The Interrelation Between Chronic Headache,Cognitive Scores,and MRI Markers Among Stroke Survivors

Abstract

Background:

Migraine is known to mildly increase the risk for ischemic stroke and is associated with vascular MRI markers. However, the potential effect of chronic headache (CH) on stroke outcomes has not been studied.

Objective:

We aimed to assess the interrelation between CH and post-stroke cognitive impairment.

Methods:

Data from 455 patients with a first ever stroke from the TABASCO study was available. All patients underwent 3T brain MRI, blood analysis, and a serial cognitive assessment at baseline and 6, 12, and 24 months after.

Results:

Eighty-five (18.7%) patients reported suffering from CH, of whom 53 (62.4%) reported symptoms of photophobia or nausea, and 34 (40%) reported an aura. CH was associated with female sex, lower prevalence of T2DM (p < 0.001), and lower HbA1C levels (p < 0.001). Multiple regression analysis, controlling for age, sex, education, vascular risk factors, and the presence of acute lesions in MRI, revealed that CH was an independent predictor of better cognitive scores 6, 12, and 24 months post-stroke (p = 0.015, p = 0.01, and p = 0.012, respectively). Stroke patients suffering from CH had also higher normalized gray, white matter, and thalamus volumes, and better white matter microstructural integrity (p < 0.001, p = 0.037, p < 0.001, p = 0.008, respectively)

Conclusion:

In this study, CH was consistently associated with better long term cognitive scores among post stroke subjects. These surprising findings may partially arise from the higher prevalence of T2DM among subjects without CH, that may represent the existence of chronic cerebrovascular disease, and may reflect mechanisms involving glucose metabolism.

Keywords

Chronic headache migraine post-stoke cognitive impairment stroke TABASCO

INTRODUCTION

The complex interrelation between migraine and ischemic stroke risk has been explored extensively [1 –3]. Most studies have shown that migraine may slightly increase the risk of ischemic stroke, especially among patients with aura [1 –3] or in combination with other factors such as smoking or use of contraceptives [4, 5]. Various mechanisms were suggested, including spreading depolarization, vasoconstriction, altered platelet function, endothelial dysfunction, coagulopathies, and patent foramen ovale [4 , 7].

The interrelation of migraine with stroke was also examined by looking at MRI features [4]. Some studies described association between migraine and silent infarct-like lesions (ILLs, i.e., T2 FLAIR hyperintense asymptomatic brain parenchymal lesions) [8, 9]. However, other studies including a large meta-analysis did not find increased prevalence of ILLs in patients with migraine when compared to controls [10, 11]. The potential effect, however, of chronic headache (CH) on stroke outcomes has not been studied.

We examined the interrelation between CH, post-stroke cognitive impairment, and MRI markers in a prospective cohort of stroke survivors.

METHODS

Study population

The present study included survivors of mild to moderate first acute ischemic stroke/transient ischemic attack (TIA) with a total NIH Stroke Scale (NIHSS)<17, who have participated in the prospective TABASCO study [12].

We excluded subjects whose stroke resulted from trauma or invasive brain procedures, hemorrhagic stroke, cognitive impairment before the stroke (determined by Informant Questionnaire on Cognitive Decline in the Elderly –IQCODE [13] score > 3.3), Type 1 diabetes mellitus (2 subjects), and severe aphasia or disability which would impair long term follow-up. The neurological assessment included verification of stroke etiology, NIH stroke scale (NIHSS), and neuroimaging. Vascular risk factors were assessed according to Framingham Stroke Risk Profile score. Metabolic syndrome was defined as previously described [14].

Information on history of chronic headache was collected upon admission due to the ischemic index event, based on patient reports and data from their medical records.

This study was registered as ClinicalTrials.gov Identifier: NCT01926691. All participants signed informed consent forms, approved by the local ethics committee.

Definition of T2DM

Type 2 diabetes mellitus (T2DM) was defined based on patient report, hemoglobin A1C (HbA1C) level above 6.5%, and/or the prior use of anti-hyperglycemic agents. Absence of T2DM was defined as none of the above. An HbA1C level below 5.7% is considered normal, between 5.7 and 6.4% signals pre-diabetes, and T2DM is diagnosed when the HbA1C is 6.5% or above [15]. Duration of T2DM was determined by self-report.

Baseline and follow-up cognitive assessments

Subjects completed a baseline neuropsychological assessment including the Montreal Cognitive Assessment (MoCA) [16] and the NeuroTrax computerized cognitive testing (NeuroTrax Corp., Bellaire, TX) [17] during their first days of hospitalization immediately after the event. These comprehensive neuropsychological evaluations were repeated 6, 12, and 24 months later. A Global Cognitive Score was computed as the average of the six index scores (memory, executive functions, visuospatial perception, verbal function, attention, and motor skills). Data for each NeuroTrax parameter were normalized according to stratifications of age and education to give a distribution with a mean of 100 and a standard deviation of 15.

MRI acquisition

All images were acquired within 7 days from onset of the acute event on a 3T GE scanner (GE Signa EXCITE, Milwaukee, WI, USA) using an 8-channel head coil. The protocol consists of the following pulse sequences: Axial fast Spin-Echo (FSE) T2-weighted images (WI) (Time to repeat (TR)/Time to echo (TE) = 13000/110 ms, Field of View (FOV) = 240, Matrix = 512×256 and slice thickness of 4 mm with no gap, fluid-attenuated inversion recovery (FLAIR), T2* Gradient Echo (GE) image, and high resolution 3D T1-WI spoiled GE sequences (SPGR) axial (TR/TE = 8.976/3.488 ms, FOV = 256, Matrix = 256×256 and slice thickness of 1 mm) and coronal (TR/TE = 6.35/1.8 ms, FOV = 250, Matrix = 256×256 and slice thickness of 2 mm). Diffusion weighted imaging (DWI) sequence (TR/TE = 6000/72.4 ms, FOV = 240, Matrix = 128×128 and slice thickness of 4 mm with no gap, b values = 0, 1000 s/mm) and diffusion tensor images (DTI) were obtained using echo planar imaging sequence. All axial slices were prescribed on the same orientation, covering the whole brain, aligned along the fourth ventricle-orbitofrontal orientation.

