The relationship between migraine and lipid sub-fractions among individuals without cardiovascular disease: A cross-sectional evaluation in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil)

Abstract

Introduction

Recent studies have explored the relationship between dyslipidemia and migraine in a cardiovascular context. Thus, we aimed to evaluate the possible association between lipids, lipoprotein subfractions and migraine according to aura symptoms in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil).

Methods

1,560 women and 1,595 men, without CVD or lipid disorders requiring medication, underwent a baseline clinical assessment. Total-cholesterol and its sub-fractions (LDL, VLDL and HDL subclass cholesterol); triglycerides and triglyceride-rich lipoprotein cholesterol [TRL-C (VLDL₁₊₂-C VLDL₃-C + IDL-C)] were determined by vertical auto profile (VAP). We also calculated logarithmic LDL density ratio [LLDR = ln ((LDL₃-C + LDL₄-C)/(LDL₁-C + LDL₂-C))], T-Chol/HDL-C and triglycerides/HDL-C ratios. Odds ratios (OR) with 95% confidence intervals (CI) were obtained to evaluate the relationship between lipids tertiles and migraine for both sexes.

Results

Main findings revealed positive associations between migraine without aura (MO) and the highest tertiles of VLDL-C (OR, 1.61; 95%CI, 1.07–2.40) and TRL-C (OR, 1.55; 95% CI, 1.03–2.34) in women. In men, the highest tertile of VLDL₃-C (OR, 3.87; 95%CI, 1.23–12.19) was positively associated with MO, as well.

Conclusions

In middle-aged participants without CVD or lipid disorders requiring medication, the worst lipid profile was determined by the highest levels of TRL-C and their cholesterol-rich remnants in migraineurs without aura for both sexes.

Keywords

Migraine with aura lipids lipoproteins observational study

Introduction

Migraine has been associated with increased cardiovascular risk (1 –4). Middle-aged women, particularly those who have migraine with aura symptoms, seem to present the highest risk of developing cardiovascular disease (CVD) (5 –7). A recent prospective cohort with more than 100,000 middle-aged women suggested a more consistent link between migraine as a risk factor for CVD, and were associated to mortality (8).

Emerging evidence suggests that dyslipidemia is a potential mediator between migraine and cardiovascular outcomes (9). However, previous data, including population-based studies (1 –3,7), have reported conflicting findings (1 –4,10 –14). Traditional evaluation of the blood lipid profile might not detect a significant part of the population at high risk for coronary heart disease (CHD). In this context, innovative techniques to better characterize plasma lipoproteins (15,16), including non-HDL-C, triglyceride-rich lipoprotein cholesterol (TRL-C) or HDL-C subfractions, which have been implicated in atherogenesis (17), might add relevant information regarding the risk of CVD outcomes in migraineurs.

Previous cross-sectional analyses performed in ELSA-Brasil, which evaluated migraine headache and cardiovascular risk factors by sex, found a slightly increased odds ratio (OR) for the relationship between dyslipidemia at baseline (LDL-C ≥ 130 mg/dl and/or use of lipid-lowering drugs) and migraine in women. However, no analyses according to aura symptoms or other lipid fractions were performed (4).

Thus, we aimed to evaluate cross-sectionally the relationship between migraine (with-MA and without aura-MO) and an advanced lipid panel using the vertical auto profile (VAP) method (15) in 3,155 participants without CVD or lipids disorders requiring medication in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil), a large cohort with a relatively high frequency of migraine and cardiovascular risk factors (4).

Methods

Study design and population

ELSA-Brasil is an ongoing prospective cohort of 15,105 civil servants aged 35 to 74 years in six state capital cities in Brazil. Baseline assessments occurred between 2008 and 2010. Further details of this cohort are described elsewhere (18 –20). In brief, all active or retired civil servants of six academic institutions were eligible for the study. Exclusion criteria were current or recent pregnancy, intention to quit working at the institution, severe communication impairment, and residence outside of a study center’s metropolitan area. Participants were selected according to occupation, classified as unskilled, technical/clerical and faculty and professional staff, permitting a gradient of socioeconomic position across the sample (20).

