Perivascular spaces and headache: A population-based imaging study (HUNT-MRI)

Abstract

Background

In four previous clinic-based MRI studies headache sufferers (in particular migraineurs) had more perivascular spaces (PVS) than individuals who were headache-free.

Methods

The present study was part of a large longitudinal, epidemiological study (Nord-Trøndelag Health Survey (HUNT)). The 1006 participants, age 50–65 years at inclusion, had participated in all previous HUNT surveys (1–3), and been randomly selected to a population-based imaging study of the head (HUNT-MRI, 2007–2009). The number of visible PVS in the basal ganglia (BG) and hemispheric white matter (HWM) was compared in headache sufferers (migraine with and without aura, non-migrainous headache) and people who were headache-free.

Results

The results showed in general small differences between headache sufferers and headache-free participants. In the cross-sectional analysis migraineurs without aura had fewer PVS than headache-free individuals in BG (OR = 0.84, 95% CI = 0.76--0.94, p value = 0.003) and in BG and HWM together (OR = 0.97, 95% CI = 0.95-1.00, p value = 0.046). No difference between long-term headache sufferers and long-term headache-free individuals with regard to number of PVS was found.

Discussion

In contrast to previous studies, the present large, blinded, population-based study showed no increase in number of dilated PVS among headache sufferers. Fewer PVS among those with migraine without aura may be a spurious finding.

Keywords

Virchow-Robin spaces headache migraine population-based magnetic resonance imaging epidemiology

Background

Migraine and headache have both been associated with increased number of dilated perivascular spaces (PVS) on magnetic resonance imaging (MRI) in four clinic-based studies (1 –4). Three studies investigated PVS among children and the fourth among adults as well. Two studies (1,2) found a positive association between migraine and PVS, the third (3) between headache and PVS and the fourth (4) between both headache and migraine and PVS. Since migraine may be associated with neurogenic perivascular inflammation (5) and PVS are linked to lymphatic drainage of the brain parenchyma (6), it seems plausible that migraine could increase the number of PVS visible on MRI.

PVS, also referred to as Virchow-Robin spaces, are small cavities surrounding the walls of blood vessels as they course from the subarachnoid space and into the brain parenchyma (7 –11). They are commonly seen on MRI as signal intensities similar to those of cerebrospinal fluid (CSF) (12) (Figure 1). Small PVS (<2 mm in cross-section) are thought to be a normal phenomenon (13), and there is a positive correlation between age and the size and number of these spaces (13 –16). It is usual to divide PVS into three different categories according to localisation: around the medullary arteries entering the cortex over the high convexities projecting down into the hemispheric white matter (HWM) (16), around the lenticulostriate arteries as they perforate the anterior border of the basal ganglia (BG) (11,16) and in the mesencephalon (17,18). This study included PVS in BG and HWM. Three of the previous studies investigated only PVS in the HWM (1 –3) whereas the fourth investigated both BG and HWM (4).

Figure 1.

Perivascular spaces (PVS) in the hemispheric white (HWM) matter visible on magnetic resonance imaging (MRI).

The aim of the present study was to investigate whether the association between headache and PVS could be demonstrated on MRI in a relatively large population-based study, where longitudinal data on headache status were available, using a blinded design. Based on the previous studies and the known involvement of the vascular system in migraine, we hypothesised that the association would be stronger for migraine than non-migrainous headache.

Material and methods

Ethical approval

The Norwegian Data Inspectorate, the Norwegian Board of Health, and the Regional Committee for ethics in Medical Research had approved all studies. All participants gave their informed, written consent.

Cohort

The participants were drawn from the Nord-Trøndelag Health Study (HUNT). HUNT is a general population study of the entire population aged > 20 years in Nord Trøndelag County, Norway. Surveys with collection of a wide range of health-related data from questionnaires and clinical measures (e.g. blood samples, blood pressure) have been conducted in 1984 to 1986 (HUNT1), 1995 to 1997 (HUNT2) and 2006 to 2008 (HUNT3). As part of HUNT3 a group of 1006 individuals (530 women), fulfilling inclusion criteria, were randomly selected to be imaged with a standardised MRI protocol (HUNT-MRI). The inclusion criteria were age between 50 and 65 years at time of consent, previous participation in HUNT1, 2 and 3, and living maximally 45 minutes by car or public transport from Levanger hospital where the scanning was performed. Exclusion criteria were limited to standard MRI safety measures including pacemaker, severe claustrophobia and weight above 150 kg. Details about the recruitment of participants to the MRI study have been published previously (19).

