Sage Journals: Discover world-class research

Abstract

Objective

To examine the relationship between white matter hyperintensities and headache.

Methods

White matter hyperintensities burden was assessed semi-quantitatively using Fazekas and Scheltens scales, and by manual and automated volumetry of MRI in a sub-study of the general population-based Nord-Trøndelag Health Study (HUNT MRI). Using validated questionnaires, participants were categorized into four cross-sectional headache groups: Headache-free (n = 551), tension-type headache (n = 94), migraine (n = 91), and unclassified headache (n = 126). Prospective questionnaire data was used to further categorize participants into groups according to the evolution of headache during the last 12 years: Stable headache-free, past headache, new onset headache, and persistent headache. White matter hyperintensities burden was compared across headache groups using adjusted multivariate regression models.

Results

Individuals with tension-type headache were more likely to have extensive white matter hyperintensities than headache-free subjects, with this being the case across all methods of white matter hyperintensities assessment (Scheltens scale: Odds ratio, 2.46; 95% CI, 1.44–4.20). Migraine or unclassified headache did not influence the odds of having extensive white matter hyperintensities. Those with new onset headache were more likely to have extensive white matter hyperintensities than those who were stable headache-free (Scheltens scale: Odds ratio, 2.24; 95% CI, 1.13–4.44).

Conclusions

Having tension-type headache or developing headache in middle age was linked to extensive white matter hyperintensities. These results were similar across all methods of assessing white matter hyperintensities. If white matter hyperintensities treatment strategies emerge in the future, this association should be taken into consideration.

Keywords

FLAIR white matter lesions white matter disease leukoaraiosis 1.5 tesla small vessel disease T1-weighted

Introduction

Tension-type headache (TTH) and migraine are the two most common primary headache disorders (1,2). Those who experience migraine seem to have more white matter hyperintensities (WMH) than those who are headache-free (3 –8). WMH refer to hyperintense lesions in the white matter visible on T2-weighted magnetic resonance imaging (MRI) (9). A limited WMH load is reported in more than 90% of individuals older than 60 years (10), and is considered normal among the elderly population (11). However, larger amounts of WMH, especially in the deep white matter, may indicate cerebrovascular disease, as the lesions often reflect histopathological changes and molecular markers typical of ischemia (12). A high WMH load is associated with cardiovascular risk factors (13), as well as increased risks of dementia, stroke, and death (14). The impact on the brain of having migraine has been addressed more in previous studies than the impact of having non-migrainous headaches. The few published studies on WMH in non-migrainous headache in adults are contradictory, showing both increased (3,7) and equal amounts (8) of WMH in comparison with controls. The longitudinal MRI studies in adults with headache are also few and conflicting, as findings of both a progressing (6,15) and stable (8) WMH load are reported.

We have previously shown that individuals with headache, and especially with TTH, have increased odds of having minor intracranial abnormalities. WMH determined by a Fazekas score >1 was the predominant pathology in that study (16). In the current study, brain MRI was used to assess the amount of extensive WMH in a random selection of individuals from the general population. Since conflicting reports in the literature may result from the varying methodology used for assessing WMH, we implemented several different methods of assessing WMH; Fazekas and Scheltens semi-quantitative scales, automated WMH volumetry, and for the first time in research investigating relationships between WMH and headache, manual delineation of WMH was performed. Self-reported headache data over a 12-year period from the Nord-Trøndelag Health Study (HUNT) (17 –19) was used to examine associations between the most common headache disorders and WMH, and to investigate what effect temporal evolution of headache had on WMH.

Methods

Participants and the MRI examination

The HUNT study was conducted in three waves in the Norwegian county of Nord-Trøndelag. These three waves ran from 1984–1986 (HUNT1), 1995–1997 (HUNT2), and 2006–2008 (HUNT3). In a neuroimaging sub-study (HUNT MRI), 1494 individuals in the age group 50–65 years who lived within 45 minutes of the location of the MRI examination and had participated in all of the three HUNT surveys were invited to participate (20,21). From 2007–2009, 1088 persons gave informed consent to participate (a response rate of 71%), although 82 subjects were not examined because of practical difficulties (21). Ultimately, 476 men and 530 women successfully underwent an MRI examination of the head. The mean age at the time of image acquisition was 58.5 years (range, 50–66) (21). All participants were scanned on the same General Electric Signa HDx 1.5 T MRI scanner equipped with an eight channel head coil (GE Healthcare). The software version pre-14.0M was used for all of the MRI examinations. Transversal fluid-attenuated inversion recovery (FLAIR), and T1-weighted sequences were used in this study (Table 1). Additional information on the conducting of the MRI examinations was published previously (16,20). Multiple sclerosis (MS) was considered to be a confounder as it may cause both cranial pain (trigeminal neuralgia) and lesions in white matter. Respondents with MS (n = 2) were therefore excluded from the current study.

Table 1.

Scan parameters for the MRI sequences.

MRI sequence	Matrix size	NSA	TR (ms)	TE	Flip-angle (°)	Slice thickness (mm)	Gap (mm)	Overlap (mm)	FOV (mm)
IR-FSPGR	192 × 192	1	10.2	4.1	10	1.2	0	0	240
FLAIR	256 × 224	1	11,002.0	122.9	90	4.0	1	0	230

IR-FSPGR: inversion recovery prepared fast spoiled grass; FLAIR: fluid attenuated inversion recovery; NSA: number of signal averages; TR: repletion time; TE: Echo time; FOV: field of view.