Ischemic infarct identification and volume

Presence of acute ischemic infarcts was assessed by a senior neuroradiologist based on the DWI images. Cortical infarcts were defined as any infarct that involves the cortex. Subtentorial infarcts were defined as cerebellar or brainstem infarctions. Volumes (mm³) of the ischemic lesions were calculated. The quantification of the ischemic lesions was performed using a semi-automated method [18].

Tissue segmentation

The identification and quantification of the ischemic lesions, total gray matter (GM) volume, white matter lesion (WML), and normal appearing white matter (NAWM) performed using multi-modal view with a semi-automatic method [18]. Volumes (mm³) of both ischemic lesions and WML were calculated across the whole brain.

Volumetric measurement of the hippocampus and thalamus

Volumetric analysis was performed on high-resolution 3D T1-WI axial images, using the FreeSurfer V5.1 image analysis suite (http://surfer.nmr.mgh.harvard.edu/) for all brain segmentation based on probabilistic atlas and intensity values, as previously described [19].

DTI analysis

Calculation of the DTI maps was performed using FMRIB Diffusion Toolbox, part of FMRIB software Library (FSL, http://www.fmrib.ox.ac.uk/fsl/), and included eddy current and motion correction. Four different indices maps were calculated in the normal appearing white matter: fractional anisotropy (FA), mean diffusivity (MD), perpendicular diffusivity (λ’), and parallel diffusivity (λ|).

Small vessel disease (SVD) markers

All MRI scans were fully evaluated for SVD radiological markers according to the STRIVE protocol [20] including: (A) White matter hyperintensities (WMH) were graded using Fazekas [21]. (B) Old lacunar infarcts were defined as sharply demarcated hypointense lesions sized between 3 mm and 15 mm in diameter on T1-weighted images with corresponding hypointense lesions with hyperintense rim on T2 dark fluid [20]. (C) Cerebral microbleeds (CMB) were defined as round hypointense lesions on susceptibility weighted imaging (SWI) with a diameter < 10 mm. In order to rule out CMB mimics, other sequences including DWI and T1 were evaluated [22]. CMBs were divided to cortical versus deep.

Statistical analyses

All variables were reported as a mean value (±SD), a median value with an interquartile range (IQR), or a proportion/percentage of the total.

Multiple linear regression analysis was used to assess the association between CH before stroke and cognitive domains as well as MRI measures, adjusted for age, sex, years of education, vascular risk factors (history of hypertension, T2DM), and the presence of acute lesion in MRI.

All subjects with incomplete information processing performances (i.e., completed some part of the subtest but discontinued by the computer program because of inadequate performance) were classified as impaired, since incomplete performance influences the total NeuroTrax score.

Furthermore, in order to examine differences in cognitive trajectories across time we used the repeated measures approach based on the general linear model (GLM) for the comparison of the cognitive scores at 6, 12, and 24 months by history of CH at admission.

In order to study the individual and joint association of CH and T2DM on neuroimaging parameters and cognitive scores, we defined four mutually exclusive categories: 1) (reference category) subjects with CH only, 2) CH and T2DM, 3) no CH and no T2DM, and 4) T2DM only. We tested for the interaction between CH and T2DM by comparing the effects of having T2DM without CH vs CH and T2DM + CH alone on cognitive scores and on total GM volume. We used the univariate analysis of variance to find significant main effects and interactions while adjusted for sex, age, years of education, and intracranial volume (ICV). Statistical analysis was performed using SPSS software (version 25.0, Chicago, IL, USA).

RESULTS

A total of 575 consecutive eligible cognitively intact patients at baseline who were admitted to the Department of Emergency Medicine at the Tel-Aviv Medical Center between April 1, 2008 to December 1, 2014, within 72 hours from onset of symptoms of TIA or ischemic stroke were initially evaluated. Data on history of chronic headache and cognitive assessments at baseline and two years thereafter was available for 455 participants, thus formed the final sample for analyses. Brain MRI scans at baseline were available for 356 subjects. No differences in cognitive results were observed between participants included and not included in the MRI study. Clinical characteristics of the study population is presented in Table 1.

Table 1

Baseline and follow up characteristics of post-stroke survivors (n = 435)

