Microstructural white matter changes preceding white matter hyperintensities in migraine

Abstract

Objective

We used magnetization transfer imaging to assess white matter tissue integrity in migraine, to explore whether white matter microstructure was more diffusely affected beyond visible white matter hyperintensities (WMHs), and to explore whether focal invisible microstructural changes precede visible focal WMHs in migraineurs.

Methods

We included 137 migraineurs (79 with aura, 58 without aura) and 74 controls from the Cerebral Abnormalities in Migraine, an Epidemiological Risk Analysis (CAMERA) study, a longitudinal population-based study on structural brain lesions in migraine patients, who were scanned at baseline and at a 9-year follow-up. To assess microstructural brain tissue integrity, baseline magnetization transfer ratio (MTR) values were calculated for whole brain white matter. Baseline MTR values were determined for areas of normal-appearing white matter (NAWM) that had progressed into MRI-detectable WMHs at follow-up and compared to MTR values of contralateral NAWM.

Results

MTR values for whole brain white matter did not differ between migraineurs and controls. In migraineurs, but not in controls, NAWM that later progressed to WMHs at follow-up had lower mean MTR (mean [SD] 0.354 [0.009] vs 0.356 [0.008], p = 0.047) at baseline as compared to contralateral white matter.

Conclusions

We did not find evidence for widespread microstructural white matter changes in migraineurs compared to controls. However, our findings suggest that a gradual or stepwise process might be responsible for evolution of focal invisible microstructural changes into focal migraine-related visible WMHs.

We previously identified migraine as a risk factor for subclinical focal deep white matter hyperintensities (WMHs).^1,2 The etiology of these lesions remains to be clarified.

In migraineurs, diffuse invisible white matter changes may be present that extend beyond the visible focal WMHs on conventional MRI. Magnetization transfer imaging (MTI) is an MRI technique that provides quantitative information on microstructural tissue integrity. MTI detects structural changes both in areas with abnormal signal intensity on conventional MRI and in normal-appearing brain tissue.^3,4 Magnetization transfer ratio (MTR) values reflect the proportion of exchange between free water protons and water protons bound to macromolecules (myelin, proteins, and cell membrane molecules). Reduced MTR values suggest lower macromolecular content or microscopic edema, indicating microstructural changes.⁵ Brain parenchyma in migraineurs has scarcely been studied with MTI^6–8 and indeed suggested migraine-related focal microstructural damage in some studies.^6,7 However, another study found no difference in MTR for whole brain and normal-appearing white matter (NAWM) in migraineurs compared to controls.⁸ These findings may have been biased by investigating more severe migraine phenotypes. Hence, we examined whole brain white matter integrity in migraineurs from the general population.

The 9-year follow-up data of the Cerebral Abnormalities in Migraine, an Epidemiological Risk Analysis (CAMERA) study revealed that the WMHs in female migraineurs were progressive also in those who no longer had migraine activity during follow-up.² Whether WMHs in migraineurs develop acutely or in a progressive way needs further assessment. Therefore, we investigated whether baseline MTR might reveal invisible brain changes in NAWM at sites that had changed to visible WMHs at follow-up.

Methods

Study population

Participants were included from the Maastricht subpopulation of the CAMERA-1 and -2 studies, a longitudinal population-based MRI study on structural brain lesions in migraine patients. Characteristics of the study population and the assessment of migraine have been described in detail elsewhere.^1,2,9 The original participants of the CAMERA-1 study included 295 well-characterized migraineurs and 140 controls, divided into 2 subpopulations from the Dutch cities of Doetinchem and Maastricht. The Maastricht subpopulation of this study consisted of 213 participants (n = 80 migraine with aura, n = 58 migraine without aura, n = 75 controls). For 211 of these 213 participants, the MRI protocol included MTI at baseline. The participants included in current analyses were more likely to smoke than those participants who were excluded because of missing MTI data (table 1). Further demographics and clinical characteristics were comparable between groups. A total of 128 participants of the original Maastricht subpopulation (n = 55 migraine with aura, n = 35 migraine without aura, n = 38 controls) participated in CAMERA-2, a 9-year follow-up study.² Reasons for participants not to participate in CAMERA-2 included inability to recontact the participant due to relocation, loss to civil registry information, no interest, inability to visit the research center, claustrophobia, non-neurologic illness, and death. The MRI protocol did not include MTI at follow-up.

