Urinary albumin excretion as a marker of endothelial dysfunction in migraine sufferers: the HUNT study, Norway.

OBJECTIVE: To investigate urine albumin leakage as a marker of endothelial dysfunction in migraine patients.
DESIGN: A population-based health study. PARTICIPANTS: 303 patients with migraine, 1009 patients with non-migraine headache and 5287 headache-free controls. OUTCOMES: The association between urine albumin- to-creatine ratio (ACR) and headache status was investigated in the Nord-Trøndelag Health Study (HUNT-2). Patients were selected in two strata, based on either (1) self-reported hypertension/diabetes (morbid sample) or (2) a random sample. Analyses were performed using analysis of covariance.
RESULTS: There was no association between headache status and ACR in the study population (p=0.23, mean ACR for migraine 1.66, 95% CI 1.31 to 2.01, for non-migraine headache 1.90, 95% CI 1.71 to 2.09 and for no headache 1.73, 95% CI 1.64 to 1.81) after relevant adjustments. Similarly, no association between headache status and ACR was seen when the analysis was stratified for morbid and random samples, or for migraine with and without aura.
CONCLUSIONS: We found no evidence of increased urine albumin leakage in migraine sufferers when compared with headache-free controls. This could indicate that systemic endothelial dysfunction is not a prominent feature of migraine.

We aimed to investigate urine albumin-to-creatine ratio (ACR) as a marker of endothelial dysfunction in patients with migraine, non-migraine headache and headache-free controls.

No associations were found between headache status and ACR, implying the absence of large differences in urine albumin leakage between migraine patients or other headache sufferers and headache-free controls.

This may indicate that systemic endothelial dysfunction is not a prominent feature of migraine.

The study is the first to assess albumin leakage as a marker of endothelial dysfunction in migraine patients.

The study includes a relatively large sample size, objective measurements of albumin leakage and the use of validated headache diagnoses.

The sample size might be too low to detect ACR differences among migraine subtypes. In particular, there was a limited number of patients having migraine with aura in this study.

Introduction

Migraine is a common disabling headache disorder affecting 6–7% of men and 18% of women.1 It is rated as one of the top 10 most disabling diseases2 3 and has been estimated to be the most costly neurological disorder to society.4 Migraine manifests in headache attacks lasting 4–72 h characterised by throbbing, pulsating and unilateral headache, often accompanied by nausea, photophobia and phonophobia.5 In up to a third of the patients6 these attacks may be accompanied by additional neurological aura symptoms separating the disease into migraine with aura (MA) and migraine without aura (MO).5

Migraine, in particular MA, has been associated with an increased risk of cardiovascular disease (CVD) including ischaemic stroke, ischaemic lesions of the brain, myocardial infarction, angina and cardiovascular death.7–9 While the mechanisms underlying this link remain unknown, multiple explanations have been suggested. Shared environmental or genetic risk factors could be present, changes in vascular function might contribute to migraine pathophysiology or the opposite; migraine pathophysiology could induce changes in vascular function, or mutual mechanistic pathways may exist.10

The vascular endothelium plays an important role in setting vascular tone, regulating vascular permeability, maintaining thrombotic balance and regulating fibrinolytic and inflammatory pathways. A dysfunctional vascular endothelium is often observed in CVD.11 Interestingly, abnormal control of systemic vascular tone,12 alterations in systemic arterial structure and function13 and impaired ability to repair systemic endothelial injury14 have also been reported in migraine patients. The increase of systemic vasoactive mediators during migraine attacks, including the vasoconstrictive factor ET-1 and the pro-inflammatory factor C reactive protein, suggests an involvement of the endothelium in migraine headache.15 16 Endothelial dysfunction could also be implicated in the generation of migraine aura as ET-1 has been shown to produce cortical spreading depression, the presumed substrate of migraine aura, in rats.17 Furthermore, migraine patients with aura have an increased risk of thrombotic events. It has been reported raised plasma levels of vWf in migraine patients,18 a factor that promotes clotting and formation when the endothelium is damaged, further supporting the presence of endothelial dysfunction in migraine. This could explain the link between migraine and CVD and might elucidate pathophysiological mechanisms of migraine.