	No CH	CH	p
	N = 370	N = 85
Age, y (SD)	67.3 (9.6)	65.9 (9.4)	0.211
Female gender, n (%)	130 (35.1)	57 (67.1)	< 0.001
Education, y (SD)	13 (4)	13.4 (3.5)	0.406
Body-mass index, kg/m2 (SD)	27.3 (4.5)	27.3 (4.3)	0.984
Current smokers, n (%)	86 (23.2)	15 (17.6)	0.331
T2DM, n (%)	125 (33.8)	11 (12.9)	< 0.001
Dyslipidemia, n (%)	201 (54.3)	46 (54.1)	0.972
Hypertension, n (%)	222 (60)	51 (60)	1.00
Atrial fibrillation, n (%)	29 (7.8)	9 (10.6)	0.413
Ischemic heart disease, n (%)	64 (17.3)	8 (9.4)	0.072
Metabolic syndrome (MetS), n (%)	123 (33.2)	19 (22.4)	0.051
Framingham Risk Score for Stroke	11.7 (5.8)	10.6 (5.6)	0.182
APOEɛ4 allele, n (%)	66 (17.8)	12 (14.1)	0.432
Admission NIHSS, median (IQR)	2 (1–4)	1 (0–2)	0.001
HbA1C, % (SD)	6.4 (1.4)	5.9 (0.6)	< 0.001
Physical activity, minutes/week (SD)	100.2 (171.6)	89.9 (130.6)	0.606
Modified Rankin 0,1, or 2	290 (78.4)	71 (83.5)	0.362
Modified Rankin 3 or 4	80 (21.6)	14 (16.5)
Recurrent cerebrovascular event/s during	10 (2.7)	6 (7.1)	0.048
24 months after the admission, n (%)
Use of beta-blockers (%)	115 (31.1)	33 (38.8)	0.169
Use of calcium channel blockers (%)	69 (18.6)	20 (23.5)	0.306
Use of non-steroidal anti-inflammatory	0	0
drugs (NSAIDs) (%)
Use of triptans (%)	0	1 (1.2)	N/A
Use of acetyl salicilic acid pre-stroke (%)	138 (37.3)	26 (30.6)	0.245
Use of oral hypoglycemic agents (%)	81 (21.9)	7 (8.2)	0.004
Use of thyroid replacement therapy (%)	27 (7.3)	12 (14.1)	0.043
TIA (%)	106 (28.6)	38 (44.7)	0.004
GDS at 12 months, median (IQR)	2 (0–4)	1 (0–4)	0.901
GDS at 24 months, median (IQR)	2 (0–4)	1 (0–4)	0.853
Etiology (TOAST criteria)	N = 264	N = 47
Lacunar strokes	169 (64)	19 (40.4)	0.004
Cardioembolic stroke	31 (11.7)	9 (19.2)	0.301
Large-artery atherosclerotic stroke	25 (9.5)	0	0.051
Other or undetermined etiology	39 (14.8)	19 (40.4)	0.002
Cognitive scores
Admission MoCA score (SD)	23.3 (3.6)	24.4 (3.6)	0.024
MoCA score at 6 months (SD)	24.9 (3.7)	26 (3.0)	0.01
MoCA score at 12 months (SD)	24.9 (3.8)	26.0 (3.6)	0.042
MoCA score at 24 months (SD)	24.6 (4.3)	26.1 (3.3)	0.001
Computerized Total cognitive	91.6 (14.4)	93.8 (11.0)	0.215
score at admission (SD)
Computerized Total cognitive score	93.7 (12.7)	98.5 (11.1)	0.003
6 months post-stroke (SD)
Computerized Total cognitive score	95.5 (13.1)	99.3 (11.3)	0.04
12 months post-stroke (SD)
Computerized Total cognitive score	95.9 (12.2)	100.0 (9.4)	0.005
24 months post-stroke (SD)
Brain MRI parameters	N = 285	N = 71
Not performed MRI, n (%)	85 (23)	14 (16.5)	0.190
Infarct type
No infarct in MRI, n (%)	71 (24.9)	31 (43.7)	0.042
Cortical infarct, n (%)	74 (26)	19 (26.8)	0.841
Sub-cortical infarct, n (%)	103 (36.1)	18 (25.4)	0.196
Subtentorial infarct, n (%)	37 (13)	3 (4.2)	0.049
Ischemic lesions volume, mm³ (SD)	7502.7 (1186.9)	7206.2 (1209.2)	0.091
Total WM volume, normalized to	30.6 (3.0)	31.5 (3.3)	0.037
ICV, mm³ (SD)
Total GM volume, normalized	39.8 (3.3)	41.6 (6)	< 0.001
to ICV, mm³ (SD)
Total thalamus volume,	0.81 (0.09)	0.86 (0.11)	< 0.001
normalized to ICV, mm³ (SD)
Total hippocampal volume,	0.52 (0.08)	0.55 (0.1)	0.0440
normalized to ICV, mm³ (SD)
NAWM MD values, mm²/s	0.00086 (0.00005)	0.00084 (0.00004)	0.023
NAWM FA values, arbitrary units	0.4 (0.01)	0.41 (0.01)	0.047
NAWMλ’ values, mm²/s	0.00067 (0.00005)	0.00065 (0.00004)	0.008
NAWMλ\| values, mm²/s	0.0012 (0.00007)	0.0012 (0.00004)	0.035
Microbleed yes/no (%)	44 (15.4)	7 (9.9)	0.339
Lacunes yes/no (%)	111 (38.9)	23 (32.4)	0.308

HbA1C, glycated hemoglobin; CH, chronic headache; Da, axial diffusivity; Dr, radial diffusivity; FA, fractional anisotropy; GDS, geriatric depression scale; GM, gray matter; ICV, intracranial volume; MD, mean diffusivity; MoCA, Montreal Cognitive Assessment; MRI, Magnetic resonance imaging; NAWM, normal-appearing white matter; NIHSS, National Institutes of Health Stroke Scale; SD, standard deviation; T2DM, Type 2 diabetes mellitus; TIA, transient ischemic attack; WM, white matter. Entries are mean (SD) or n and %, as indicated. Significant results are shown in bold (p < 0.05).

Of the 455 subjects, 85 (18.7%) reported suffering from CH, of whom 53 (62.4%) had symptoms of photophobia or nausea, and 34 (40%) reported experiencing an aura prior to headache. Subjects with CH were more likely to be female (67.1% versus 32.9%, p < 0.001), had lower prevalence of T2DM (12.9% versus 33.8%, p < 0.001) and lower levels of hemoglobin HbA1C (5.9±0.6 versus 6.4±1.4, p < 0.001). Additionally, stroke severity at admission, as assessed by the NIHSS, was lower in the CH group (1[0–2] versus 2[1–4], p = 0.001), with lower rates of ischemic stroke versus TIA compared to patients without CH (55.3% versus 71.4%, p = 0.004).