This is a cross-sectional analysis using 5,061 participants from the ELSA-Brasil investigation center in São Paulo who had a lipid profile measured by VAP method. From 5,061 participants in Sao Paulo, 4,935 had complete information about migraine headaches and aura symptoms. As our focus was to evaluate the association between lipid profile and migraine headache, we excluded 315 participants who reported a previous history of CVD (CHD, myocardial revascularization, percutaneous coronary intervention, heart failure and stroke).

Since the information about all classes of lipid-lowering drugs was available for ELSA-Brasil participants at baseline, we were able to exclude those participants who were taking lipid-lowering drugs (564 individuals), leaving 4,056 participants. In addition, we excluded 901 participants with probable migraine, which left us with 3,155 participants (2,780 individuals without migraine, 164 who had MA and 211 who had MO).

Lipid profile

Lipids were extracted from venous blood samples, which were obtained after a 12-hour overnight fast. Here, we evaluated an extended lipid panel using the VAP test (Atherotech®), an inverted rate zonal, single vertical spin, density gradient ultracentrifugation (UC) technique that simultaneously measures cholesterol concentrations after fraction separation, including T-Chol and its sub-fractions: Real LDL-C [(LDLr-C), which is the biochemically defined LDL-C fraction from the UC separation of the lipids by VAP; dense LDL-C, (LDL₃-C + LDL₄-C); low density (buoyant) LDL-C, (LDL₁-C + LDL₂-C); total VLDL (very low density lipoprotein cholesterol) and VLDL₃-C (small dense cholesterol-rich VLDL subfraction); non-HDL-C (non-high-density lipoprotein cholesterol) that corresponds to a sum of LDLr-C, VLDL-C, IDL-C and Lp(a)-C; HDL-C and its sub-fractions HDL₂-C (larger, buoyant subclass) and HDL₃-C (smaller, denser subclass). Triglyceride-rich lipoprotein cholesterol (TRL-C) was defined as IDL-C + VLDL₃-C + VLDL₁₊₂-C subfractions (15). We calculated T-Chol/HDL-C and triglyceride/HDL-C ratios and also the logarithm of LDL-C density ratio [LLDR, ln ((LDL₃-C + LDL₄-C)/(LDL₁-C + LDL₂-C))], which is closely related to ultracentrifugation-derived LDL-C density phenotype (16). Total plasma triglycerides were measured by traditional enzymatic methods (Figure 1).

Figure 1.

Median (Interquartile range) of lipoproteins subfractions cholesterol, ELSA-Brasil. Source: Segrest JP et al., 1994; 45, 303–369 (39).

Migraine definition

All participants who answered “yes” to the question “In the last 12 months, did you have a headache?” at the ELSA-Brasil baseline evaluation were invited to answer a detailed headache questionnaire based on the International Headache Society criteria (IHS-2004) (21), which has been adapted, validated and previously used in Brazil (22). All migraineurs underwent a telephone interview about aura symptoms performed by a medical doctor with great expertise in headache classification based on IHS criteria (21,22). We classified individuals according to IHS (codes 1.1: MO or 1.2: MA). Participants with other headaches or without headache were considered as having no migraine.

Finally, we collected information about prophylactic medications for migraine based on the American Academy of Neurology guideline, as follows. Level A: Divalproex sodium, sodium valproate, topiramate, propranolol, metoprolol and timolol; level B: Amitriptyline, venlafaxine, atenolol and nadolol (23).

Other study variables

Each participant underwent an interview at their workplace and a visit to the Research Center for clinical exams according to standard protocols (18 –20). The interview and examination were performed by trained personnel, with strict quality control (20). Questionnaires addressed age (median, interquartile range), years of education (up to incomplete high school, complete high school or more than high school), self-reported race (White, Brown, Black, Asian and Indigenous), smoking status (never, former and current), physical activity at leisure time (low, mild or vigorous) using the International Physical Activity Questionnaire (IPAQ-Long Form). Alcohol intake was evaluated by a specific questionnaire that captures alcoholic type, frequency and pattern of alcohol intake in the last 12 months (24). We considered a high hazardous alcohol consumption per week as ≥ 140 grams for women and ≥ 210 grams for men.