The presence of subclinical increases in urinary albumin excretion (UAE) has been associated with impaired endothelium-dependent vasodilation19 and is believed to reflect endothelial dysfunction.20 21 No previous studies have, however, investigated albumin excretion in migraine patients. The aim of this study was therefore to compare urine albumin-to-creatine ratio (ACR), as a measure of UAE, in patients with migraine, non-migraine headache and headache-free controls.

Methods

Participants and headache diagnoses

Participants were selected from the second Nord-Trøndelag Health Study (HUNT 2), a general health survey conducted in 1995–1997. All inhabitants aged ≥20 years residing in the county at the time of screening were invited (n=92 703) and 65 258 patients (70.4%) participated.22 The population in Nord-Trøndelag County is ethnically homogenous with less than 3% non-Caucasians. All participants were mailed a questionnaire and attended a clinical examination (for details see ref. 22). The questionnaire included questions on previous or current CVD, hypertension, diabetes mellitus and smoking habits as well as 13 headache questions. The clinical examination included measurements of height, weight and blood pressure. When attending the health examination, participants received a second questionnaire to complete and return by mail. The 13 headache questions of the questionnaire were designed mainly to determine whether the individual had headache, the frequency of headache, and to diagnose migraine according to a modified version of the ICHD migraine criteria.23 24 Participants who answered ‘yes’ to the question ‘Have you suffered from headache during the last 12 months?’ were classified as headache sufferers. Those who answered ‘no’ comprised the headache-free control group. Based on the subsequent 12 headache questions, headache sufferers were classified as having migraine if they fulfilled the following three criteria: (1) headache attacks lasting 4–72 h (less than 4 h was accepted for those who reported often visual disturbances before headache); (2) headache with at least one of the following characteristics: pulsating quality, unilateral location, aggravation by physical activity; (3) during headache, at least one of the following: nausea, photophobia or phonophobia. Those within this group who reported often having visual disturbance prior to headache were classified as having MA. Headache sufferers who did not fulfil the criteria for migraine and did not have self-reported migraine were classified as having non-migraine headache. Headache diagnoses were mutually exclusive. The headache diagnoses were validated by clinical interviews by physicians with long clinical experience in headache disorders. The positive and negative predictive values of the questionnaire-based diagnoses were, respectively, 68% and 76% for non-migraine headache, 87% and 75% for migraine, and 100% and 91% for MA. A total of 9598 participants were included in a further study in which they donated three urine samples for the determination of ACR, originally designed for investigating UAE in hypertension and diabetes. Participants were included based on (1) the presence of self-reported diabetes mellitus and/or medically treated hypertension (morbid sample) and (2) a 5% randomly selected sample of the total population (random sample). Participants who contributed with information regarding both headache status and ACR were eligible for the present study. Participants with overt proteinuria, ACR ≥25 mg/mmol in men and ≥35 mg/mmol in women, one participant with migraine (with aura), 16 participants with non-migraine headache and 141 participants without headache, were excluded. In total, 303 participants with migraine, 1009 participants with non-migraine headache and 5287 headache-free participants were included in the study, in which the morbid sample constituted 3688 medically treated hypertensive participants, 647 diabetic participants and 413 participants with both medically treated hypertension and diabetes mellitus.

Urine sampling

Participants received a unit with three plastic receptacles for three first morning urine samples, three transport tubes and one envelope for returning the samples by mail to the laboratory. Of 11 661 packs handed out, 9598 (82.3%) participants mailed three samples back to the laboratory.25 Those who failed to return three samples were excluded. A written instruction describing how to collect urine, information about the screening and a questionnaire concerning urinary tract infection in the previous week, persistent haematuria in the last year and menstruation at the time of collection was included. While the most consistent method for determining UAE is 24 h urine sampling, it has been shown that measuring ACR in one or more morning spot urine samples provides good specificity and sensitivity.26

Laboratory procedures

Fresh, non-frozen urine samples were analysed at the Central Laboratory at Levanger Hospital on a Hitachi 91 Autoanalyzer (Hitachi, Mito, Japan). Urine albumin and creatine levels were determined using an immunoturbimetric method (antihuman serum albumin; Dako Norway, Oslo) and Jaffé method, respectively. The ACR, measured in mg/mmol, was used as an expression of UAE.