Measurement of WMH

Using FLAIR images, WMH were independently evaluated by two experienced neuroradiologists using the semi-quantitative Fazekas (22) and Scheltens (23) scales. The Fazekas scale rates WMH on a four-point scale: Absent (0), periventricular caps or thin lining or punctuate foci in deep white matter (1), periventricular halo or beginning confluence in deep white matter (2), and periventricular WMH extending into deep white matter or large confluent areas of WMH in deep white matter (3) (22). The Scheltens scale rates periventricular WMH from zero to six, and deep WMH from 0 to 24 (23). According to the standards of the neuroradiology lab of Barkhof and Scheltens, WMH less than 10 mm from the ventricular system were categorized as periventricular. Only lesions with a diameter ≥2 mm were categorized as WMH. Infratentorial hyperintensities or those located in the basal ganglia were not included, as this is normal reading procedure in our clinic and recommended in a position paper on neuroimaging of small vessel disease (9). Differences in opinion regarding the Fazekas score were resolved by consensus following discussion. Manual volumetry was performed in all participants with a Fazekas score above 0. This was performed by a specially trained research assistant who delineated the WMH on the FLAIR images using the software tool Multi-image Analysis Graphical User Interface (MANGO) (24). Each transverse slice was analyzed separately, and the volume of WMH in mm³ was calculated (WMH area × 5 mm slice thickness) and combined across slices. In the main volumetric analyses, individuals with WMH graded as 0 on the Fazekas scale by both neuroradiologists were counted as not having WMH (0 mm³) on manual volumetry. Cerebral infarctions, perivascular spaces, and microbleeds were excluded from visual and manual volumetric measurement of WMH. Automated volumetry was based on white matter hypointensity volumes obtained from T1-weighted volumes using the FreeSurfer 4.50 suite (25). Limited WMH are common in older populations (10). Therefore, to compare those who probably had a clinically relevant amount of WMH to those with findings within normal limits, WMH were categorized as limited and extensive in the current study. This dichotomization also made no assumption of a linearity of the data. The cut-off score for the Fazekas scale for extensive WMH was set to ≥2 according to normal clinical practice (11). For the Scheltens scale and WMH volumetry, there is no consensus on what cut-off should be used to indicate an extensive amount of WMH. The cut-offs for extensive WMH were therefore a priori set to WMH load in the upper quartile, to have groups sizes that allow reasonable statistical power. Using a method based on the Statistical Parametric Mapping (SPM) software package, intracranial volumes were calculated using the adapted reverse and automatic reverse brain-masking techniques (26).

Headache classification

The headache classifications in HUNT3 were described in an earlier paper (16). In short, answers to HUNT3 questionnaires were used to categorize respondents into one of four validated (18) and mutually exclusive groups: Headache-free, TTH, migraine, or unclassified headache (Figure 1). As the “infrequent TTH” criteria (headache <1 day a month) had low specificity (18,27), only those reporting frequent or chronic (≥1 day/month) headache were categorized as having TTH in the main analyses. Those reporting infrequent TTH were categorized as having unclassified headache in the main analyses. Longitudinal headache classification was based on identical and validated headache screening questions in HUNT2 and HUNT3 (18,17). Respondents were classified into four mutually exclusive groups according to the temporal evolution of headache over the approximately 12-year period from HUNT2 to HUNT3: Stable headache-free (reporting freedom from headache in both surveys), past headache (headache in HUNT2 but not in HUNT3), new onset headache (no headache in HUNT2 but headache in HUNT3), and persistent headache (reporting headache in both surveys) (Figure 1).

Figure 1.

Flowchart describing the cohort.

Demographics, health-related information, and clinical measurements

Total Hospital Anxiety and Depression Scale (HADS) score and self-reported information on sex, employment status, alcohol habits, health, smoking, and history of ischemic heart disease (myocardial infarction and angina pectoris) were derived from HUNT3. A value above zero on the CAGE questionnaire (28) (a screening test for alcoholism) was counted as potential alcohol overuse. Body mass index, blood pressure, total serum cholesterol, serum triglycerides, and non-fasting serum glucose were measured at the medical examination that was a part of the HUNT surveys. Blood pressure was measured three times, with the mean of the two last measurements being used. The age used in the analyses was that at the time of MRI acquisition. Those with cerebral infarctions on examination of the MRI were categorized as having suffered a cerebral infarction (20).

Statistical methods

The means of the Scheltens scores of the two neuroradiologists were calculated for all individuals. The WMH volume was adjusted for intracranial volume using the following formula: Vol – b(ICV – meanICV), where Vol is the original WMH volume, ICV is the intracranial volume, b is the slope from the linear regression of Vol on ICV, and meanICV is the mean intracranial volume (29). In the main analyses, the odds of having extensive WMH were compared across headache groups using two binary logistic regression models. Both models included WMH measured with the Fazekas scale (<2 vs. ≥2), Scheltens scale (quartiles 1–3 vs. 4), or manual/automated volumetry (quartiles 1–3 vs. 4), entered as dichotomized dependent variables. All covariates were selected a priori based on prior research and entered in one step. Both models had headache status as a categorical covariate, and included age and sex as categorical covariates, as their association with headache (19) and WMH (10) meant that they were very likely to be confounders. Model B included additional covariates that were likely to be confounders as well, but where the theoretical evidence was less substantial than for model A (see footnote to Table 5). Individuals with missing information on a covariate were excluded from the analyses using that covariate. The number of individuals with missing information on each covariate is presented in tables (see footnote to Tables 2–3). For the sensitivity analyses, those analyses that included manual volumetry were repeated, with those individuals who were not subjected to manual measurement of WMH (Fazekas = 0) treated as missing. As cerebral infarctions were not manually excluded in analyses using automated volumetry, some cerebral infarctions could have been counted as WMH. These analyses were therefore repeated after excluding those with subcortical or cortical hemispheric infarctions. Cross-sectional analyses were also repeated, after placing those respondents with infrequent TTH into the TTH group, and after excluding those with medication-overuse headache. Tests of significance were two-tailed, and p values < 0.05 were considered significant. All statistical analyses were carried out using IBM SPSS Statistics version 23.

Table 2.

HUNT3 background data for respondents and non-respondents to the headache questionnaire, and for the different headache groups in the MRI study.