Table 1.

Baseline characteristics of included and excluded Cerebral Abnormalities in Migraine, an Epidemiological Risk Analysis (CAMERA)–1 study participants for current magnetization transfer ratio substudy

Standard protocol approvals, registrations, and patient consents

The study protocol was approved by the Leiden University Medical Center ethics committee. All participants gave written informed consent.

Magnetic resonance imaging

Fast field echo MTI (repetition time [TR] 106 ms, echo time [TE] 5.9 ms, 28 axial 5-mm slices, 256 × 256 acquisition matrix, field of view 220 mm, flip angle 12°, in-plane resolution 0.86 × 0.86 mm) images were acquired on a 1.5T MRI scanner (Philips Gyroscan Intera ACS-NT, Best, the Netherlands) at baseline. MTI comprised 2 consecutive sequences, one without (M0) (resulting in proton density contrast) and one with (Ms) radiofrequent saturation pulse (1,100 Hz upfield of H₂O resonance). In addition, dual echo T2 (TR 3,000 ms, TE 27–120 ms, echo train length 10) and fluid-attenuated inversion recovery (FLAIR) (TR 8,000 ms, TE 100 ms, inversion time 2,000 ms, echo train length 19) images were acquired to check images for structural abnormalities and to be able to segment WMHs at baseline and after a 9-year follow-up, using the same MRI scanner and protocols.

Image postprocessing

MTI data were postprocessed using FSL (FMRIB Software Library, FMRIB Center, Oxford, UK)¹⁰ and ELASTIX.¹¹ First, M0 images were linearly registered to the Ms images using FMRIB's linear image registration tool (FLIRT).¹² Non-brain tissue was removed from M0 images using brain extraction tool¹³; a binary mask created in this processing step was applied to remove non-brain tissue from Ms images. The MTR was then calculated by the equation (M₀ − M_s)/M₀. The MTR images were linearly registered to the Montreal Neurological Institute (MNI) 152 stereotactic standard space implemented in FSL using FLIRT again. To optimize image registration in the white matter around the ventricles, these normalized MTR images were registered to MNI152 space once again using standard measures in ELASTIX. The MNI152 template was segmented into gray matter, white matter, and CSF binary masks using FAST.¹⁴ MTR histogram measures (mean and normalized peak height) were retrieved for white matter for all participants by overlaying the binary masks on the normalized MTR images. The mean MTR reflects the average MTR in a region of interest (i.e., white matter in this case). The peak height of the MTR histogram shows the number of voxels with the most common MTR value and reflects the uniformity of the underlying voxels in terms of MTR values. As the peak height depends on the total number of voxels, normalized peak height is calculated by dividing the number of voxels with the most common MTR by the total number of voxels in the region of interest. To decrease influence of extreme outliers, only MTI measures within 3 SD of the mean were included in the statistical analysis.

To study the baseline MTR of brain tissue that developed into WMHs at follow-up, supratentorial WMHs were segmented semi-automatically as hyperintense lesions on proton density, T2, and FLAIR images (figure) both at baseline and follow-up using QBrain 1.1.² As we had a particular interest in the development of deep WMHs, which were more prevalent and more progressive in (female) migraineurs in CAMERA-2,² we differentiated between deep and periventricular WMHs. WMHs attached to the lateral ventricles were classified as periventricular WMHs. For follow-up analyses, we included all study participants with deep WMH progression, defined as increase in size or number of WMHs, or both. Segmented deep WMHs at baseline and at follow-up were registered to normalized MTR images in MNI152 space using FLIRT. Mean MTR was computed for deep WMHs at baseline and for tissue that had progressed to deep WMH at follow-up. As deep WMHs often occur as asymmetric, punctate lesions, we considered that white matter contralateral to the deep WMHs could serve as NAWM for within-participant comparisons. MTR values of this contralateral white matter were computed by per slice mirroring deep WMH maps about the sagittal axis.

Figure. Fluid-attenuated inversion recovery images of one participant.