Patients with CH were less likely to use oral hypoglycemic agents (8.2% versus 21.9%; χ² = 8.26, p = 0.004) but more likely to use thyroid replacement therapy (14.1% versus 7.3%; χ² = 4.10, p = 0.043) prior to the index event. No significant difference in other medications was noted, including beta-blockers, calcium channel blocker, non-steroidal anti-inflammatory drugs (NSAIDs) and triptans (Table 1).

Stroke etiologies in the complete cohort (based on TOAST criteria) included 188 patients with lacunar stroke (60.5%), 39 patients with cardioembolic stroke (12.5%), 24 patients with large-artery atherosclerotic stroke (7.7%), 60 patients with stroke of other or undetermined etiology (19.3%), and 144 patients (31.6%) with a TIA. Patients with CH had fewer lacunar strokes than patients not suffering from CH (40.4% versus 64%, p = 0.004), none of the CH patients had stroke secondary to large artery arteriosclerosis, and more had stroke of other determined or undetermined etiology (40.4% versus 14.8%, p = 0.002).

Chronic headache and MRI measurements

CH was associated with higher normalized GM, WM, thalamus, and hippocampal volumes (p = 0.001, p = 0.045, p = 0.003, p = 0.02, respectively). DTI analysis demonstrated better global NAWM microstructural integrity (reflected as higher FA and lower MD,λ’, and λ| indices, p = 0.054, p = 0.008, p = 0.003, p = 0.013, respectively). No differences were observed regarding the presence of ischemic lesions and lacunas, or other measures of small vessel disease (Table 1). As our cohort included both stroke and TIA patients, we performed a sub-analysis in patients with evidence of acute DWI lesion in neuroimaging (evidence of acute stroke), and results were similar (Supplementary Table 2).

A longer duration of CH was associated with higher normalized total GM and hippocampal volumes (r = 0.387, p = 0.026; r = 0.397, p = 0.022, respectively), as well as with higher memory scores 12 months post-stroke (r = 0.343, p = 0.035). No association was observed between headache duration in years and WM volume or integrity.

Chronic headache and cognitive performance post-stroke

Table 2 presents the results of the longitudinal cognitive assessments at admission, 6, 12, and 24 months post-stroke. CH was independently associated with better results in memory and global cognitive scores at all time points, better executive function scores at 12 months and better verbal function scores at 24 months post-stroke. These associations remained significant after adjustment for age, sex, education, vascular risk factors including T2DM, ICV, and the presence of acute lesions in MRI (Table 2). No association was found between number of lacunes, lesion location and volume, and development of post-stroke cognitive impairment. Moreover, further adjustment to stroke etiology revealed similar results.

Table 2

Associations between CH and cognitive measures

Cognitive domain at admission	β (95% CI)	Standardized β	p
Memory score	5.5 (–0.92–11.92)	0.12	0.092
Executive function score	3.21 (–1.79–8.21)	0.09	0.207
Visuo-spatial score	–0.65 (–6.59–5.29)	–0.02	0.830
Verbal memory score	4.99 (–3.82–13.81)	0.08	0.265
Attention score	5.32 (–1.33–11.97)	0.12	0.116
Total cognitive score	3.81 (–1.04–8.67)	0.11	0.123
Cognitive domain at 6 months	β (95% CI)	Standardized β	p
Memory score	10.26 (4.8–15.71)	0.23	< 0.001
Executive function score	2.23 (–1.47–5.93)	0.08	0.237
Visuo-spatial score	1.47 (–3.9–6.83)	0.03	0.590
Verbal memory score	7.14 (–0.06–14.33)	0.14	0.052
Attention score	4.0 (–0.68–8.68)	0.11	0.094
Total cognitive score	4.59 (0.9–8.29)	0.16	0.015
Cognitive domain at 12 months	β (95% CI)	Standardized β	p
Memory score	0.24 (4.13–14.74)	2.69	0.001
Executive function score	0.13 (–0.21–8.58)	4.28	0.05
Visuo-spatial score	0.078 (–2.4–9.74)	3.78	0.234
Verbal memory score	0.12 (–0.98–13.7)	3.6	0.089
Attention score	0.11 (–0.81–9.25)	2.55	0.099
Total cognitive score	0.17 (1.33–9.5)	2.07	0.01
Cognitive domain at 24 months	β (95% CI)	Standardized β	p
Memory score	0.19 (2.39–13.84)	2.9	0.006
Executive function score	0.12 (–0.51–7.82)	2.11	0.085
Visuo-spatial score	0.04 (–4.16–7.86)	3.04	0.538
Verbal memory score	0.14 (–0.06–14.97)	3.81	0.05
Attention score	0.06 (–2.72–6.61)	2.36	0.412
Total cognitive score	0.16 (1.12–8.99)	2.0	0.012

β coefficient (95% CI) adjusted for age, sex, education, vascular risk factors (history of hypertension, T2DM), and the presence of acute lesion in MRI. Standardized β coefficient of CH for regression against each cognitive measure. Significant results are shown in bold (p < 0.05).

To examine differences in cognitive trajectories across time, we used the repeated measures approach (GLM) of the four observation points of the memory scores for the comparison between patients with and without CH (Fig. 1). As expected, most patients cognitive function improved from baseline to 6 months (72.5%) and from 6 to 12 months post-stroke (64.8%), but only the CH group retained this improvement at 12 to 24 months. Furthermore, the cognitive scores of CH subjects remained significantly higher compared to the non-CH group at 6, 12, and 24 months post-stroke (p < 0.001, p = 0.001, p = 0.006, respectively, Fig. 1).

Fig. 1

General linear model (GLM) analysis of repeated measures of memory scores along the follow-up, comparing patients with CH to patients without. ^*p < 0.05, ^**p < 0.001.