			Respondents by headache status over previous year
	Non- respondents	Respondents	Headache-free	TTH	Migraine	Unclassified
Number of participants per group	141	862	551	94^a	91^b	126
Demographic
Age in years (mean [SD])	57.9 [4.3]	58.6 [4.2]	58.8 [4.2]	57.9 [4.1]	57.7 [4.3]	58.6 [4.2]
Women (n [%])	78 [55.3]	451 [52.3]	260 [47.2]	54 [57.4]	67 [73.6]	70 [55.6]
Non-employed (n [%])	25 [17.9]	149 [17.4]	88 [16.1]	14 [15.1]	16 [18.0]	31 [24.6]
Health-related
Alcohol consumption (n [%])	7 [24.1]	150 [19.0]	111 [21.9]	11 [12.8]	12 [15.0]	16 [13.9]
Daily smoking (n [%])	28 [19.9]	133 [15.4]	83 [15.1]	14 [14.9]	18 [19.8]	18 [14.3]
Fair or poor health (n [%])	41 [30.1]	230 [27.1]	113 [21.0]	35 [37.2]	37 [41.1]	45 [35.7]
Ischemic heart disease (n [%])	3 [2.1]	28 [3.2]	21 [3.8]	2 [2.1]	0 [0.0]	5 [4.0]
Cerebral infarction (n [%])	4 [2.8]	39 [4.5]	29 [5.3]	1 [1.1]	4 [4.4]	5 [4.0]
HADS (mean [SD])	7.5 [6.1]	6.8 [5.3]	6.1 [4.9]	7.6 [5.7]	7.9 [5.7]	8.2 [5.8]
Clinical measurements
BMI (mean [SD])	27.3 [3.7]	26.9 [3.7]	27.0 [3.6]	27.1 [4.3]	26.6 [4.0]	26.8 [3.4]
SBP (mean [SD])	134.0 [17.6]	131.5 [16.8]	131.2 [16.2]	132.8 [17.8]	131.0 [18.5]	131.9 [17.8]
DBP (mean [SD])	78.2 [10.9]	76.0 [10.7]	75.8 [9.9]	77.4 [11.0]	74.6 [11.9]	77.1 [12.3]
Cholesterol (mean [SD])	6.0 [1.0]	5.7 [1.0]	5.7 [1.0]	5.7 [1.0]	5.9 [1.2]	5.7 [0.9]
Triglyceride (mean [SD])	1.8 [0.9]	1.6 [0.9]	1.7 [0.9]	1.6 [0.8]	1.5 [0.7]	1.7 [1.0]
Glucose (mean [SD])	5.6 [1.6]	5.6 [1.7]	5.7 [1.9]	5.6 [1.3]	5.3 [1.1]	5.5 [1.3]

Ninety subjects with TTH had headache 1–14 days per month, and four had headache > 14 days per month.

Forty-six subjects with migraine reported visual aura, and sixteen reported sensory aura.

TTH: tension-type headache ≥ 1 day per month; Alcohol consumption: positive response to a CAGE question; Fair or poor health: fair or poor self-perceived health; Ischemic heart disease: self-reported myocardial infarction or angina pectoris; HADS: Total Hospital Anxiety and Depressions Scale; BMI: body mass index; SBP: systolic blood pressure in mmHg; DBP: diastolic blood pressure in mmHg; Cholesterol: total serum cholesterol in mmol/L; Triglyceride: serum triglyceride in mmol/L; Glucose: non-fasting serum glucose in mmol/L.

Numbers and percentages of eligible individuals answering the headache screening questions (862) who did not answer the respective questions or did not undergo the measurement procedures: Employment status (n = 7, 0.8%), body mass index (n = 1, 0.1%), alcohol consumption (n = 73, 8.5%), self-reported health (n = 13, 1.5%), HADS (n = 6, 0.7%), mean systolic blood pressure (n = 9, 1.0%), mean diastolic blood pressure (n = 8, 0.9%), and cholesterol/triglycerides/glucose (n = 49, 5.7%). The question regarding daily smoking was asked in such a manner that missing data were most sensibly interpreted as a negative answer.

Table 3.

HUNT3 background data for respondents and non-respondents to the headache screening question in HUNT2 and HUNT3, and for the different longitudinal headache groups.

			Respondents by temporal headache evolution
	Did not answer both screening questions	Answered both screening questions	Stable headache-free	Past headache (only HUNT2)	New onset headache (only HUNT3)	Persistent headache (both HUNT2 and HUNT3)
Number in group	225	778	342	147	61	228
Demographic
Age in years (mean [SD])	58.3 [4.3]	58.5 [4.2]	58.8 [4.2]	58.7 [4.1]	58.3 [4.6]	58.0 [4.1]
Women (n [%])	124 [55.1]	405 [52.1]	135 [39.5]	92 [62.6]	31 [50.8]	147 [64.5]
Non-employed (n [%])	38 [17.0]	136 [17.6]	50 [14.7]	28 [19.3]	8 [13.3]	50 [22.1]
Health-related
Alcohol consumption (n [%])	20 [18.7]	137 [19.3]	78 [24.8]	21 [15.3]	13 [23.2]	25 [12.3]
Daily smoking (n [%])	45 [20.0]	116 [14.9]	48 [14.0]	21 [14.3]	11 [18.0]	36 [15.8]
Fair or poor health (n [%])	67 [30.6]	204 [26.6]	59 [17.7]	37 [25.5]	19 [31.1]	89 [39.2]
Ischemic heart disease (n [%])	5 [2.2]	26 [3.3]	12 [3.5]	7 [4.8]	2 [3.3]	5 [2.2]
Cerebral infarction (n [%])	8 [3.6]	35 [4.5]	18 [5.3]	7 [4.8]	1 [1.6]	9 [3.9]
HADS (mean [SD])	6.9 [5.3]	6.8 [5.3]	5.9 [4.7]	6.4 [5.1]	7.6 [5.4]	8.1 [5.9]
Clinical measurements
BMI (mean [SD])	27.2 [3.6]	26.9 [3.7]	27.1 [3.6]	26.9 [3.7]	27.1 [4.0]	26.6 [3.8]
SBP (mean [SD])	132.5 [16.7]	131.6 [17.0]	131.2 [16.0]	131.7 [17.0]	135.4 [19.0]	131.2 [18.0]
DBP (mean [SD])	77.5 [10.2]	76.0 [10.8]	76.4 [10.0]	74.6 [10.1]	78.5 [12.8]	75.6 [11.8]
Cholesterol (mean [SD])	5.9 [1.0]	5.7 [1.0]	5.7 [1.0]	5.7 [1.0]	5.8 [0.9]	5.8 [1.1]
Triglyceride (mean [SD])	1.7 [0.8]	1.7 [0.9]	1.7 [1.0]	1.6 [0.8]	1.6 [0.8]	1.6 [0.9]
Glucose (mean [SD])	5.6 [1.5]	5.6 [1.7]	5.7 [2.1]	5.6 [1.4]	5.6 [1.4]	5.4 [1.2]