Consent

Participation in HUNT2 was voluntary and each participant signed a written consent. All surveys and analyses were approved by the Norwegian Data Inspectorate and by the Regional Committee for Medical and Health Research Ethics.

Statistical analysis

For all analyses we used mean ACR from three delivered urine samples, measured in mg/mmol. Comparisons of mean ACR between cases with migraine, non-migraine headache and headache-free controls in the morbid, random and combined samples were performed by analysis of covariance. The confounding effect of potential confounders, that is, age, sex, body mass index (BMI), smoking, self-reported diabetes and self-reported antihypertensive medication, in the combined sample was calculated by the formula (mean ACRcrude−mean ACRadj)/mean ACRcrude, where mean ACRcrude was the mean ACR in the crude model and mean ACRadj the mean ACR in the model adjusted for the relevant variable. Variables with a confounding effect >5% in one or more headache group were included in the final model. A p value less than 0.05 was used to indicate statistical significance.

Results

Baseline characteristics for the random and morbid samples are shown in table 1. The morbid sample was characterised by a higher age and BMI and a lower proportion of smokers as compared with the random sample. For both samples a higher proportion of women had headache than men, with the highest percentage of women found among the migraine participants. The mean age was lower in the headache groups than in controls.

Table 1

Clinical profile of the study participants

	No headache	Non-migraine headache	Migraine
Morbid sample (n=4748)
Women (%)	51.7	60.2	69.8
Mean age (SD)	67.0 (11.2)	58.7 (12.9)	51.5 (12.5)
Antihypertensive use (%)	86.9	84.4	83.0
Diabetes (%)	22.5	22.5	20.9
Daily smokers (%)	19.3	24.6	23.1
Mean body mass index (SD)	28.6 (4.5)	29.0 (4.9)	28.7 (4.8)
Random sample (n=1851)
Women (%)	46.8	63.2	75.0
Mean age (SD)	53.9 (16.5)	43.6 (13.7)	39.6 (11.1)
Antihypertensive use (%)	13.1	7.2	7.0
Diabetes (%)	4.1	1.2	0.0
Daily smokers (%)	27.3	33.3	34.8
Mean body mass index (SD)	26.5 (4.0)	26.1 (4.0)	25.8 (3.8)

Crude analyses indicated an association between ACR and headache status in the morbid sample (p=0.013, mean ACR for migraine 1.22, for non-migraine headache 1.95 and for no headache 2.03) and the random sample (p=0.011, mean ACR for migraine 0.85, for non-migraine headache 0.96 and for no headache 1.22) (table 2). However, when adjusting for age and sex this effect disappeared (model I). No significant associations were observed between headache status and ACR in the morbid sample (p=0.11, mean ACR for migraine 1.87, for non-migraine headache 2.27 and for no headache 1.96) or the random sample (p=0.55, mean ACR for migraine 1.00, for non-migraine headache 1.07 and for no headache 1.16) when adjusted for age, sex, self-reported diabetes and self-reported use of antihypertensive medication (model II).