Body weight was measured to the nearest 0.1 kg with the participant wearing light clothes using a digital scale (Toledo® Columbus, Ohio, US). Height was measured to the nearest 0.1 cm using a wall-mounted stadiometer that rises from floor height with a footplate for increased stability (Seca®, Hamburg, Germany). Anthropometry was performed in all six ELSA-Brasil centers using the same equipment during baseline assessment. The research time, which is composed by nurses, was trained in a centralized way to ensure uniformity across ELSA-Brasil centers under strict quality control.

Body mass index (BMI) was calculated as weight (in kilograms) divided by the square of height (in meters) and organized in strata, based on the World Health Organization criteria. For obesity, we considered BMI 30.0 kg/m² or above. Blood pressure (BP) was taken using a validated oscillometric device (Omron HEM 705CPINT). Three measurements were taken at one-minute intervals. The mean of the two latest BP measurements was considered as the BP values. Hypertension was defined as the use of medications to treat hypertension, a systolic blood pressure ≥ 140 mmHg or a diastolic blood pressure ≥ 90 mmHg. Diabetes was defined as a medical history of diabetes, the use of medications to treat diabetes, a fasting serum glucose ≥ 126 mg/dl, a 2 h oral glucose tolerance test ≥ 200 mg/dl or HbA1c levels ≥ 6.5% (25).

Statistics

Baseline characteristics were compared according to the migraine headache status (no migraine, MA and MO). Categorical variables were presented in proportions using a Chi-square test, and continuous variables were presented in median values (interquartile range) using the Kruskal-Wallis test. p-values for linear trends for each association with lipid subfraction according to migraine subgroups were obtained by the Jonckheere-Terpstra test. We also calculated OR (95%CI) for the association between each lipid component (in tertiles) as the dependent variables and migraine subgroups, using the “no migraine” subgroup as reference. We performed multinomial logistic regression analyses in crude, age-adjusted and multivariable models adjusted by confounders such as age, sex, educational level, smoking, hypertension, diabetes, and body mass index (BMI). We performed additional adjustments by high sensitive-C-Reactive Protein (hs-CRP) and migraine prophylaxis (23).

Previous analyses in ELSA-Brasil showed that migraine and cardiovascular risk factors are heterogeneous among women and men (4). Based on these findings, we proposed sex-stratified analysis in the present paper.

For all analyses, p-values < 0.05 were also considered significant. The statistical analyses were performed with the statistical software SPSS version 22.0.

Results

Table 1 shows baseline characteristics according to migraine status for all of the sample and by sex. Out of 3,155 participants, those who had MA were younger than MO participants. This difference in age according to migraine was also observed for both sexes. Overall, migraine was more often found in women compared to men. Aura symptoms were present in 40.5%, and most migraineurs with aura were women (85.8%). Hypertension and diabetes were more frequent among non-migraineurs than migraineurs. However, the frequency of obesity across migraine subgroups was not statistically different.

Table 1.

Baseline clinical characteristics of 3,155 participants from ELSA-Brasil according to migraine status.