Similar results were demonstrated in a sub-analysis of patients with radiologically proven acute ischemic stroke based on diffusion weighted imaging (254/356 = 71.3%). In this sub-group, CH was associated with better results in memory, verbal function, and global cognitive scores at 6, 12, and 24 months after the index event (p < 0.001, p = 0.018, p = 0.004, for 6 months; p = 0.005, p = 0.004, p = 0.027, for 12 months, and p = 0.038, p = 0.025, p = 0.031, for 24 months, respectively), as well as higher total GM and thalamus volumes, normalized to ICV (p = 0.001, p = 0.04, respectively), and better global NAWM microstructural integrity (lowerλ’ values, p = 0.046, Supplementary Table 1).

Migraines and MRI measurements

Seventy-seven (16.9%) patients reported suffering from symptoms of chronic migraine (CM), of whom 68.8% (53/77) reported symptoms of photophobia or nausea and 44.2% (34/77) reported an aura. Migraine was diagnosed based on patient report alone. Patients with CM had higher normalized GM, WM, thalamic, and hippocampal volumes (p = 0.001, p = 0.049, p = 0.02, p = 0.037, respectively), and better global NAWM, microstructural integrity (reflected as higher NAWM FA and lower NAWM MD,λ’, and λ| indices, p = 0.003, p = 0.005, p = 0.003, p = 0.019, respectively) when compared to patients without CM or with non-migrainous headache.

T2DM, MRI measurements, and cognitive performance post-stroke

T2DM was associated with inferior cognitive scores 6 and 24 months post-stroke (p = 0.048, p = 0.002, respectively), as well as lower normalized GM and WM volumes (p = 0.009, p = 0.016, respectively), and worse NAWM microstructural integrity (reflected as lower NAWM FA and higher NAWM MD,λ’, and λ| indices, p = 0.005, p = 0.002, p = 0.002, p = 0.017, respectively) when compared to subjects without T2DM (Supplementary Table 2).

Higher HbA1C was associated with lower normalized GM volume, after adjustment for age, sex, education, vascular risk factors, ICV, and the presence of acute lesion in MRI (p = 0.038) and with worse NAWM microstructural integrity (reflected as higher NAWM MD,λ’ and λ| indices, p = 0.024, p = 0.044, p = 0.026, respectively).

Effect of CH and glycemic control on MRI measurements and cognitive performance

To study the interaction between CH, T2DM and HbA1C levels we divided the study population into four subgroups: 1) CH and no T2DM (n = 74); 2) CH and T2DM (n = 11); 3) no CH and no T2DM (n = 245); 4) no CH and T2DM (n = 125). Overall, patients with T2DM (groups 2 and 4) had higher NIHSS scores at admission (p < 0.001), showed a clear trend in risk elevation for brain atrophy: lower normalized total GM, WM, thalamus, and hippocampus volumes (p = 0.003, p = 0.006, p = 0.046, p < 0.001, respectively), worse WM integrity (p = 0.001), and lower long term cognitive scores 2 years post stroke/TIA (p = 0.002) when compared to patients without T2DM (groups 1 and 3). However, among patients with T2DM, the combination with CH (group 2) was associated with better outcomes—lower NIHSS scores (p = 0.036), greater normalized GM volume (p = 0.023), and a non-significant trend toward better cognitive scores 2 years post-stroke. Similar results were observed in patients without T2DM. CH was associated with lower NIHSS scores (p = 0.012), higher total GM (p = 0.008), and better cognitive scores 2 years post-stroke (p = 0.041), (Fig. 2A, B). Group 4 (no CH and T2DM) had significantly greater brain atrophy and worse cognitive scores 2 years post-stroke than all other groups (p < 0.001). We further tested for interaction between CH and T2DM. A two-way ANOVA was conducted that examined the effect of CH and T2DM on total cognitive score. A statistically significant interaction between the effects of CH and T2DM on cognitive scores was demonstrated one and two years post-stroke in a model adjusted for sex, age, years of education, and ICV [F (1,314) = 3.75, p = 0.05, and β (for interaction) = 10.47 (95% CI: –0.2–21.14)]. No significant interaction effect was observed on normalized GM volume.

Fig. 2

Gray matter volume and cognitive results according to group stratification of HC and T2DM. A1C, glycated hemoglobin; ICV, intracranial volume; T2DM, Type 2 diabetes mellitus.

We then performed a repeated analysis incorporating pre-diabetic patients and diabetic patients into one group (based on HbA1C≥5.7%) and divided into four subgroups: 1) CH and HbA1C < 5.7% (n = 25); 2) CH and HbA1C > 5.7% (n = 51); 3) no CH and HbA1C < 5.7% (n = 90); 4) no CH and HbA1C > 5.7% (n = 230). Group 4 had a lower normalized total GM volume (p = 0.048) and worse cognitive scores 2 years post-stroke (p = 0.014), and group 3 had a lower normalized total GM volume (p < 0.001) and lower cognitive scores 2 years post-stroke (p = 0.045), versus groups 1 and 2 (CH) (Fig. 2C, D). Association of GM volume with age revealed significant negative association for all sub-groups. However, groups 3 and 4 (no CH) presented lower GM volumes starting at a younger age than groups 1 and 2 (CH) (Supplementary Figure 1).

DISCUSSION

In the current study, chronic headache was consistently associated with better baseline and follow-up cognitive scores and baseline MRI measurements among post-stroke patients. The difference was more robust when the CH exhibited migrainous features. These surprising results most likely arise from differences in the pathophysiology and mechanism beyond the stroke. Migraine, especially with aura, is known to mildly elevate the risk for ischemic stroke [1, 2]. Studies examining the interrelation of migraine with silent radiological infarcts and other MRI features demonstrated conflicting results [8 –11]. It should be noted that in these studies the elevated risk for stroke among migrainous patients was of course compared to healthy controls, and not to patients with previous stroke or cardiovascular risk factors.