Alcohol consumption: positive response on a CAGE question; Fair or poor health: fair or poor self-perceived health; Ischemic heart disease: self-reported myocardial infarction or angina pectoris; HADS: total Hospital Anxiety and Depressions Scale; BMI: body mass index; SBP: systolic blood pressure in mmHg; DBP: diastolic blood pressure in mmHg; Cholesterol: total serum cholesterol in mmol/L; Triglyceride: serum triglyceride in mmol/L; Glucose: non-fasting serum glucose in mmol/L.

Numbers and percentages of eligible individuals answering both headache screening questions (778) who did not answer the respective questions or did not undergo the measurement procedures: employment status (n = 7, 0.9%), body mass index (n = 1, 0.1%), alcohol consumption (n = 67, 8.6%), self-reported health (n = 12, 1.5%), HADS (n = 5, 0.6%), mean systolic blood pressure (n = 6, 0.8%), mean diastolic blood pressure (n = 5, 0.6%), cholesterol/triglycerides/glucose (n = 41, 5.3%). The question regarding daily smoking was asked in such a manner that missing data were most sensibly interpreted as a negative answer.

Table 4.

Prevalence of white matter hyperintensities according to headache groups.

	Headache status in HUNT3				Temporal evolution of headache from HUNT2 to 3
	Headache- free	TTH	Migraine	Unclassified	Stable headache- free	Past headache	New onset headache	Persistent headache
Fazekas score
0–1 (n [%])	508 [92.2]	79 [84.0]	82 [90.1]	114 [90.5]	318 [93.0]	131 [89.1]	53 [86.9]	202 [88.6]
2–3 (n [%])	43 [7.8]	15 [16.0]	9 [9.9]	12 [9.5]	24 [7.0]	16 [10.9]	8 [13.1]	26 [11.4]
Scheltens score
(Median [IQR])	3.0 [6.0]	3.5 [9.0]	2.5 [6.0]	2.8 [7.1]	2.5 [5.6]	3.0 [7.5]	3.0 [7.5]	2.5 [7.5]
1–3 quartile (n [%])	417 [75.7]	58 [61.7]	69 [75.8]	94 [74.6]	264 [77.2]	103 [70.1]	40 [65.6]	166 [72.8]
4 quartile (n [%])	134 [24.3]	36 [38.3]	22 [24.2]	32 [25.4]	78 [22.8]	44 [29.9]	21 [34.4]	62 [27.2]
Manual volumetry (mm³)
(Median [IQR])	69 [773]	312 [1361]	63 [806]	226 [806]	50 [735]	220 [944]	265 [1174]	119 [896]
1–3 quartile (n [%])	391 [75.9]	59 [67.0]	67 [77.0]	92 [75.4]	250 [78.4]	99 [71.7]	41 [68.3]	159 [73.6]
4 quartile (n [%])	124 [24.1]	29 [33.0]	20 [23.0]	30 [24.6]	69 [21.6]	39 [28.3]	19 [31.7]	57 [26.4]
Automated volumetry (mm³)
(Median [IQR])	2004 [1033]	2166 [1570]	2044 [1111]	2099 [976]	1980 [965]	2077 [1120]	2096 [1436]	2117 [1213]
1–3 quartile (n [%])	402 [76.9]	56 [62.2]	67 [75.3]	94 [76.4]	256 [79.3]	102 [72.3]	39 [65.0]	162 [73.3]
4 quartile (n [%])	121 [23.1]	34 [37.8]	22 [24.7]	29 [23.6]	67 [20.7]	39 [27.7]	21 [35.0]	59 [26.7]

TTH: Tension-type headache ≥1 day per month; IQR: Interquartile range.

Note: Due to missing responses, the cut-offs defining the upper quartile varied slightly for cross-sectional vs. longitudinal data for manual volumetry (≥834 mm³ vs. ≥847 mm³), and automated volumetry (≥2680 mm³ vs. ≥2682 mm³), but cut-offs were identical for cross-sectional and longitudinal data for the Scheltens score (≥7).

Table 5.

Headache status and odds ratios for extensive white matter hyperintensities.