Total sample (n = 3,155)	Non-migraine n = 2,780	Migraine without aura n = 211	Migraine with aura n = 164
Median age, years (IQR)	50 (45–57)	48 (44–52)*	46 (42–56)*
Female gender (%)	1243 (44.7)	181 (85.8)*	136 (82.9)*
White race (%)	1623 (59.2)	117 (56.3)	84 (51.2)
Years of education (%)
Up to incomplete high school	462 (16.6)	21 (10.0)*	15 (9.1)*
BMI, kg/m² (IQR)	27 (24–30)	26 (24–29)	30 (24–31)
hs-CRP, mg/L (IQR)	1.38 (0.68–3.22)	1.51 (0.82–3.67)	1.76 (0.81–4.43)*
Hypertension (%)^†	800 (28.8)	46 (21.8)**	31 (18.9)**
Diabetes (%)^‡	492 (17.7)	27 (12.8)**	16 (9.8)**
Obesity (%)	629 (22.6)	45 (21.3)	46 (28.0)
Current smoking (%)	451 (16.2)	40 (19.0)	29 (17.7)
High hazardous alcohol consumption per week (%)^¥	327 (11.8)	12 (5.7)*	3 (1.8)*
Moderate-intense physical activity (%)	614 (22.9)	30 (14.8)**	22 (16.8)**
Migraine prophylaxis medication (%)	190 (6.8)	18 (8.5)**	14 (8.5)**
Women (n = 1,560)	n = 1,243	n = 181	n = 136
Median age, years (IQR)	51 (45–58)	48 (43–52)*	46 (42–51)*
White race (%)	761 (61.8)	102 (56.7)**	74 (54.4)**
Years of education (%)
Up to incomplete high school	155 (12.5)	16 (8.8)*	155 (12.5)*
BMI, kg/m² (IQR)	26 (23–30)	26 (24–29)	27 (24–31)
hs-CRP, mg/L (IQR)	1.59 (0.75–3.84)	1.60 (0.81–3.80)	2.25 (0.88–5.16)
Hypertension (%)^†	293 (23.6)	37 (20.4)	26 (19.1)
Diabetes (%)^‡	176 (14.2)	25 (13.8)	13 (9.6)
Obesity (%)	315 (25.3)	39 (21.5)	42 (30.9)
Current smoking (%)	172 (13.8)	35 (19.3)	25 (18.4)
High hazardous alcohol consumption per week (%)^¥	60 (4.8)	8 (4.4)	2 (1.5)
Moderate-intense physical activity (%)	233 (19.6)	24 (13.7)	15 (11.3)
Migraine prophylaxis medication (%)	86 (6.9)	16 (8.8)	12 (8.8)
Men (n = 1,595)	n = 1,537	n = 30	n = 28
Median age, years (IQR)	50 (44–57)	49 (45–55)*	47 (41–51)*
White race (%)	862 (57.1)	15 (53.6)	10 (35.7)
Years of education (%)
Up to incomplete high school	307 (20.0)	5 (16.7)	3 (10.7)
BMI, kg/m² (IQR)	27 (24–29)	25 (23–29)	25 (23–29)
hs-CRP, mg/L (IQR)	1.25 (0.65–2.86)	1.24 (0.85–2.14)	1.38 (0.63–2.13)
Hypertension (%)^†	507 (33.0)	9 (30.0)	5 (17.9)
Diabetes (%)^‡	316 (20.6)	2 (6.7)	3 (10.7)
Obesity (%)	314 (20.4)	6 (20.0)	4 (14.3)
Current smoking (%)	279 (18.2)	5 (16.7)	4 (14.3)
High hazardous alcohol consumption per week (%)^¥	267 (17.4)	4 (13.3)	1 (3.6)
Moderate-intense physical activity (%)	381 (25.5)	6 (21.4)	7 (25.9)
Migraine prophylaxis medication (%)	104 (6.8)	2 (6.7)	2 (7.1)

Some proportions might not add up to 100% due to rounding or missing values (at most 1.8%).

Continuous variables are demonstrated in median with interquartile ranges (IQR).

All migraine includes migraine with and without aura, controls include non-migraine cases.

p-values were from Chi-Square test for categorical variables and from Kruskal-Wallis test for continuous variables with Bonferroni correction.

p-value < 0.001 compared to non-migraine subgroup.

p-value < 0.05 compared to non-migraine subgroup.

†

Hypertension was defined by systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg or current treatment.

‡

Diabetes was defined as previous medical history of diabetes, use of medication to treat diabetes, a fasting plasma glucose ≥126 mg/dl), a 2-hour plasma glucose ≥ 200 mg/dl), and/or HbA1C ≥ 6.5%.

High hazardous alcohol consumption per week: ≥ 140 grams for women and ≥ 210 grams for men.

IQR: Interquartile range; BMI: Body mass index; hs-CRP: High sensitive-C-Reactive Protein.