When looking at CH and cognition, the CAMERA and EVA studies found no association between brain abnormalities and cognitive decline in persons with migraine or other severe headache [23, 24]. This is in accordance with large population-based studies that have shown no long-term effects of migraine on cognitive function [25]. In contrast, in our cohort, patients without CH had a higher prevalence of T2DM and worse baseline various MRI features including GM atrophy. These findings may represent the existence of ongoing progressive cerebral vascular disease, putting these patients at higher risk for recurrent events and neurological sequela such as cognitive impairment. Cerebral vascular disease is the second most common cause for dementia, and post-stroke cognitive decline is common. Moreover, T2DM and altered glucose metabolism were found to be associated with reduced post-stroke cognitive function [26, 27]. The importance of stroke mechanism is highlighted when looking at the distribution of stroke etiology in our cohort. Among CH patients, etiology by TOAST was other determined or undetermined in 40.4% of patients as compared to only 14.8% in patients without CH. Lacunar strokes and stroke secondary to large arteriosclerosis, were less common among CH patients—40.4% lacunar strokes in headache patients comparing to 64% in patients without CH. None of the CH patients had stroke secondary to large arteriosclerosis versus 9.5% in patients without CH. This may also point out to possible differences in stroke recurrence rates.

As our cohort included both stroke and TIA patients, it may be speculated that more CH patients had actually “TIA mimics”. Indeed, there were more cases of negative DWI scans among CH patients. We therefore performed a sub-analysis in patients with evidence of acute DWI lesion in neuroimaging (evidence of acute stroke), which demonstrated similar results. This further strengthens our hypothesis and reduces the risk of bias. The results also show mild difference in admission NIHSS between CH and non-CH groups (1 versus 2 respectively). It is our belief that the difference found in NIHSS is not the causative factor beyond the cognitive results, rather it may arise from the impact of underlying brain pathology and previous damage, which also affects cognition. This could be supported by the difference described in baseline MRI markers including GM and WM volume, and the fact that no difference was found comparing cortical and subcortical lesions or lesion volume between groups.

The reduced GM and WM volumes found in non-CH patients may represent brain damage secondary to various processes and may imply on mixed pathologies. This may be supported by the slightly larger hippocampal volume among CH patients. It should be noted, however, that no difference was observed in the distribution of APOE4 allele between non-CH and CH patients, nor in age and education. Thus, difference in the underlying neurodegenerative pathology cannot be excluded, although less likely in our opinion.

When comparing pre-stroke cardiovascular risk factors of CH and no-CH patients, the only difference noted was the prevalence of T2DM. T2DM has been shown to increase the risk for cognitive decline and dementia and was previously found to predict post-stroke cognitive impairment [28]. Furthermore, in a recent study, Moran et al. demonstrated association between T2DM and brain atrophy in a cohort of patients with low cerebrovascular disease burden [29]. Therefore, the low prevalence of T2DM among patients with migraines may serve as a protective factor for post-stroke cognitive impairment. However, this cannot be the sole explanation beyond our results, as even in the cohort of patients with prior T2DM the ‘protective’ effect of CH remained. This protective effect could neither be explained by the use of chronic medications commonly prescribed for migraines which could potentially exert a protective effect (including beta-blockers, calcium channel blocker, non-steroidal anti-inflammatory drugs (NSAIDs) and triptans) [30], as no significant difference was found in the rate of chronic usage of these medications.

It is still puzzling why CH patients in our cohort had lower prevalence of T2DM and altered glucose metabolism, despite a similar mean age and clinical profile. Similar findings were described in a recent study from the E3N cohort [31]. In that cohort, patients with migraines had a lower risk for T2DM, and the prevalence of active migraine decreased prior to the diagnosis of T2DM [31]. Some studies found similar results, while others described positive association between CH and obesity or insulin resistance [32, 33]. Migraine was previously found to be associated with polymorphisms in the insulin receptor gene [34], higher levels of plasma insulin [35], and elevation in free fatty acid and ketone bodies concentration before a migraine attack [31]. Fagherazzi et al. suggest that relative hypoglycemia secondary to fasting, increased insulin secretion, or high insulin sensitivity may trigger migraine attacks [31]. The inverse association between T2DM and CH may also arise from predisposing damage to the vasa-neurosum of cerebrovascular and meningeal structures, leading to nerve damage and thus lower prevalence of CH. This may be supported by studies which described lower frequency of post-stroke headache among T2DM patients [36].

Another interesting possible mechanism involves the calcitonin gene-related peptide (CGRP). This neuropeptide, which is expressed in sensory nerves, was found to be a strong vasodilator and is considered to have an important role in the pathophysiology of migraine [37]. CGRP was also found to be associated with glucose metabolism [38]. Animal studies have demonstrated a strong link between CGRP, migraine, and glucose metabolism [39]. Serum levels of CGRP are decreased in patients with diabetes [40]. CGRP was found to be associated with decrease insulin secretion, increase insulin sensitivity, and suggested effect assisting weight loss. As CGRP is linked to migraine and T2DM, is may be the missing connecting underlying factor. Whether genetic variations influence CGRP secretion or the response to CGRP is still unknown. Another possible mechanism is that the development of insulin resistance and diabetes result in damage to the sensory neurons producing CGRP. This decrease in CGRP secretion may lead to resolution of migraine [41]. These findings may support an inverse association between migraine and T2DM and altered glucose metabolism, thus explaining the ‘protective’ effect of CH on brain atrophy and cognition seen in our cohort.

The strengths of our study include the systematic prospective follow-up and comprehensive data on patients’ clinical status and structural MRI measures, as well as the use of an extensive and validated computerized cognitive tool battery of multi-domain cognitive tests. The main limitations of our study are the lack of non-stroke control group and the lack of follow-up neuroimaging. Other limitations stem from the limited information obtained from the chronic headache questionnaire, which lacks documentation for some headache features such as frequency, duration of symptoms, autonomic features, and specific doses of anti-headache medications. Another limitation is the exclusion of patients who had severe stroke, who may potentially suffer from more brain atrophy and cerebral vascular disease burden. Thus, our findings cannot be generalized to this sub-group of patients and further research among them is warranted.