	Fazekas score 0–1 vs. 2–3		Scheltens score quartile 1–3 vs. 4		Manual volumetry quartile 1–3 vs. 4		Automated volumetry quartile 1–3 vs. 4
	Model A^a	Model B^b	Model A^a	Model B^b	Model A^a	Model B^b	Model A^a	Model B^b
Headache-free	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
TTH
Odds ratio	2.41	3.07	2.18	2.46	1.74	2.00	2.49	2.64
95% CI	1.24–4.65	1.51–6.21	1.34–3.56	1.44–4.20	1.04–2.92	1.13–3.55	1.50–4.14	1.49–4.67
p value	0.009^c	0.002^c	0.002^c	0.001^c	0.037^c	0.017^c	<0.001^c	0.001^c
Migraine
Odds ratio	1.26	1.14	1.02	0.76	0.96	0.80	1.25	1.39
95% CI	0.57–2.80	0.46–2.84	0.59–1.77	0.39–1.47	0.54–1.72	0.40–1.60	0.71–2-19	0.72–2.66
p value	0.56	0.78	0.94	0.41	0.90	0.54	0.43	0.33
Unclassified headache
Odds ratio	1.28	1.65	1.08	1.23	1.07	1.33	1.06	1.19
95% CI	0.64–2.55	0.79–3.44	0.68–1.72	0.73–2.09	0.66–1.74	0.77–2.31	0.65–1.72	0.67–2.10
p value	0.49	0.18	0.75	0.43	0.78	0.30	0.82	0.55

TTH: Tension-type headache ≥ 1 day per month.

White matter hyperintensities measured with Fazekas or Scheltens scale, or manually, or automated as dependent variable. Headache status (headache-free as reference), age, and sex included as categorical covariates.

As ^a, but also with daily smoking, responses to CAGE questions, ischemic heart disease, and cerebral infarction included as categorical covariates, and total HADS score, total serum cholesterol, serum triglyceride, non-fasting serum glucose, body mass index, and mean systolic blood pressure included as continuous covariates.

p value < 0.05.

Ethical approval

The study was approved by the Norwegian Data Inspectorate, the Norwegian Board of Health, and the Regional Committee for Ethics in Medical Research. All participants provided written informed consent.

Results

Subjects responding to the headache questionnaire in HUNT3

Of the 862 MS-free respondents to the HUNT3 headache questionnaire, 11% had TTH ≥ 1 day per month, 11% had migraine, and 15% had unclassified headache (Table 2). Of the 126 with unclassified headache, eight had medication-overuse headache and 52 had TTH < 1 day per month. Compared with the respondents to the HUNT3 headache questionnaire, the 141 individuals who did not answer the headache questions (non-respondents) were somewhat more likely to smoke and to give a positive answer to the CAGE questions (Table 2). Compared with the headache-free subjects, those reporting headache showed a larger proportion of females, reported worse general health, and were less likely to give a positive answer to CAGE questions (Table 2).

Subjects responding to the headache questionnaires in both HUNT2 and HUNT3

Of the 778 MS-free individuals answering the headache screening question in both HUNT2 and HUNT3, 44% were classified as stable headache-free, 19% as having past headache, 8% as having new onset headache, and 29% as having persistent headache (Table 3). The respondents who answered the headache questionnaires in both HUNT2 and HUNT3 reported better general health and a lower number of them smoked in comparison with the 225 subjects who did not answer both headache questionnaires. Stable headache-free subjects were more likely to be male and to report good general health than those having headache at any time point (Table 3).

Measurement of WMH

The WMH were graded with the Fazekas and Scheltens scales for all respondents (Figure 2). Thirty seven individuals were excluded from both automated and manual WMH volumetry because of motion or image artifacts. Automated measurements were therefore obtained in 825 individuals. Manual measurements could not be performed on a further 13 individuals, as FLAIR images were either not available or of poor quality. Of the remaining 812 individuals, manual volumetry was performed on 407 individuals who received a Fazekas score ≥1 from at least one neuroradiologist.

Figure 2.

Examples of white matter hyperintensities of Fazekas grade 1–3.

Cross-sectional headache data and WMH

Individuals with TTH had the highest prevalence of extensive WMH of all the groups (Table 4). In both the A and B regression models (Table 5), those with TTH had increased odds of having extensive WMH compared with headache-free subjects, with this being the case across all measuring methods: Fazekas odds ratio (OR), 3.07 and 95% CI, 1.51–6.21; Scheltens OR, 2.46 and 95% CI, 1.44–4.20; manual volumetry OR, 2.00 and 95% CI, 1.13–3.55; automated volumetry OR, 2.64 and 95% CI, 1.49–4.67. No association was found between the other headache types and extensive WMH in the main analyses (Table 5). In the sub-analyses of deep and periventricular WMH (Supplemental Tables 1–2), individuals with TTH had increased odds of extensive WMH in the deep (OR, 2.01; 95% CI, 1.18–3.44), but not periventricular, white matter. Respondents with migraine were less likely to have extensive periventricular WMH than those who were headache-free (OR, 0.48; 95% CI, 0.24–0.98). Sensitivity analyses treating participants without manual volumetry (Fazekas = 0) as missing, and analyses including those with infrequent TTH in the TTH group, produced similar results to the main analyses (Supplemental Tables 2–4). Excluding 25 individuals with hemispheric infarctions from analyses using automated volumetry, and excluding eight individuals with medication-overuse headache from cross-sectional analyses, did not change the main results substantially (data not shown).

Longitudinal headache data and WMH

Individuals with new onset headache between HUNT2 and HUNT3 had a higher prevalence of extensive WMH than those with persisting headache, past headache, and those who were stable headache-free (Table 4). In some of the regression models (Table 6), subjects with new onset headache had increased odds of having extensive WMH compared with those who were stable headache-free: Scheltens OR, 2.24; 95% CI, 1.13–4.44; automated volumetry OR, 2.74; 95% CI, 1.35–5.57. Sub-analyses on deep and periventricular WMH showed that individuals with past headache had increased odds of having extensive periventricular WMH (OR, 1.65; 95% CI, 1.00–2.73) compared with headache-free subjects (Supplemental Table 5). Excluding 21 individuals with hemispheric infarctions from analyses using automated volumetry produced similar results to the main analyses (data not shown).

Table 6.

Longitudinal headache groups and odds ratios for extensive white matter hyperintensities.