Table 2 shows the lipid profile according to migraine subtype stratified by sex. For women, those with MO had the lowest values of total-HDL-C [56 (47–66 mg/dl), p = 0.003], its subfractions HDL2-C [15 (12–21 mg/dl), p trend = 0.01] and HDL3-C [41 (30–46 mg/dl), p trend = 0.003]. The TG/HDL-ratio was progressively higher across migraine subgroups with MA presented with higher concentrations [1.76 (1.14–2.54), p trend = 0.03]. Among men, we found no significant differences for the relationship with lipid subfractions and migraine subtypes.

Table 2.

VAP Lipid profile in the ELSA-Brasil baseline assessment, according to migraine status.

	Non-migraine (n = 2,780)	Migraine without aura (n = 211)	Migraine with aura (n = 164)	p for
	Median (IQR)	Median (IQR)	Median (IQR)	trend
Women (n = 1,560)	n = 1,243	n = 181	n = 136
Total cholesterol, mg/dl	218 (194–247)	214 (187–248)	215 (195–237)	0.11
LDL-real cholesterol, mg/dl	110 (92–131)	113 (88–131)	108 (92–122)	0.33
VLDL-cholesterol, mg/dl	22 (17–28)	23 (18–31)	21 (18–30)	0.16
Total HDL-cholesterol, mg/dl	59 (50–70)**	56 (47–66)**	57 (47–67)**	0.003
Triglycerides, mg/dl	96 (73–130)	97 (73–143)	98 (74–138)	0.27
Non-HDL cholesterol, mg/dl	157 (133–185)	159 (132–187)	156 (135–179)	0.67
VLDL₃ –cholesterol, mg/dl	13 (11–16)	14 (11–17)	13 (11–17)	0.24
HDL₂-cholesterol, mg/dl	17 (13–21)**	15 (12–21)**	16 (12–20)**	0.01
HDL₃-cholesterol, mg/dl	42 (37–48)**	41 (30–46)**	41 (36–46)**	0.003
TRL-cholesterol, mg/dl	29 (22–37)	29 (23–38)	29 (22–39)	0.50
Dense LDL cholesterol, mg/dl	46 (34–63)	47 (35–67)	48 (36–62)	0.41
Low density LDL cholesterol, mg/dl	46 (33–60)	43 (33–60)	44 (32–57)	0.27
Logarithmic LDL density ratio (LLDR)	0.10 (−0.36;0.49)	0.19 (−0.29;0.55)	0.14 (−0.28;0.54)	0.18
T-Chol/HDL-C ratio	3.65 (3.10–4.36)	3.87 (3.163–4.54)	3.78 (3.11–4.48)	0.08
TG/HDL-C ratio	1.61 (1.10–2.36)	1.70 (1.137–2.89)	1.76 (1.14–2.54)	0.03
Men (n = 1,595)	n = 1,537	n = 30	n = 28
Total cholesterol, mg/dl	215 (189–243)	222 (187–239)	213 (186–233)	0.81
LDL-real cholesterol, mg/dl	114 (93–133)	120 (84–133)	120 (92–136)	0.99
VLDL-cholesterol, mg/dl	26 (21–35)	30 (24–39)	23 (19–34)	0.90
Total HDL-cholesterol, mg/dl	48 (41–56)	50 (39–55)	46 (40–55)	0.53
Triglycerides, mg/dl	121 (88–171)	129 (88–213)	112 (78–143)	0.75
Non-HDL cholesterol, mg/dl	165 (140–193)	175 (140–193)	167 (134–184)	0.89
VLDL₃ –cholesterol, mg/dl	15 (12 –18)	16 (14 –19)	13 (12 –17)	0.98
HDL₂-cholesterol, mg/dl	11 (9 –15)	11 (9 –15)	10 (9 –14)	0.46
HDL₃-cholesterol, mg/dl	36 (32–42)	38 (31–41)	37 (31–41)	0.60
TRL-cholesterol, mg/dl	31 (24–39)	33 (28–41)	30 (23–33)	0.62
Dense LDL cholesterol, mg/dl	65 (46–84)	62 (37–82)	55 (40–91)	0.42
Low density LDL cholesterol, mg/dl	40 (31–51)	42 (29–51)	42 (28–48)	0.67
Logarithmic LDL density ratio (LLDR)	0.53 (0.19; 0.80)	0.49 (−.029;0.93)	0.52 (0.21; 0.79)	0.81
T-Chol/HDL-C ratio	3.66 (4.48–5.31)	3.87 (4.58–5.62)	3.61 (4.26–5.42)	0.88
TG/HDL-C ratio	2.54 (1.63–3.94)	2.59 (1.67–4.71)	2.23 (1.48–3.72)	1.00