Example of deep white matter hyperintensities (WMHs) at a similar level shows deep WMHs at baseline (A, long thin arrows) and progressive and new deep WMHs at follow-up (B, short thick arrows).

Statistical analyses

Demographic characteristics were compared applying one-way analyses of variance (normal distribution), nonparametric tests (skewed distribution), and Fisher exact tests (Statistical Package for Social Science 20.0; SPSS Inc., Chicago, IL). The primary analysis of MTR histogram measures of whole brain white matter and NAWM comprised the comparison among migraineurs, migraine subgroups, and controls for the whole population that underwent MTI (n = 211) using general linear models adjusting for age, sex, hypertension, diabetes, systolic and diastolic blood pressure, total cholesterol, high-density lipoprotein (HDL) cholesterol, and body mass index (BMI).¹⁵

To decrease the effect of possible misregistration between conventional T2 and MTI images, which is theoretically largest for the smallest WMHs, paired sample t tests comparing mean MTR of deep WMHs and contralateral NAWM were weighted for the deep WMH volume. Mean MTR of deep WMHs was also compared between migraine (subgroups) and controls using general linear models adjusting for age, sex, hypertension, diabetes, systolic and diastolic blood pressure, total cholesterol, HDL cholesterol, and BMI. p Values <0.05 were considered statistically significant. An explorative subanalysis was performed for subgroups of migraineurs based on disease activity (>1 year free of attacks [inactive] vs at least 1 attack within the last year [active]).

Data availability statement

Anonymized data will be shared by request from any qualified investigator, only for purposes of replicating procedures and results.

Results

MTI data at baseline were available for 137 migraineurs (n = 79 with aura; n = 58 without aura) and 74 controls. Demographics and clinical characteristics were similar between migraineurs and controls, except for a higher BMI in migraineurs (mean [SD] 25.7 [4.8]) vs controls (24.3 [3.3], p = 0.02; table 2), particularly in migraineurs with aura (26.2 [5.1], p = 0.005).

Table 2.

Baseline Cerebral Abnormalities in Migraine, an Epidemiological Risk Analysis (CAMERA)–1 characteristics of CAMERA magnetization transfer ratio study participants

Whole brain analyses

Mean MTR and normalized MTR peak height in baseline whole brain white matter (including areas with WMHs at baseline and follow-up) and NAWM (excluding areas with WMHs at baseline and follow-up) did not differ between migraineurs (or subgroups of migraine patients) and controls (table 3).

Table 3.

Volumes and magnetization transfer ratio (MTR) histogram measures for whole brain white matter, deep white matter hyperintensities (WMHs), and contralateral deep white matter

Focal WMHs analyses

At 9-year follow-up, 49 migraineurs (29 with aura, 20 without aura) and 19 controls had increased deep WMH volume due to new or progressive lesions (table 3). In migraineurs, areas that had undergone transition to deep WMHs at follow-up had lower baseline mean MTR (0.354 [0.009] vs 0.356 [0.008], p = 0.047), compared to their own contralateral white matter. This contralateral difference was not seen in controls. Baseline MTR in areas that had progressed to WMHs at follow-up was not lower than in contralateral white matter in subgroups of active (0.354 [0.008] vs 0.356 [0.009], p = 0.12) and inactive (0.353 [0.011] vs 0.356 [0.006], p = 0.21) migraineurs. At the sites of deep WMHs present at baseline, mean MTR was lower in migraineurs compared to controls (0.355 [0.008] vs 0.358 [0.012], p = 0.048). Baseline mean MTR of areas that progressed to WMHs at follow-up in migraineurs (0.354 [0.009]) was not significantly lower than in similar areas in controls (0.357 [0.010]; p = 0.08).