CONCLUSION

In this study, chronic headache, especially with migrainous features, was consistently associated with better baseline and long-term cognitive scores and baseline MRI features among stroke survivors. These surprising findings may arise from differences in the mechanism beyond the stroke—as patients without chronic headache had higher prevalence of T2DM and worse baseline MRI markers; this may represent the existence of chronic cerebral vascular disease mitigated by mechanisms involving glucose metabolism and possibly CGRP. The results of our trial and previous studies also emphasize the negative interrelation between migraine and T2DM, which may shed further light as to the elusive causes of these common diseases.

DISCLOSURE STATEMENT

Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/21-0077r2).

Footnotes

The supplementary material is available in the electronic version of this article: .

References

Etminan

, Takkouche

, Isorna

, Samii

(2005) Risk of ischaemic stroke in people with migraine: Systematic review and meta-analysis of observational studies. BMJ 330, 63.

Lantz

, Sieurin

, Sjölander

, Waldenlind

, Sjöstrand

, Wirdefeldt

(2017) Migraine and risk of stroke: A national population-based twin study. Brain 140, 2653–2662.

, Zhou

, Zhao

, Peng

(2017) Migraine and the risk of stroke: An updated meta-analysis of prospective cohort studies. Neurol Sci 38, 33–40.

Agostoni

, Longoni

(2018) Migraine and cerebrovascular disease: Still a dangerous connetction? Neurol Sci 39, 533–537.

Sacco

, Merki-Feld

, Ægidius

, Bitzer

, Canonico

, Kurth

, Lampl

, Lidegaard

, Anne MacGregor

, MaassenVanDenBrink

, Mitsikostas

, Nappi

, Ntaios

, Sandset

, Martelletti

, European Headache Federation (EHF) and the European Society of Contraception and Reproductive Health (ESC) (2017) Hormonal contraceptives and risk of ischemic stroke in women with migraine: A consensus statement from the European Headache Federation (EHF) and the European Society of Contraception and Reproductive Health (ESC). J Headache Pain 18, 108.

Dreier

, Reiffurth

(2015) The stroke-migraine depolarization continuum. Neuron 86, 902–922.

Gryglas

, Smigiel

(2017) Migraine and stroke: What’s the link? What to do? Curr Neurol Neurosci Rep 17, 22.

Kruit

, van Buchem

, Hofman

, Bakkers

, Terwindt

, Ferrari

, Launer

(2004) Migraine as a risk factor for subclinical brain lesions. JAMA 291, 427–434.

Scher

, Gudmundsson

, Sigurdsson

, Ghambaryan

, Aspelund

, Eiriksdottir

, van Buchem

, Gudnason

, Launer

(2009) Migraine headache in middle age and late-life brain infarcts. JAMA 301, 2563–2570.

10.

Bashir

, Lipton

, Ashina

(2013) Migraine and structural changes in the brain: A systematic review and meta-analysis. Neurology 81, 1260–1268.

11.

Kurth

, Tzourio

(2009) Migraine and cerebral infarct-like lesions on MRI: An observation, not a disease. JAMA 301, 2594–2595.

12.

Ben Assayag

, Korczyn

, Giladi

, Goldbourt

, Berliner

, Shenhar-Tsarfaty

, Kliper

, Hallevi

, Shopin

, Hendler

, Baashat

, Aizenstein

, Soreq

, Katz

, Solomon

, Mike

, Usher

, Hausdorff

, Auriel

, Shapira

, Bornstein

(2012) Predictors for poststroke outcomes: The Tel Aviv brain acute stroke cohort (TABASCO) study protocol. Int J Stroke 7, 341–347.

13.

Jorm

, Jacomb

(1989) The Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE): Socio-demographic correlates, reliability, validity and some norms. Psychol Med 19, 1015–1022.

14.

American Heart Association; National Heart, Lung, and Blood Institue, Grundy

, Cleeman

, Daniels

, Donato

, Eckel

, Franklin

, Gordon

, Krauss

, Savage

, Smith

Jr , Spertus

, Costa

(2005) Diagnosis and management of the metabolic syndrome. An American Heart Association/National Heart, Lung, and Blood Institute Scientific Statement. Circulation 112, 2735–2752.

15.

Inzucchi

(2013) Diagnosis of diabetes. N Engl J Med 368, 193.

16.

Nasreddine

, Phillips

, Bédirian

, Charbonneau

, Whitehead

, Collin

, Cummings

, Chertkow

(2005) The Montreal Cognitive Assessment, MoCA: A brief screening tool for mild cognitive impairment. J Am Geriatr Soc 53, 695–699.

17.

Doniger

, Dwolatzky

, Zucker

, Chertkow

, Crystal

, Schweiger

, Simon

(2006) Computerized cognitive testing battery identifies mild cognitive impairment and mild dementia even in the presence of depressive symptoms. Am J Alzheimers Dis Other Demen 21, 28–36.

18.

Artzi

, Aizenstein

, Jonas-Kimchi

, Myers

, Hallevi

, Ben Bashat

(2013) Flair lesion segmentation: Application in patients with brain tumors and acute ischemic stroke. Eur J Radiol 82, 1512–1518.

19.

Dale

, Fischl

, Sereno

(1999) Cortical surfacebased analysis. I. Segmentation and surface reconstruction. Neuroimage 9, 179–194.

20.

Wardlaw

, Smith

, Biessels

, Cordonnier

, Fazekas

, Frayne

, Lindley

, O’Brien

, Barkhof

, Benavente

, Black

, Brayne

, Breteler

, Chabriat

, Decarli

, de Leeuw

, Doubal

, Duering

, Fox

, Greenberg

, Hachinski

, Kilimann

, Mok

, Oostenbrugge

, Pantoni

, Speck

, Stephan

, Teipel

, Viswanathan

, Werring

, Chen

, Smith

, van Buchem

, Norrving

, Gorelick

, Dichgans

(2013) Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 12, 822–838.