Table 2

Albumin-to-creatine ratio (ACR) by headache status and selection criterium

	Crude		Model I‡		Model II§
	Mean ACR (95% CI) N	p Value	Mean ACR (95% CI) N	p Value	Mean ACR (95% CI) N	p Value
Morbid sample		0.013*		0.13*		0.11*
No headache	2.03 (1.92 to 2.14) 3998		1.96 (1.85 to 2.07) 3998		1.96 (1.85 to 2.06) 3978
Non-migraine headache	1.95 (1.68 to 2.23) 591	0.61†	2.26 (1.97 to 2.54) 591	0.057†	2.27 (1.99 to 2.55) 585	0.044†
Migraine	1.22 (0.68 to 1.75) 159	0.003†	1.85 (1.31 to 2.40) 159	0.70†	1.87 (1.33 to 2.41) 158	0.75†
Random sample		0.011*		0.60*		0.55*
No headache	1.22 (1.11 to 1.32) 1289		1.16 (1.06 to 1.27) 1289		1.16 (1.05 to 1.26) 1285
Non-migraine headache	0.96 (0.77 to 1.14) 418	0.017†	1.08 (0.89 to 1.26) 418	0.44†	1.07 (0.89 to 1.26) 417	0.44†
Migraine	0.85 (0.53 to 1.16) 144	0.028†	1.02 (0.70 to 1.34) 144	0.41†	1.00 (0.68 to 1.31) 143	0.35†

*p Value for overall effect of headache groups.

†p Values are for comparison against no headache group within the same selection criterium.

‡Model I: adjusted for age and sex.

§Model II: adjusted for age, sex, self-reported diabetes and self-reported use of antihypertensive medication.

A similar pattern was observed when morbid and random samples were combined; crude data indicated an association between ACR and headache status, but this effect was explained by differences in age and sex, table 3. No significant associations were observed when adjusted for age, sex, self-reported diabetes and self-reported use of antihypertensive medication (p=0.23, mean ACR for migraine 1.66, for non-migraine headache 1.90 and for no headache 1.73). Mean differences for ACR in the combined sample (model II) were −0.07 (95% CI −0.43 to 0.30) between migraine participants and headache-free controls and 0.18 (95% CI −0.04 to 0.39) between non-migraine headache and headache-free controls. We also examined the association between headache status and ACR after stratification of migraine into its two subtypes MA and MO. In the adjusted model, no differences in ACR scores were found between migraine subtype and controls (p=0.68, mean ACR for MA 1.36, for MO 1.73 and for no headache 1.73; table 4). There was no interaction between headache status and self-reported diabetes, or between headache status and self-reported use of antihypertensive medication on ACR. Since the measurement ACR was positively skewed, we repeated the analyses after square root transformation, which provided similar results. To ease interpretation only non-transformed results are presented.

Table 3

Albumin-to-creatine ratio (ACR) by headache status in combined sample

	Crude		Model I‡		Model II§
	Mean ACR (95% CI) N	p Value	Mean ACR (95% CI) N	p Value	Mean ACR (95% CI) N	p Value
		<0.001*		0.29*		0.23*
No headache	1.83 (1.75 to 1.92) 5287		1.73 (1.65 to 1.81) 5287		1.73 (1.64 to 1.81) 5263
Non-migraine headache	1.54 (1.35 to 1.73) 1009	0.007†	1.89 (1.69 to 2.09) 1009	0.15	1.90 (1.71 to 2.09) 1002	0.11†
Migraine	1.04 (0.69 to 1.39) 303	<0.001†	1.67 (1.31 to 2.03) 303	0.74	1.66 (1.31 to 2.01) 301	0.72†

*p Value for overall effect of headache groups.

†p Values are for comparison against no headache group.

‡Model I: adjusted for age and sex.

§Model II: adjusted for age, sex, self-reported diabetes and self-reported use of antihypertensive medication.

Table 4

Albumin-to-creatine ratio (ACR) by migraine subtypes

	Crude		Model I‡		Model II§
	Mean ACR (95% CI) N	p Value	Mean ACR (95% CI) N	p Value	Mean ACR (95% CI) N	p Value
		<0.001*		0.78*		0.68*
No headache	1.83 (1.75 to 1.92) 5287		1.73 (1.65 to 1.81) 5287		1.73 (1.64 to 1.81) 5263
Migraine without aura	1.05 (0.66 to 1.44) 244	<0.001†	1.73 (1.33 to 2.12) 244	0.88†	1.73 (1.34 to 2.13) 242	0.85†
Migraine with aura	1.00 (0.21 to 1.80) 59	0.042†	1.43 (0.64 to 2.22) 59	0.49†	1.36 (0.59 to 2.13) 59	0.39†

*p Value for overall effect of headache groups.