Of the 1006 people who participated in HUNT-MRI, 864 (86%) had answered the headache questionnaire in HUNT3 and were included in a cross-sectional study. A total of 780 (78%) of the participants in HUNT-MRI had answered both the headache questionnaire in HUNT2 and 3, making it possible to compare long-term headache sufferers and long-term headache-free participants with regard to number of PVS.

MRI scanning

All imaging was performed on the same 1.5 T General Electric Signa HDx 1.5 T MRI scanner equipped with an eight-channel head coil and software version pre-14.0 M (GE Healthcare). The following protocol was used: a T1-weighted magnetisation prepared rapid gradient echo (MPRAGE) volume, transverse T2, T2*, a fluid-attenuated inversion recovery (FLAIR) sequence, a time-of-flight three-dimensional angio sequence through the base of the brain and diffusion tensor imaging.

MRI analysis

Number of visible PVS was assessed based on the MPRAGE volume, and the transaxial T2 and FLAIR images. In order to obtain reliable counting of PVS the MPRAGE volume was used for counting PVS while the transaxial T2 and FLAIR images were used to verify whether a possible PVS on the MPRAGE was a PVS or represented pathology. The FLAIR modality was used to distinguish PVS from white matter abnormalities, and FLAIR and T2 scans to help in the differentiation between PVS and lacunar infarctions. For each participant the entire MPRAGE volume was scrolled through while viewing the images in three planes to visualise the PVS and to avoid counting one PVS more than once due to many PVS’ tortuous course. To further facilitate the PVS count, the MRI readings of two experienced neuroradiologists who had reviewed the images concerning cysts, tumour, lacunar infarctions, contusions, etc., were consulted. All visible PVS with signal intensities similar to CSF were counted separately for BG and HWM. The sum of PVS from BG and HWM was also calculated to constitute the total number of PVS. One postgraduate medical student who was blinded with regard to the headache status of the individuals performed the PVS counting. The diameter of the spaces was not measured.

Headache diagnoses

For the cross-sectional analysis of PVS in HUNT3, the four mutually exclusive categories (see Table 1) were made for migraine, tension-type headache (TTH) ≥ 1 day per month, unclassified headache and being headache-free (defined as not suffering from headache during the past year). The migraine and TTH diagnoses were based on the criteria of the second edition of the International Classification of Headache Disorder (ICHD-II). The questionnaire did not allow individuals to have several diagnoses. From a study evaluating the screening question used in the questionnaire with face-to-face interviews of a random sample of participants, it is known that 7.1% of the migraineurs also had TTH (20). The classification and validity of the questionnaire-based diagnoses has been described in detail previously (21). For TTH sensitivity was 96% and specificity 69% and for migraine sensitivity was 51% and specificity 95%, making the diagnosis suitable to evaluate factors suspected to be associated with migraine. Furthermore the migraineurs were categorised as having either migraine with or without aura based on the question ‘Have you experienced transient visual disturbances before or during the headache?’ The migraine with aura diagnosis had a sensitivity of 50% and specificity of 95%. To test for the possibility that number of PVS might be related to persistency of headache, those having headache, migraine and non-migrainous headache in both HUNT2 and 3 were compared to those who were headache-free in both surveys. Those not suffering from headache the past year in both HUNT2 and 3 were categorised as headache-free in this analysis. For migraine, the liberal criteria from the validation studies (21,22) were used showing a sensitivity of 49%, specificity of 96% and a Cohen’s kappa of 0.51 in HUNT3. In HUNT2 the criteria showed a positive predictive value of 84%, a negative predictive value of 78% and a Cohen’s kappa of 0.59. These criteria were used to make uniform criteria because the headache questionnaire in HUNT2 lacked a question regarding the intensity of the headache. Individuals reporting headache but not fulfilling the criteria for migraine were categorised as non-migraineurs, showing a sensitivity of 61% and a specificity of 81% in HUNT2 (22).