21.

Fazekas

, Kleinert

, Offenbacher

, Schmidt

, Kleinert

, Payer

, Radner

, Lechner

(1993) Pathologic correlates of incidental MRI white matter signal hyperintensities. Neurology 43, 1683–1689.

22.

Greenberg

, Vernooij

, Cordonnier

, Viswanathan

, Al-Shahi Salman

, Warach

, Launer

, Van Buchem

, Breteler

; Microbleed Study Group (2009) Cerebral microbleeds: A guide to detection and interpretation. Lancet Neurol 8, 165–174.

23.

Gaist

, Pedersen

, Madsen

, Tsiropoulos

, Bak

, Sindrup

, McGue

, Rasmussen

, Christensen

(2005) Long-term effects of migraine on cognitive function: A population-based study of Danish twins. Neurology 64, 600–607.

24.

Rist

, Dufouil

, Glymour

, Tzourio

, Kurth

(2011) Migraine and cognitive decline in the population-based EVA study. Cephalalgia 31, 1291–1300.

25.

Baars

MAE

, van Boxtel

MPJ

, Jolles

(2010) Migraine does not affect cognitive decline: Results from the Maastricht aging study. Headache 50, 176–184.

26.

Verdile

, Fuller

, Martins

(2015) The role of type 2 diabetes in neurodegeneration. Neurobiol Dis 84, 22–38.

27.

Ben Assayag

, Eldor

, Korczyn

, Kliper

, Shenhar-Tsarfaty

, Tene

, Molad

, Shapira

, Berliner

, Volfson

, Shopin

, Strauss

, Hallevi

, Bornstein

, Auriel

(2017) Type 2 diabetes mellitus and impaired renal function are associated with brain alterations and poststroke cognitive decline. Stroke 48, 2368–2374.

28.

Hanyu

(2019) Diabetes-related dementia. Adv Exp Med Biol 1128, 147–160.

29.

Moran

, Beare

, Wang

, Callisaya

, Srikanth

; Alzheimer’s Disease Neuroimaging Initiative (ADNI) (2019) Type 2 diabetes mellitus, brain atrophy, and cognitive decline. Neurology 92, e823–e830.

30.

Hayden

, Zandi

, Khachaturian

, Szekely

, Fotuhi

, Norton

, Tschanz

, Pieper

, Corcoran

, Lyketsos

, Breitner

, Welsh-Bohmer

; Cache County Investigators (2007) Does NSAID use modify cognitive trajectories in the elderly? The Cache County study. Neurology 69, 275–282.

31.

Fagherazzi

, El Fatouhi

, Fournier

, Gusto

, Mancini

, Balkau

, Boutron-Ruault

, Kurth

, Bonnet

(2019) Associations between migraine and type 2 diabetes in women: Findings from the E3N Cohort Study. JAMA Neurol 76, 257–263.

32.

Rainero

, Limone

, Ferrero

, Valfrè

, Pelissetto

, Rubino

, Gentile

, Lo Giudice

, Pinessi

(2005) Insulin sensitivity is impaired in patients with migraine. Cephalalgia 25, 593–597.

33.

Wang

, Li

, Diao

, Meng

, Xing

, Zhou

, Yang

, Sun

, Chen

, Zhao

(2017) Are glucose and insulin metabolism and diabetes associated with migraine? a community-based, case-control study. J Oral Facial Pain Headache 31, 240–250.

34.

McCarthy

, Hosford

, Riley

, Bird

, White

, Hewett

, Peroutka

, Griffiths

, Boyd

, Lea

, Bhatti

, Hosking

, Hood

, Jones

, Handley

, Rallan

, Lewis

, Yeo

, Williams

, Priest

, Khan

, Donnelly

, Lumsden

, O’Sullivan

, See

, Smart

, Shaw-Hawkins

, Patel

, Langrish

, Feniuk

, Knowles

, Thomas

, Libri

, Montgomery

, Manasco

, Xu

, Dykes

, Humphrey

, Roses

, Purvis

(2001) Single-nucleotide polymorphism alleles in the insulin receptor gene are associated with typical migraine. Genomics 78, 135–149.

35.

Kokavec

(2015) Effect of sucrose consumption on serum insulin, serum cortisol and insulin sensitivity in migraine: Evidence of sex differences. Physiol Behav 142, 170–178.

36.

Hansen

, Marcussen

, Klit

, Kasch

, Jensen

, Finnerup

(2015) Development of persistent headache following stroke: A 3-year follow-up. Cephalalgia 35, 399–409.

37.

Vollesen

ALH

, Snoer

, Beske

, Guo

, Hoffmann

, Jensen

, Ashina

(2018) Effect of infusion of calcitonin gene-related peptide on cluster headache attacks: A randomized clinical trial. JAMA Neurol 75, 1187–1197.

38.

Molina

, Cooper

, Leighton

, Olefsky

(1990) Induction of insulin resistance in vivo by amylin and calcitonin gene-related peptide. Diabetes 39, 260–265.

39.

Siva

, Uluduz

, Keskin

, Erenler

, Balcı

, Uygunoğlu

, Saip

, Göksan

, Siva

(2018) Determinants of glucose metabolism and the role of NPY in the progression of insulin resistance in chronic migraine. Cephalalgia 38, 1773–1781.

40.

Wang

, Zhou

, Li

, Zheng

, Zhu

, Hu

, Sun

(2012) Serum levels of calcitonin gene-related peptide and substance P are decreased in patients with diabetes mellitus and coronary artery disease. J Int Med Res 40, 134–140.

41.

Gelfand

, Loder

(2019) Potential benefits of migraine-what is it good for? JAMA Neurol 76, 250–251.