	Fazekas score 0–1 vs. 2–3		Scheltens score quartile 1–3 vs. 4		Manual volumetry quartile 1–3 vs. 4		Automated volumetry quartile 1–3 vs. 4
	Model A^a	Model B^b	Model A^a	Model B^b	Model A^a	Model B^b	Model A^a	Model B^b
Stable headache-free	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Past headache
Odds ratio	1.42	1.20	1.29	1.29	1.36	1.40	1.49	1.71
95% CI	0.71–2.83	0.54–2.65	0.82–2.05	0.76–2.18	0.84–2.22	0.80–2.45	0.92–2.41	0.98–2.99
p value	0.33	0.65	0.27	0.35	0.21	0.24	0.11	0.057
New onset headache
Odds ratio	2.10	2.40	1.89	2.24	1.87	2.05	2.42	2.74
95% CI	0.88–5.05	0.95–6.06	1.02–3.50	1.13–4.44	0.99–3.55	1.00–4.19	1.29–4.55	1.35–5.57
p value	0.096	0.064	0.042^c	0.020^c	0.056	0.051	0.006^c	0.005^c
Persistent headache
Odds ratio	1.67	1.79	1.24	1.20	1.33	1.47	1.51	1.63
95% CI	0.90–3.10	0.90–3.55	0.82–1.88	0.75–1.94	0.86–2.06	0.89–2.44	0.98–2.32	0.98–2.73
p value	0.10	0.095	0.31	0.45	0.21	0.14	0.064	0.060

White matter hyperintensities measured with Fazekas or Scheltens scale, or manually, or automated as dependent variable. Temporal evolution of headache (stable headache-free as reference), age, and sex included as categorical covariates.

p value < 0.05.

Discussion

The main findings of this study were that individuals with TTH had more extensive WMH than those who were headache-free, while individuals with migraine or unclassified headache did not have more extensive WMH than headache-free subjects. Analyses using longitudinal headache data revealed that those reporting onset of headache in adult life had more extensive WMH than individuals who were stable headache-free. The results were similar when using semi-quantitative and volumetric methods of measuring WMH.

Comparison with earlier studies on WMH in headache populations

Our findings with regard to the association between extensive WMH and TTH are in line with a case-control study that specifically included individuals with TTH (3), as well as a population-based study on non-migrainous headache (7). However, they differ from another population-based study that did not find a difference in WMH occurrence between people with non-migrainous headache and those who were headache-free (8). Our finding of a lack of association between migraine and WMH is in line with another population-based study (30), but differs from four other population-based studies that did report more WMH in individuals with migraine (4,6 –8). The age ranges in these previous studies and the present study are quite similar and cannot explain the discrepant results. There were, however, differences in sample size, and previous studies that detected an association between WMH and migraine included more people with migraine than in this study (4,6 –8). Therefore, we may have lacked the power to detect a potentially increased risk of WMH in people with migraine. Nevertheless, the sizes of the TTH and migraine groups were similar, and we found a significant association between TTH and WMH, and not between migraine and WMH. It therefore appears that WMH are more strongly associated to TTH than to migraine. This underlines the importance of including individuals with non-migrainous headache in future research.

We found that WMH had a stronger and more consistent association with new onset headache (after the age of 40) than with persistent headache. This suggests that headache starting in adulthood have a greater impact on white matter than headache starting in younger years, rather than there being a dose-response relationship between headache burden and WMH over time. This finding is at variance with the previously reported WMH progression in people with migraine (6,15), which suggests a dose-response relationship between headache and WMH. However, others have also reported that WMH are stable in individuals with migraine (8). The current findings imply that different pathophysiological mechanisms may be involved in headache during the lifespan. Headaches that start later in life are more likely to be secondary to other diseases, which may explain their relation to WMH.

Potential pathophysiological mechanisms

WMH may have a vascular etiology (12), and cerebral ischemia is usually highlighted as a potential cause of WMH in individuals with headache (4,6,7). Interestingly, it has also been reported that those who experience migraine are more likely to be affected by cerebral infarctions than those who are headache-free (4,7,31), but this finding is not consistent across all populations (32). The finding of more extensive WMH in TTH in the current and previous studies (3,7) suggests that ischemic and inflammatory processes in white matter are either caused by or accompany TTH. The fact that WMH in this study were more pronounced in deep white matter indicates a vascular cause. However, vascular factors are currently not viewed as a major factor in the pathophysiology of TTH. Furthermore, we attempted to adjust for cardiovascular risk factors in the present study, which supports the notion that WMH in TTH is not associated with cardiovascular health in general. WMH may also have non-ischemic causes such as degenerative changes, blood-brain barrier leakage, inflammation, and amyloid angiopathy, as seen with histopathological investigations of WMH in autopsy studies (12).

WMH measuring methods

In the present study, semi-quantitative scales and volumetric methods provided very similar results. Our findings imply that the easily available semi-quantitative methods for WMH assessment, which are often a standard part of the neuroradiological reading above a certain age, are acceptable measures to use in cross-sectional headache research. The more advanced Scheltens scale provides additional information by separating lesions into periventricular and deep WMH, which can improve accuracy since the latter are associated with ischemic histopathology and molecular markers of hypoxic injury (12). Analyses based on automated volumetry produced higher odd ratios than those based on manual volumetry. Automated volumetry was performed on more individuals than manual volumetry, which may explain the difference in the results. However, we would like to point out that the results were overall comparable, and that, in general, these methods are difficult to compare as they estimate WMH in different ways. WMH measuring methods cannot explain the difference in results between the present study and the previous studies, as similar methods used in earlier studies were explored in the current study, that is, automated volumetric methods (6 –8,30) and visual grading (3,4,6,8).