Table 1.

Headache diagnoses in HUNT2 and HUNT3 in the HUNT-MRI population.

		Both genders	Male	Female	Mean age at scanning (SD)
Cross-sectional analyses^a
Answering headache questionnaire HUNT3		864	412	452	58.56 (4.21)
Headache-free individuals		552	292	260	58.81 (4.19)
Headache sufferers		312	120	192	58.12 (4.20)
Migraine^c		91	24	67	57.74 (4.34)
TTH		94	40	54	57.94 (4.08)
Unclassified headache		127	56	71	58.53 (4.18)
Longitudinal analyses^b
Answering both headache questionnaires		780	374	406	58.51 (4.19)
HUNT2	HUNT3
Headache-free	Headache-free	342	207	135	58.80 (4.19)
Headache	Headache	229	81	148	57.99 (4.13)
Migraine^c	Migraine^c	69	16	41	57.63 (4.37)
Non-migrainous headache	Non-migrainous headache	101	47	61	58.05 (3.92)

The cross-sectional analysis only included individuals answering the headache questionnaire in HUNT3. ^bThe longitudinal analysis included only individuals answering the headache questionnaire in both HUNT2 and HUNT3. TTH: tension-type headache; MRI: magnetic resonance imaging; HUNT: Nord-Trøndelag Health Survey. ^cIn the cross-sectional analysis the restrictive migraine criteria from the validation study were used whereas the liberal criteria were used in the longitudinal analysis.

Potential confounders

All participants in HUNT2 and 3 had their blood pressure measured three times in each survey. The mean arterial blood pressure was calculated as a mean of the second and third measurement. The participants had also, in both surveys, reported their alcohol consumption. In HUNT2 the variable was continuous reporting the frequency of drinking during a month, whereas in HUNT3 it was categorically divided into eight different groups reporting the frequency of alcohol consumption the last year (group 1: four to seven times a week, group 2: two to three times a week, group 3: once a week, group 4: two to three times a month, group 5: once a month, group 6: few times last year, group 7: never within last year, group 8: never consumed alcohol).

Statistics

The relationship between PVS and headache was examined with binary logistic regression using odds ratios (ORs) with 95% confidence interval (CI). The model had headache status as a dependent variable and the number of PVS (continuous variable) as covariate. In addition age at time of the scanning (continuous), gender, alcohol consumption and blood pressure (continuous) were included as covariates. These adjustments were made because it has been demonstrated that both headache and PVS are associated with age (13 –16,23), gender (16,23), alcohol consumption (24,25) and blood pressure (15,26). For the longitudinal analysis of long-term headache sufferers and long-term headache-free individuals, blood pressure and alcohol consumption at baseline, i.e. HUNT2, were used, whereas in the cross-sectional analysis blood pressure and alcohol consumption in HUNT3 were used. Ordinal regression was used to relate the number of PVS to headache frequency in HUNT3 because the variable representing days with headache in HUNT3 was ordinal (<1 day/month, one to six days, seven to 14 days, >14 days). Two-tailed p values <0.05 (not corrected for multiple comparisons) were considered significant. SPSS version 21 (SPSS IBM, NY, USA) was used for all analyses.

Results

PVS visible on MRI were found in 997 of the 1006 participants. In the remaining nine participants (three women) the MRI quality was too poor for acceptable PVS counts and they were excluded from the analysis. Lack of information on alcohol consumption and blood pressure led to the exclusion of another 12 individuals from the cross-sectional analysis and 88 individuals from the longitudinal analysis.

Cross-sectional analysis (HUNT3)

There were in general small differences between the headache groups and the headache-free individuals with regard to the number of PVS. Individuals suffering from migraine without aura were found to have fewer PVS than headache-free participants in BG (OR = 0.84, 95% CI = 0.76–0.94, p value = 0.003) and in BG and HWM together (OR = 0.97, 95% CI = 0.95–1.00, p value = 0.046). There were no associations between PVS and any of the other headache categories in HUNT3 (Table 2). No correlation between the frequency of headache attacks and the total number of PVS was found (OR = 1.00, 95% CI = 0.96–1.05, p value = 0.846, ordinal regression).