p-values for linear trend across subgroups were obtained by Jonckheere-Terpstra test.

IQR-interquartile range; TRL: Triglyceride-rich lipoprotein remnants; Logarithmic LDL density ratio (LLDR) = ln [(LDL₃-C + LDL₄-C)/LDL₁-C + LDL₂-C)].

We found positive associations between MA and the second tertiles of T-Chol (OR, 1.56; 95% CI, 1.02–2.41) and non-HDL-C (OR, 1.63; 95% CI, 1.05–2.53), but no associations with the highest tertiles were observed in women (Table 3). Positive associations between MO and the highest tertiles of VLDL-C (OR, 1.61; 95% CI, 1.07–2.40) and TRL-C (OR, 1.55; 95% CI, 1.03–2.34) were also observed among women.

Table 3.

Odds ratio (95% CI) for the association with migraine and VAP lipid profile of 1,560 women in ELSA-Brasil.

	Non-migraine	Migraine without aura	Migraine with aura
Total cholesterol
Second tertile (202–236 mg/dl)^†
Crude	Reference (1.0)	0.91 (0.62–1.34)	1.28 (0.85–1.95)
Age and BMI adjusted	Reference (1.0)	1.07 (0.73–1.58)	1.59 (1.04–2.44)**
Multivariable adjusted	Reference (1.0)	1.07 (0.72–1.58)	1.56 (1.02–2.41)**
Third tertile (237–474 mg/dl)^†
Crude	Reference (1.0)	0.91 (0.62–1.32)	0.76 (0.48–1.20)
Age and BMI adjusted	Reference (1.0)	1.25 (0.84–1.86)	1.16 (0.72–1.89)
Multivariable adjusted	Reference (1.0)	1.27 (0.85–1.89)	1.19 (0.73–1.93)
LDL real-cholesterol
Second tertile (99–122 mg/dl)^†
Crude	Reference (1.0)	0.72 (0.49–1.07)	1.38 (0.91–2.08)
Age and BMI adjusted	Reference (1.0)	0.82 (0.55–1.22)	1.58 (1.03–2.41)**
Multivariable adjusted	Reference (1.0)	0.80 (0.53–1.19)	1.53 (0.99–2.35)
Third tertile (123–303 mg/dl)^†
Crude	Reference (1.0)	0.96 (0.67–1.39)	0.76 (0.47–1.22)
Age-adjusted	Reference (1.0)	1.21 (0.83–1.76)	1.03 (0.64–1.68)
Age and BMI adjusted	Reference (1.0)	1.22 (0.84–1.78)	0.99 (0.61–1.61)
Multivariable adjusted	Reference (1.0)	1.22 (0.83–1.79)	1.00 (0.61–1.64)
VLDL-cholesterol
Second tertile (19–25 mg/dl)^†
Crude	Reference (1.0)	0.95 (0.63–1.42)	0.99 (0.64–1.55)
Age and BMI adjusted	Reference (1.0)	1.05 (0.70–1.58)	1.07 (0.68–1.68)
Multivariable adjusted	Reference (1.0)	1.02 (0.68–1.55)	1.06 (0.67–1.67)
Third tertile (26–96 mg/dl)^†
Crude	Reference (1.0)	1.37 (0.94–1.99)	1.15 (0.75–1.77)
Age and BMI adjusted	Reference (1.0)	1.74 (1.17–2.58)**	1.33 (0.85–2.10)
Multivariable adjusted	Reference (1.0)	1.61 (1.07–2.40)**	1.23 (0.77–1.97)
Total HDL-cholesterol
First tertile (21–53 mg/dl)^†