Clinical relevance

WMH burden is associated with several negative outcomes in aging (14), and the mechanisms behind the association between WMH and TTH need to be elucidated to optimize population brain health. The previous finding in HUNT of a significantly increased risk of later dementia in people with non-migrainous headache further supports the notion that this headache type is linked to detrimental pathophysiological changes in the brain and needs closer scrutiny (33). As there are no established strategies on how to manage WMH, the present results do not imply that brain MRI is warranted in individuals with headache to map WMH. However, correcting cardiovascular risk factors have been shown to delay the progression of WMH (34), and treatment strategies may emerge in the future. In that case, having TTH or late onset headache should strengthen the indication of performing an MRI examination. Since headache affects approximately 50% of the adult population during the course of a single year (35), the results of this study have public health relevance.

Strengths and limitations

This study has several strengths. It has a population-based design and includes a large number of participants. The study population is representative of the middle-aged Norwegian population (21), and the results are therefore generalizable to similar populations. Furthermore, participants in the HUNT MRI study were selected at random, independently of headache status, and the questionnaires allowed for categorization of respondents into headache groups on the basis of validated primary headache diagnoses (18,27,17). Also, documentation of headache at two time points is probably more reliable than a single retrospective description of both periods. The MRI protocol was standardized, and several methods for measuring WMH were used in the analyses. Manual volumetry gives the most accurate measurement of WMH volume, and this is the first population-based study where manual volumetry was used to investigate associations between WMH and headache.

The study also has several limitations. Participants had fewer cardiovascular risk factors than those who did not participate (21). This could result in a systematical underestimation of the prevalence of WMH (13). However, the differences in risk factors, and therefore their potential effect on our estimates, were small (21). There is no consensus on what cut-off should be used to indicate an extensive amount of WMH when using volumetry and Scheltens scale, which make it difficult to compare results across studies. Continuous data was dichotomized based on the assumption that limited and extensive WMH have different clinical relevance. This approach leads to a loss of data, which reduces statistical power and limits our potential to correct for confounding factors. As some reports indicate that the association between WMH and migraine is specific for women (4,6), it would have been interesting to stratify for sex. However, as only 24 men had migraine, this sub-analysis would have low power in the current cohort. The measurement of headache by a self-administered questionnaire is inferior to a clinical interview performed by a trained physician, although the headache criteria had fair sensitivity and specificity (18,27). The migraine diagnosis was highly specific (95%), but had a sensitivity of only 50% (18). Therefore, some respondents with migraine were probably categorized as having TTH or unclassified headache. Our questionnaire also allowed for only one primary headache diagnosis per respondent. If WMH have a dose-response relationship with duration of pain, selecting those with TTH ≥ 1 day/month may exaggerate the association between TTH and WMH. However, sensitivity analyses that included those with infrequent TTH in the TTH group did not give different results. In the main analyses, each headache group was compared with the headache-free group using two models across four methods of measuring WMH. Performing multiple comparisons raises the likelihood of making type I errors. We have tried to make a judicious evaluation of the results instead of performing a correcting procedure that increases the likelihood of performing type II errors. The association between TTH and WMH appears to be genuine, as only TTH was significantly associated with WMH in the cross-sectional analyses. Furthermore, this association was consistently observed in every analysis, making it unlikely to be a spurious finding. In longitudinal analyses, associations were less clear, and some could represent spurious findings. The modest number of individuals with TTH and migraine reduces the statistical power of some analyses. This could also mean that model B is overfitted, which limits our potential to correct for confounding factors.

Conclusion

In this population-based neuroimaging study, individuals with TTH and those who developed a new headache in adulthood had more extensive WMH than headache-free subjects. These results were similar across all methods of assessing WMH. Our findings suggest that these headaches are accompanied by structural alterations in the brain. This underlines the importance of differentiating between headache sub-types and taking changes in headache occurrence over time into consideration when future investigations into the relationship between WMH and headache are undertaken.

Supplemental Material

Supplementary tables -Supplemental material for White matter hyperintensities and headache: A population-based imaging study (HUNT MRI)

Supplemental material, Supplementary tables for White matter hyperintensities and headache: A population-based imaging study (HUNT MRI) by Lasse-Marius Honningsvåg, Asta Kristine Håberg, Knut Hagen, Kjell Arne Kvistad, Lars Jacob Stovner and Mattias Linde in Cephalalgia

Footnotes

Article highlights

Having tension-type headache or developing headache in adulthood were linked to the occurrence of extensive white matter hyperintensities.

The increased risk of extensive white matter hyperintensities was observed across all methods of WMH assessment.

There were no associations between white matter hyperintensities and migraine or unclassified headache.

Acknowledgements

The staff at the Department of Radiology, Levanger hospital, performed the MRIs. Neuroradiologists Kjell Arne Kvistad and Jana Rydland at the Department of Radiology and Nuclear Medicine read all MRIs. Research assistant Axel Kvistad manually delineated the WMH. The Nord-Trøndelag Health Study (HUNT) is a collaboration between the HUNT Research Centre, the Faculty of Medicine at the Norwegian University of Science and Technology (NTNU), the Norwegian Institute of Public Health, and the Nord-Trøndelag County Council.

Declaration of conflicting interests

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

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors, but was supported by NTNU Norwegian University of Science and Technology; National Norwegian Advisory Unit for functional MRI; and St. Olav's University Hospital.

References

Vos

Flaxman

Naghavi

et al.

Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: A systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012; 380: 2163–2196.

Headache Classification Committee of the International Headache Society (IHS). The International Classification of Headache Disorders, 3rd edition (beta version). Cephalalgia 2013; 33: 629–808.

De Benedittis

Lorenzetti

Sina

et al.

Magnetic resonance imaging in migraine and tension-type headache. Headache 1995; 35: 264–268.

Kruit

van Buchem

Hofman

PAM

et al.

Migraine as a risk factor for subclinical brain lesions. JAMA 2004; 291: 427–434.

Kruit

Launer

Ferrari

et al.

Brain stem and cerebellar hyperintense lesions in migraine. Stroke 2006; 37: 1109–1112.