Table 2.

Comparison of different headache groups and headache-free individuals in HUNT3 (cross-sectional analysis) with regard to number of PVS in various parts of the brain.

Headache type	No. of individuals	Mean of PVS	OR (95% CI)	p value
Basal ganglia
Headache-free	543	10.13	1.00 (Reference)
Headache	304	10.24	1.01 (0.97–1.04)	0.721
Migraine	88	9.59	0.95 (0.88–1.01)	0.108
Migraine with aura	45	10.69	1.03 (0.95–1.11)	0.525
Migraine without aura	43	8.44	0.84 (0.76–0.94)	0.003
TTH	93	10.25	1.00 (0.95–1.06)	0.877
Unclassified headache	123	10.69	1.04 (0.99–1.09)	0.125
Hemispheric white matter
Headache-free	543	16.33	1.00 (Reference)
Headache	304	15.43	1.00 (0.99–1.01)	0.885
Migraine	88	13.39	0.99 (0.97–1.01)	0.224
Migraine with aura	45	14.62	0.99 (0.97–1.02)	0.557
Migraine without aura	43	12.09	0.98 (0.96–1.01)	0.168
TTH	93	17.11	1.00 (0.99–1.02)	0.602
Unclassified headache	123	15.63	1.01 (0.99–1.03)	0.319
Basal ganglia and hemispheric white matter merged
Headache-free	543	26.46	1.00 (Reference)
Headache	304	25.67	1.00 (0.99–1.01)	0.973
Migraine	88	22.98	0.99 (0.97–1.00)	0.130
Migraine with aura	45	25.31	1.00 (0.97–1.02)	0.707
Migraine without aura	43	20.53	0.97 (0.95–1.00)	0.046
TTH	93	27.35	1.01 (0.99–1.02)	0.352
Unclassified headache	123	26.32	1.00 (0.99–1.01)	0.904

PVS: perivascular spaces; OR: odds ratio; CI: confidence interval; TTH: tension-type headache; HUNT: Nord-Trøndelag Health Survey.

Values in bold are significant results (p<0.05).

Longitudinal analysis (HUNT2–3)

No significant associations between long-term headache sufferers and number of PVS were found (Table 3). As in the cross-sectional analysis the HUNT2-3 longitudinal analysis showed small differences in number of PVS between the headache groups and headache-free individuals.

Discussion

In contrast to the previous studies (1 –4), the present study did not demonstrate more PVS among those with headache (especially migraine) than among headache-free individuals. To the contrary, there were slightly fewer PVS among those with migraine without aura. However, we consider this finding quite likely to be spurious because fewer PVS among those with migraine without aura in HUNT3 (cross-sectional analysis) was the only significant difference in the migraine groups and the OR difference was modest. We found no association between long-term sufferers (migraine in both HUNT2 and 3) and PVS, which would have been likely if migraine and PVS were causally related. Furthermore considering that both migraine with and without aura are thought to be closely related disorders with similar pathophysiology (27), at least with regard to the headache phase, one would not expect PVS numbers to differ between the two groups. We consider this study to be an exploratory study, and have therefore not corrected the significance level for multiple comparisons. If we had used the Bonferroni-correction the results would have failed to reach statistical significance. Two of the former studies found a positive association between migraine and PVS (1,2), the third one between headache and PVS (3) and the fourth between both migraine and headache and PVS (4). However, there are notable differences between our study and the four previous ones (see Table 4). The present study was based on a more representative sample and a better methodology because it was population based, allowed adjustment for several covariates with multivariate statistical analyses, all participants were scanned with the same protocol and, in contrast to the studies of Rollins et al. (3) and Machado et al. (1), image analysis was performed blinded to clinical status. The previous studies were clinic-based studies, based on available clinical MRI images, and no adjustments for covariates were made. Clinic-based studies are known to introduce the risk of selection bias (28). To select controls randomly from the same population as the cases is considered to be the best method (28). Furthermore, the improvement in MRI technology in the years between our study (2007–2009) and three of the former studies (1993 (3), 1999 (2), 2001 (1)), and higher magnetic field strength in our study (see Table 4; 0.5 T was used in the study of Rollins et al. and some scans of Schick et al.) provides significantly improved image quality and more certain identification of PVS in the present study. In this respect it is of interest that Rollins et al. (3) found visible PVS in only 37 participants (3%) whereas it was found in all individuals in the present study.