Crude	Reference (1.0)	1.58 (1.07–2.33)**	1.57 (0.99–2.47)
Age, BMI adjusted	Reference (1.0)	1.44 (0.97–2.16)	1.17 (0.73–1.88)
Multivariable adjusted	Reference (1.0)	1.31 (0.87–1.98)	1.06 (0.66–1.71)
Second tertile (54–65 mg/dl)^†
Crude	Reference (1.0)	1.25 (0.84–1.89)	1.51 (0.95–2.38)
Age-adjusted	Reference (1.0)	1.13 (0.74–1.69)	1.30 (0.82–2.06)
Age, BMI adjusted	Reference (1.0)	1.15 (0.76–1.73)	1.22 (0.76–1.94)
Multivariable adjusted	Reference (1.0)	1.08 (0.71–1.64)	1.12 (0.70–1.80)
Triglycerides
Second tertile (80–115 mg/dl)^†
Crude	Reference (1.0)	0.99 (0.67–1.48)	0.89 (0.57–1.38)
Age, BMI-adjusted	Reference (1.0)	1.18 (0.79–1.76)	1.02 (0.65–1.61)
Multivariable adjusted	Reference (1.0)	1.16 (0.77–1.74)	1.10 (0.64–1.59)
Third tertile (116–1535 mg/dl)
Crude	Reference (1.0)	1.23 (0.84–1.79)	1.05 (0.69–1.61)
Age and BMI-adjusted	Reference (1.0)	1.61 (1.08–2.40)**	1.21 (0.77–1.91)
Multivariable adjusted	Reference (1.0)	1.48 (0.98–2.23)	1.13 (0.71–1.80)
Non-HDL-cholesterol
Second tertile (142–174 mg/dl)^†
Crude	Reference (1.0)	0.82 (0.55–1.21)	1.40 (0.91–2.14)
Age and BMI adjusted	Reference (1.0)	0.96 (0.64–1.43)	1.66 (1.07–2.56)**
Multivariable adjusted	Reference (1.0)	0.96 (0.64–1.43)	1.63 (1.05–2.53)**
Third tertile (175–380 mg/dl)^†
Crude	Reference (1.0)	0.99 (0.69–1.44)	0.91 (0.57–1.44)
Age and BMI adjusted	Reference (1.0)	1.33 (0.90–1.96)	1.25 (0.77–2.01)
Multivariable adjusted	Reference (1.0)	1.30 (0.88–1.92)	1.24 (0.76–2.01)
Multivariable adjusted	Reference (1.0)	1.27 (0.86–1.90)	1.04 (0.66–1.65)
HDL₂-cholesterol
First tertile (4–13 mg/dl)^†
Crude	Reference (1.0)	1.57 (1.10–2.30)**	1.45 (0.92–2.30)
Age, BMI-adjusted	Reference (1.0)	1.52 (1.02–2.27)**	1.13 (0.69–1.82)
Multivariable adjusted	Reference (1.0)	1.39 (0.92–2.09)	1.04 (0.63–1.69)
Second tertile (14–19 mg/dl)^†
Crude	Reference (1.0)	0.99 (0.67–1.49)	1.40 (0.89–2.19)
Age, BMI-adjusted	Reference (1.0)	0.94 (0.62–1.41)	1.16 (0.73–1.83)
Multivariable adjusted	Reference (1.0)	0.89 (0.59–1.36)	1.08 (0.68–1.73)
HDL₃-cholesterol
First tertile (17–38 mg/dl)^†
Crude	Reference (1.0)	1.57 (1.06–2.34)**	1.70 (1.08–2.67)**
Age and BMI-adjusted	Reference (1.0)	1.39 (0.93–2.10)	1.27 (0.79–2.02)
Multivariable adjusted	Reference (1.0)	1.26 (0.83–1.92)	1.15 (0.71–1.84)
Second tertile (39–45 mg/dl)^†
Crude	Reference (1.0)	1.38 (0.93–2.05)	1.39 (0.88–2.18)
Age, BMI-adjusted	Reference (1.0)	1.26 (0.84–1.88)	1.14 (0.71–1.81)