Palm-Meinders

Koppen

Terwindt

et al.

Structural brain changes in migraine. JAMA 2012; 308: 1889–1897.

Kurth

Mohamed

Maillard

et al.

Headache, migraine, and structural brain lesions and function: Population based Epidemiology of Vascular Ageing-MRI study. BMJ 2011; 342: c7357.

Hamedani

Rose

Peterlin

et al.

Migraine and white matter hyperintensities: The ARIC MRI study. Neurology 2013; 81: 1308–1313.

Wardlaw

Smith

Biessels

et al.

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

10.

de Leeuw

de Groot

Achten

et al.

Prevalence of cerebral white matter lesions in elderly people: A population based magnetic resonance imaging study. The Rotterdam Scan Study. J Neurol Neurosurg Psychiatry 2001; 70: 9–14.

11.

Wahlund

L-O

Westman

Van Westen

et al.

Structural brain imaging may improve diagnostics in dementia. Lakartidningen 2013; 110: 2116–2118.

12.

Gouw

Seewann

van der Flier

et al.

Heterogeneity of small vessel disease: A systematic review of MRI and histopathology correlations. J Neurol Neurosurg Psychiatry 2011; 82: 126–135.

13.

van Dijk

Breteler

MMB

Schmidt

et al.

The association between blood pressure, hypertension, and cerebral white matter lesions: Cardiovascular determinants of dementia study. Hypertension 2004; 44: 625–630.

14.

Debette

Markus

. The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: systematic review and meta-analysis. BMJ 2010; 341: c3666.

15.

Erdélyi-Bótor

Aradi

Kamson

et al.

Changes of migraine-related white matter hyperintensities after 3 years: A longitudinal MRI study. Headache 2015; 55: 55–70.

16.

Honningsvåg

L-M

Hagen

Håberg

et al.

Intracranial abnormalities and headache: A population-based imaging study (HUNT MRI). Cephalalgia 2016; 36: 113–121.

17.

Hagen

Zwart

Vatten

et al.

Head-HUNT: Validity and reliability of a headache questionnaire in a large population-based study in Norway. Cephalalgia 2000; 20: 244–251.

18.

Hagen

Zwart

J-A

Aamodt

et al.

The validity of questionnaire-based diagnoses: The third Nord-Trøndelag Health Study 2006–2008. J Headache Pain 2010; 11: 67–73.

19.

Linde

Stovner

Zwart

J-A

et al.

Time trends in the prevalence of headache disorders. The Nord-Trondelag Health Studies (HUNT 2 and HUNT 3). Cephalalgia 2011; 31: 585–596.

20.

Håberg

Hammer

Kvistad

et al.

Incidental intracranial findings and their clinical impact; the HUNT MRI study in a general population of 1006 participants between 50–66 Years. PLoS One 2016; 11: e0151080.

21.

Honningsvåg

L-M

Linde

Håberg

et al.

Does health differ between participants and non-participants in the MRI-HUNT study, a population based neuroimaging study? The Nord-Trøndelag health studies 1984–2009. BMC Med Imaging 2012; 12: 23.

22.

Fazekas

Chawluk

Alavi

et al.

MR signal abnormalities at 1.5 T in Alzheimer's dementia and normal aging. AJR Am J Roentgenol 1987; 149: 351–356.

23.

Scheltens

Barkhof

Leys

et al.

A semiquantative rating scale for the assessment of signal hyperintensities on magnetic resonance imaging. J Neurol Sci 1993; 114: 7–12.

24.

Lancaster JL and Martinez MJ. Mango: Multi-image Analysis GUI, version 3.0.4.http://ric.uthscsa.edu/mango/ (accessed 29 March 2017).

25.

FreeSurfer. FreeSurfer 4.50http://surfer.nmr.mgh.harvard.edu/ (accessed 29 March 2017).

26.

Hansen

Brezova

Eikenes

et al.

How does the accuracy of intracranial volume measurements affect normalized brain volumes? Sample size estimates based on 966 subjects from the HUNT MRI cohort. AJNR Am J Neuroradiol 2015; 36: 1450–1456.

27.

Hagen

Zwart

J-A

Aamodt

et al.

A face-to-face interview of participants in HUNT 3: The impact of the screening question on headache prevalence. J Headache Pain 2008; 9: 289–294.

28.

Mayfield

McLeod

Hall

. The CAGE questionnaire: Validation of a new alcoholism screening instrument. Am J Psychiatry 1974; 131: 1121–1123.

29.

Buckner

Head

Parker

et al.

A unified approach for morphometric and functional data analysis in young, old, and demented adults using automated atlas-based head size normalization: Reliability and validation against manual measurement of total intracranial volume. Neuroimage 2004; 23: 724–738.

30.

Monteith

Gardener

Rundek

et al.

Migraine, white matter hyperintensities, and subclinical brain infarction in a diverse community: The northern Manhattan study. Stroke 2014; 45: 1830–1832.

31.

Schürks

Rist

Bigal

et al.

Migraine and cardiovascular disease: Systematic review and meta-analysis. BMJ 2009; 339: b3914.

32.

Gaist

Garde

Blaabjerg

et al.

Migraine with aura and risk of silent brain infarcts and white matter hyperintensities: An MRI study. Brain 2016; 139: 2015–2023.

33.

Hagen

Stordal

Linde

et al.

Headache as a risk factor for dementia: A prospective population-based study. Cephalalgia 2014; 34: 327–335.

34.

Dufouil

Chalmers

Coskun

et al.

Effects of blood pressure lowering on cerebral white matter hyperintensities in patients with stroke: The PROGRESS (Perindopril Protection Against Recurrent Stroke Study) Magnetic Resonance Imaging Substudy. Circulation 2005; 112: 1644–1650.

35.

Stovner

Andree

. Prevalence of headache in Europe: A review for the Eurolight project. J Headache Pain 2010; 11: 289–299.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.27 MB