Table 3.

Comparison of long-term headache sufferers and long-term headache-free (longitudinal analysis) with regard to number of PVS in various parts of the brain.

HUNT2	HUNT3	No. of individuals	Mean of PVS	OR (95% CI)	p value
Basal ganglia
Headache-free	Headache-free	306	10.11	1.00 (Reference)
Headache	Headache	208	10.17	1.00 (0.95–1.05)	0.920
Migraine	Migraine	60	10.07	0.98 (0.90–1.06)	0.537
Non-migrainous headache	Non-migrainous headache	91	10.33	1.02 (0.96–1.08)	0.564
Hemispheric white matter
Headache-free	Headache-free	306	17.01	1.00 (Reference)
Headache	Headache	208	15.06	1.00 (0.98–1.01)	0.523
Migraine	Migraine	60	14.98	0.99 (0.97–1.02)	0.572
Non-migrainous headache	Non-migrainous headache	91	16.29	1.00 (0.98–1.02)	0.973
Basal ganglia and hemispheric white matter merged
Headache-free	Headache-free	306	27.13	1.00 (Reference)
Headache	Headache	208	25.23	1.00 (0.98–1.01)	0.581
Migraine	Migraine	60	25.05	0.99 (0.97–1.01)	0.500
Non-migrainous headache	Non-migrainous headache	91	26.62	1.00 (0.98–1.02)	0.859

PVS: perivascular spaces; OR: odds ratio; CI: confidence interval; TTH: tension-type headache; HUNT: Nord-Trøndelag Health Survey.

Table 4.

Comparison of different studies on PVS and headache.

Author (year)	Study design	No. of participants (age range)	MRI strength	Definition of dilated PVS	Statistics	Findings
Husøy et al. (2015) (present study)	Population based, cross sectional and longitudinal	997 (50–66)	1.5 T	Visible on MRI	Multivariate logistic regression	No clear association between headache and PVS, and if any a negative correlation for migraine without aura (cross-sectional analysis). PVS present among all participants.
Biedroń et al. (2014) (4)	Clinic based	1348	1.5 T	–	–	More PVS among paediatric patients suffering from headache than paediatric patients without headache. Nine of 53 patients with PVS had migraine and six of 53 patients with PVS had other types of headache.
Machado et al. (2001) (1)	Clinic based case control	140 (13–54)	1.5 T	>5 mm in diameter	–	More PVS among migraineurs than headache-free individuals. Found PVS in 40% of migraineurs and in 7% of individuals in the control group.
Schick et al. (1999) (2)	Clinic based case control	118 (13–55)	0.5, 1.0 and 1.5 T	Visible on MRI	Chi-square test	PVS were found in 61% of children with migraine, in 22% of children with TTH, in 27% of headache-free peers and in 13% of adult migraineurs.
Rollins et al. (1993) (3)	Clinic based	1250 (–)	0.5 T	Visible on MRI	Fischer’s exact test	More PVS among paediatric patients suffering from headache than paediatric patients without headache. 12 of 37 patients with PVS had headache.

Differences in the definition of a PVS may also have contributed to the discrepant results between the studies. Rollins et al. (3) and Schick et al. (2) counted all visible PVS, similar to the present study, while Machado et al. (1) counted only PVS > 5 mm. It is likely that this latter approach results in fewer PVS, but the impact of this on the association to headache should be insignificant. Biedroń et al. (4) did not specify how they defined dilated PVS. As to the regions where the PVS were counted, the present study and Biedroń et al. (4) included data from the BG and HWM regions whereas the three other studies investigated only the HWM. We did not count PVS in the brain stem because the MR images had too low resolution in the mesencephalon area.

Looking at the sample sizes in the previous studies, Rollins et al. (3) and Biedroń et al. (4) had larger samples than the present study, while those of Schick et al. (2) and Machado et al. (1) had far smaller. Except for Machado et al., with 70 migraine patients, our study was far larger than the others with regard to participants in the different headache categories. This should give more statistical power in our study, consequently questioning the positive relationships reported in the other studies.

