sensibility to hypoxia in these patients leading to migraine attacks. We suggest, this may represent a hypoxia-induced changed neuronal excitability or abnormal vascular response to visual stimulation in the visual cortex in migraine with aura patients. The mechanisms behind hypoxia-induced changes in BOLD response in migraine patients need to be further investigated, especially the possible role of changes in regional CBF, CBV, CMRO2 and oxygen extraction fraction.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was supported by the Capital Region of Denmark Foundation for Health Research (A4620), the Lundbeck Foundation (R155-2014-171), the Novo Nordic Foundation (NNF11OC1014333), the Augustinus Foundation (13-3794), Det Frie Forskningsråd (DFF-4004-00169B), Simon Fougner Hartmanns Familiefond, the European Union’s Seventh Framework programme (2007–2103) under grant agreement no. 602633.

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.

Authors’ contributions

NA contributed to study design; protocol development; participant enrolment; data acquisition, data processing, analysis, statistics, and interpretation; and drafting and revision of the paper. AH analyzed the fMRI BOLD data and contributed to drafting of the paper. HWS contributed to study design, protocol development and statistical tests. MV analysed the CBF data and contributed to drafting of the paper. JB contributed to participant recruitment, data acquisition and assisted during scanning procedures. MA initiated the study and contributed to study design, protocol development, statistics, data interpretation, and drafting and revision of the paper. AH, HWS, MBV, JB, FMA, KSO, HBWL, JO were involved in design of the study, interpretation of data and critical revision of the manuscript. All authors have given final approval of the version to be published.

Supplementary material

Supplementary material for this paper can be found at http://journals.sagepub.com/home/JCB