†p Values are for comparison against controls.

‡Model I: adjusted for age and sex.

§Model II: adjusted for age, sex, self-reported diabetes and self-reported use of antihypertensive medication.

Discussion

The present study did not find an association between headache status and ACR, indicating the absence of large differences in urine albumin leakage between migraine participants or other headache sufferers and headache-free controls. The mechanisms linking migraine and CVD are likely complex, but dysregulation of vascular endothelium as expressed by albumin leakage is not supported by the present data to be a major factor.

The study includes a relatively large sample size, objective measurements of UAE, analyses based on fresh urine samples and the use of validated headache diagnoses. Furthermore, the design of the study allowed us to evaluate the association between headache status and UAE in a random sample as well as a morbid sample, which included a high proportion of participants with hypertension and diabetes types I and II, well-known risk factors for increased UAE.27–29 A limitation of the study is, however, that hypertension and diabetes information was based on self-report. Data on participants’ migraine status (ictal or interictal phase) during urine sampling was not available and it cannot be excluded that there may be a difference in urine albumin leakage during and outside a migraine attack. In addition, the sample size might be too low to detect ACR differences among migraine subtypes. In particular, there was a limited number of participants having MA in this study.

A high level of urine ACR, most likely representing excessive glomerular capillary leakage of albumin,19 is a risk marker for cardiovascular morbidity and mortality30 and is believed to reflect endothelial dysfunction.19 21 Increased UAE has been associated with impaired endothelium-dependent vasodilatation,19 impaired peripheral vasoreactivity to endothelial agonists31 as well as with circulating biomarkers of endothelial dysfunction.32 Abnormal ACR levels in urine have also been reported in inflammatory, non-cardiovascular conditions such as rheumatoid arthritis, inflammatory bowel syndrome and surgery, which has been hypothesised to be a result of circulating inflammatory molecules damaging parenchymal functions of the kidneys.20 Interestingly, ACR levels even within the normal range have been associated with an increased rate of kidney disease, CVD and mortality.33–35 As UAE is a biological continuum, we used continuous ACR values in the present study rather than applying microalbuminuria commonly defined as ACR ≥2.5 mg/mmol. The mean ACR levels in the morbid group of the present study were twice as high as for the randomly chosen sample, which is explained by the presence of hypertension and diabetes. In the random sample, however, ACR was generally low for all headache and headache-free groups.

The unadjusted analysis indicated an association between ACR and headache status in the morbid, random and combined samples. However, when adjusting for age and sex the association disappeared, suggesting these variables to be the major contributing factors for differences in ACR in the present study. Given the fact that women have less muscular mass and lower creatine excretion than men, and that creatine excretion decreases with age,25 a confounding effect of age and sex on ACR was not unexpected.

Contrary to our hypothesis we did not find urine albumin leakage to be more prominent among migraine sufferers than among headache-free participants. If endothelial dysfunction with albumin leakage was a feature of migraine, we would expect elevated ACR levels among migraine participants, and highest among MA. Although we could not find an association between ACR and migraine status, there was a non-significant tendency in the opposite direction; participants with MA had the lowest ACR levels. While many previous studies favour the presence of endothelial dysfunction in migraine patients, some studies argue against it. In fact, systemic endothelium-dependent vasodilation, nitric oxide levels and stimulated t-PA levels have been found similarly expressed in migraine participants and healthy controls.36–38 Furthermore, it was recently suggested that extracranial dilatation may not be relevant for migraine pain during attacks.39

In summary, in this first study examining albumin leakage in migraine sufferers, we did not find urine albumin leakage to be more prominent among migraine participants than among headache-free participants, which could indicate that systemic endothelial dysfunction is not a prominent feature of migraine.