If there is an association between migraine and PVS, either positive or negative, one might speculate about a possible link to the pathophysiology. Increased numbers of PVS have previously been found in multiple sclerosis (29,30), an inflammatory disease. Considering that neurogenic inflammation in the vessels of the meninges is a likely pathogenic mechanism of migraine pain (5), one might expect to find a higher number of PVS in those with migraine. Frequent episodes of inflammation around the vessels conceivably should result in a higher rate of lymphatic drainage, which again ought to result in dilation and increased number of visible PVS, which was found in the previous studies but not in the present. An important difference between our study and the other studies was the higher age (Table 4). Schick et al. (2) speculated that a possible disturbance in the serotonin metabolism during brain development might be causally related to both PVS and migraine. Conceivably only derangement of serotonin metabolism related to migraine in childhood has the potential to influence brain development. This explanation of course assumes that almost all the individuals with migraine in the present study had adult-onset migraine, which is unlikely.

The present study has several strengths. It is the only population-based imaging study relating PVS to headache. We have categorised the sufferers into different headache groups, and longitudinal data on headache status were available allowing selection of a group with long-lasting headache problems. The participants were randomly drawn among a large population to keep selection bias at a minimum. A comparison of the non-invited, the non-participants and the participants revealed that they were not widely different from the general population, with the possible exception of cardiovascular risk factors (19). Some limitations need to be considered. Estimating the headache status of individuals with a questionnaire is inferior to a clinical interview. However, the headache criteria were validated (21,22), showing acceptable accuracy. The migraine diagnoses was highly specific, but had lower sensitivity. This relationship was opposite for the non-migrainous headache diagnoses, probably incorrectly rendering some true migraineurs as having a non-migrainous headache. The effect of potential diagnostic misclassification is to make any differences less than they are in reality. Another limitation of this study is the time interval between when participants answered the headache questionnaire (1995–1997 in HUNT2 and 2006–2008 in HUNT3) and when they were scanned (2007–2009). Furthermore, it can be difficult to define PVS when examining the MR images. For instance, in the areas near the cortex, PVS may be mistaken for blood vessels or vice versa. However, one would expect this misclassification to be non-differential, i.e. not affecting the relationship between PVS number and headache status.

In conclusion, this population-based study with a blinded design of individuals aged 50–66 years showed no clear association between number of dilated PVS and headache. This is in contrast to the findings of four previous studies in younger populations, indicating a positive correlation between headaches, in particular migraine, and PVS. We believe our results are more reliable due to the design of this study, being population based, blinded and with many participants in different headache categories, and because of development of better MRI technology in the past few years, enabling more certain identification of PVS.

Public health relevance

Previous clinic-based, case-control magnetic resonance imaging (MRI) studies have demonstrated more perivascular spaces (PVS) among headache sufferers than headache-free individuals.

In this population-based sample of 1006 individuals between the ages of 50 and 66 years investigated with a 1.5 T MRI scanner, participants with or without headache were compared with regard to visible PVS in the regions of basal ganglia and hemispheric white matter.

No clear association between headache and PVS was found in this study.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest

None declared.

Acknowledgements

The Nord-Trøndelag Health Study (the HUNT study) is a collaboration between the HUNT Research Centre (Faculty of Medicine, Norwegian University of Science and Technology NTNU), Nord-Trøndelag County Council, Central Norway Health Authority, and the Norwegian Institute of Public Health.

References

Machado

Jr Matos

Goyanna

, et al. Dilatation of Virchow-Robin spaces in patients with migraine [article in Portuguese]. Arq Neuropsiquiatr 2001; 59: 206–209.

Schick

Gahleitner

Wober-Bingol

, et al. Virchow-Robin spaces in childhood migraine. Neuroradiology 1999; 41: 283–287.

Rollins

Deline

Morriss

. Prevalence and clinical significance of dilated Virchow-Robin spaces in childhood. Radiology 1993; 189: 53–57.

Biedroń

Steczkowska

Kubik

, et al. Dilatation of Virchow-Robin spaces in children hospitalized at pediatric neurology department. Neurol Neurochir Pol 2014; 48: 39–44.