Discussion

In this study using MTI in migraineurs from the general population, we found no significant differences in baseline whole brain white matter MTR values between migraine patients and controls. Thus, in contrast to 4 studies using diffusion-weighted MRI^16,17 and MTI,^6,7 but in line with another MTI study,⁸ compared to controls, we did not find evidence for more diffuse microstructural changes in the white matter of migraineurs from the general population. A possible explanation might be that compared to the clinic-based patients in previous studies, our participants likely had less severe migraine. They also were less exposed to the potentially confounding effects of prophylactic and abortive medications and concomitant anxiety and depressive disorders, factors that might influence brain architecture as well.^18,19

In migraineurs, we found that mean MTR at baseline was significantly lower in normal-appearing areas on conventional T2-weighted images at baseline that progressed to deep WMHs at 9-year follow-up. This suggests the presence of focal occult microstructural alterations or damage in brain tissue integrity prior to the appearance of visible deep WMHs on conventional T2-weighted MRI. This difference was not obviously explained by differences between migraineurs who had active migraine and those with inactive disease (with most being attack-free for >5 years), which is in line with our previous findings of WMHs being progressive in female migraineurs regardless of disease activity.² Although the nature of these microstructural changes remains elusive, our findings suggest that deep WMHs in migraine develop gradually or stepwise, possibly due to complex interplay of electrophysiologic and vascular phenomena. We may only speculate about underlying pathophysiology. Cortical hyperexcitability is thought to lower the threshold for cortical spreading depression in migraine. This cortical spreading depression, possibly triggered by ischemic events, leads to a phase of self-expanding oligemia,²⁰ which may trigger the release of proinflammatory and prothrombotic substances. Together with preexistent modulating factors as an increased cardiovascular risk profile, circulating substances associated with vascular impairment, and dysfunctional endothelium,²¹ this may ultimately contribute to a state of (chronic) ischemia, leading to axonal loss, demyelination, or gliosis, appearing as WMHs on conventional MRI.

Our study was underpowered so we cannot exclude the possibility that WMHs in control participants may develop in a similar gradual or stepwise manner. After all, 50% of controls who participated in the follow-up study had progressive white matter lesions in the same follow-up period as well, compared to 54% in migraineurs. In some controls, these white matter lesions may have been preceded by similar MTR differences as seen in migraineurs. However, it should also be mentioned that the average increase in lesion load was (nonsignificantly) lower in controls, further decreasing the possibility to find similar MTR changes. Since MTR is not homogeneous throughout the white matter due to local differences in myelination,³ differences in distribution of deep WMHs between migraineurs and controls may explain why we found lower mean MTR in deep WMHs at baseline in migraineurs compared to controls. A previous study based on the whole CAMERA-2 study population indeed showed that WMHs were more diffusely distributed in female migraineurs compared to controls, also affecting subcortical white matter in upper frontal regions.² Alternatively, microstructure within WMHs in migraineurs might also just be more severely affected than in controls. Further, we cannot exclude that MTR values might have decreased even further after our baseline measurement, as MTR changes may show evolution in time.

Nevertheless, the MTR changes in our and older migraine studies⁸ are mild compared to those found in, for instance, multiple sclerosis⁸ or elderly subjects.^22,23 This might explain why deep WMHs are not associated with clinically relevant cognitive decline in migraine.²

Thus, despite being visible, these WMHs only show minor changes in microstructural integrity in migraineurs and may thus well be clinically insignificant. Therefore, when incidentally discovered on conventional MRI, they may be no reason for alarm. Our finding that migraine is not accompanied by occult extensive white matter disease might provide further reassurance for migraineurs and their physicians.

Glossary

BMI

body mass index

CAMERA

Cerebral Abnormalities in Migraine, an Epidemiological Risk Analysis

FLAIR

fluid-attenuated inversion recovery

FLIRT

FMRIB's linear image registration tool

HDL

high-density lipoprotein

MNI

Montreal Neurological Institute

MTI

magnetization transfer imaging

MTR

magnetization transfer ratio

NAWM

normal-appearing white matter

TE

echo time

TR

repetition time

WMH

white matter hyperintensities

Appendix. Authors

Study funding

Supported by grants from the NIH (1R01NS061382-01) and the Netherlands Heart Foundation (2007B016). The sponsors had no role in the design or conduct of the study.

Disclosure

E. Arkink, I. Palm-Meinders, H. Koppen, J. Milles, B. van Lew, L. Launer, and P. Hofman report no disclosures relevant to the manuscript. G. Terwindt: European Community (FP7-EUROHEADPAIN, no. 602633). M. van Buchem reports no disclosures. M. Ferrari: Netherlands Organization for Scientific Research (NWO: VICI 91856601 and Spinoza 2009). M. Kruit reports no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.