Olesen

Goadsby

Ramadan

, et al. The headaches, 3rd edn. Philadelphia, PA, USA: Lippincott Williams & Wilkins, 2006.

Schley

Carare-Nnadi

Please

, et al. Mechanisms to explain the reverse perivascular transport of solutes out of the brain. J Theor Biol 2006; 238: 962–974.

Virchow

. Ueber die Erweiterung kleiner Gefaesse. Archiv Pathol Anat Physiol Klin Med 1851; 3: 427–462.

Robin

. Recherches sur quelques particularités de la structure des capillaires de l’encephale. J Physiol Homme Anim 1859; 2: 537–548.

Hutchings

Weller

. Anatomical relationships of the pia mater to cerebral blood vessels in man. J Neurosurg 1986; 65: 316–325.

10.

Zhang

Inman

Weller

. Interrelationships of the pia mater and the perivascular (Virchow-Robin) spaces in the human cerebrum. J Anat 1990; 170: 111–123.

11.

Pollock

Hutchings

Weller

, et al. Perivascular spaces in the basal ganglia of the human brain: Their relationship to lacunes. J Anat 1997; 191(Pt 3): 337–346.

12.

Kwee

. Virchow-Robin spaces at MR imaging. Radiographics 2007; 27: 1071–1086.

13.

Heier

Bauer

Schwartz

, et al. Large Virchow-Robin spaces: MR-clinical correlation. AJNR Am J Neuroradiol 1989; 10: 929–936.

14.

Potter

Doubal

Jackson

, et al. Enlarged perivascular spaces and cerebral small vessel disease. Int J Stroke 2015; 10: 376–381.

15.

Gutierrez

Rundek

Ekind

, et al. Perivascular spaces are associated with atherosclerosis: An insight from the northern Manhattan study. AJNR Am J Neuroradiol 2013; 34: 1711–1716.

16.

Zhu

Dufouil

Mazoyer

, et al. Frequency and location of dilated Virchow-Robin spaces in elderly people: A population-based 3D MR imaging study. AJNR Am J Neuroradiol 2011; 32: 709–713.

17.

Papayannis

Saidon

Rugilo

, et al. Expanding Virchow Robin spaces in the midbrain causing hydrocephalus. AJNR Am J Neuroradiol 2003; 24: 1399–1403.

18.

Poirier

Barbizet

Gaston

, et al. Thalamic dementia. Expansive lacunae of the thalamo-paramedian mesencephalic area. Hydrocephalus caused by stenosis of the aqueduct of Sylvius [article in French]. Rev Neurol (Paris) 1983; 139: 349–358.

19.

Honningsvåg

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: 23–23.

20.

Hagen

Zwart

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.

21.

Hagen

Zwart

Aamodt

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

22.

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.

23.

Linde

Stovner

Zwart

, 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.

24.

Aamodt

Stovner

Hagen

, et al. Headache prevalence related to smoking and alcohol use. The Head-HUNT Study. Eur J Neurol 2006; 13: 1233–1238.

25.

Liu

Zhang

, et al. Morphological changes of cerebral vessels and expression patterns of MMP-2 and MMP-9 on cerebrovascular wall of alcoholic rats. Int J Clin Exp Pathol 2014; 7: 1880–1888.

26.

Hagen

Stovner

Vatten

, et al. Blood pressure and risk of headache: A prospective study of 22 685 adults in Norway. J Neurol Neurosurg Psychiatry 2002; 72: 463–466.

27.

Purdy

. Migraine with and without aura share the same pathogenic mechanisms. Neurol Sci 2008; 29(Suppl 1): S44–S46.

28.

Nishioka

Gyorkos

Joseph

, et al. Selection of subjects for hospital-based epidemiologic studies based on outward manifestations of disease. Clin Invest Med 2001; 24: 299–303.

29.

Conforti

Cirillo

Saturnino

, et al. Dilated Virchow-Robin spaces and multiple sclerosis: 3 T magnetic resonance study. Radiol Med 2014; 119: 408–414.

30.

Etemadifar

Hekmatnia

Tayari

, et al. Features of Virchow-Robin spaces in newly diagnosed multiple sclerosis patients. Eur J Radiol 2011; 80: e104–e108.