Multivariable adjusted	Reference (1.0)	1.17 (0.78–1.76)	1.05 (0.66–1.68)
TRL-cholesterol
Second tertile (25–34 mg/dl)^†
Crude	Reference (1.0)	1.10 (0.75–1.63)	0.85 (0.55–1.31)
Age, BMI-adjusted	Reference (1.0)	1.32 (0.89–1.96)	0.98 (0.63–1.54)
Multivariable adjusted	Reference (1.0)	1.23 (0.82–1.83)	0.91 (0.58–1.44)
Third tertile (35–88 mg/dl)
Crude	Reference (1.0)	1.26 (0.85–1.85)	1.06 (0.69–1.62)
Age and BMI-adjusted	Reference (1.0)	1.68 (1.12–2.51)**	1.39 (0.89–2.17)
Multivariable adjusted	Reference (1.0)	1.55 (1.03–2.34)**	1.29 (0.81–2.04)
VLDL₃-cholesterol
Second tertile (12-14 mg/dl)^†
Crude	Reference (1.0)	0.85 (0.56–1.28)	1.26 (0.81–1.95)
Age and BMI adjusted	Reference (1.0)	0.96 (0.63–1.46)	1.38 (0.88–2.16)
Multivariable adjusted	Reference (1.0)	0.93 (0.61–1.42)	1.35 (0.86–2.12)
Third tertile (15-46 mg/dl)^†
Crude	Reference (1.0)	1.23 (0.86–1.77)	1.08 (0.70–1.67)
Age-adjusted	Reference (1.0)	1.46 (1.01–2.12)**	1.36 (0.87–2.12)
Age and BMI adjusted	Reference (1.0)	1.52 (1.04–2.21)**	1.24 (0.79–1.95)
Multivariable adjusted	Reference (1.0)	1.40 (0.95–2.05)	1.14 (0.72–1.82)
Logarithmic LDL density ratio (LLDR)
Second tertile (−0.17; 0.36)
Crude	Reference (1.0)	1.18 (0.80–1.74)	1.13 (0.73–1.75)
Age, BMI-adjusted	Reference (1.0)	1.13 (0.77–0.68)	1.00 (0.64–1.57)
Multivariable adjusted	Reference (1.0)	1.11 (0.75–1.65)	0.96 (0.61–1.51)
Third tertile (0.36; 3.16)^†
Crude	Reference (1.0)	1.25 (0.85–1.84)	1.19 (0.77–1.84)
Age, BMI-adjusted	Reference (1.0)	1.16 (0.78–1.72)	0.99 (0.64–1.56)
Multivariable adjusted	Reference (1.0)	1.12 (0.75–1.66)	0.99 (0.63–1.54)
T-Chol/HDL-C ratio
Second tertile (3.30–4.11)^†
Crude	Reference (1.0)	1.17 (0.79–1.74)	1.21 (0.78–1.88)
Age and BMI-adjusted	Reference (1.0)	1.23 (0.82–1.84)	1.16 (0.74–1.82)
Multivariable adjusted	Reference (1.0)	1.18 (0.79–1.77)	1.12 (0.71–1.77)
Third tertile (4.11–12.52)^†
Crude	Reference (1.0)	1.43 (0.97–2.09)	1.22 (0.78–1.89)
Age, BMI-adjusted	Reference (1.0)	1.64 (1.10–2.44)**	1.22 (0.77–1.93)
Multivariable adjusted	Reference (1.0)	1.53 (1.02–2.29)**	1.15 (0.72–1.83)
Triglycerides/HDL-C ratio
Second tertile (1.28–2.09)^†
Crude	Reference (1.0)	1.16 (0.78–1.72)	1.13 (0.72–1.76)
Age, BMI-adjusted	Reference (1.0)	1.22 (0.82–1.82)	1.09 (0.69–1.71)
Multivariable adjusted	Reference (1.0)	1.17 (0.78–1.76)	1.05 (0.66–1.67)
Third tertile (2.09–52.93)^†
Crude	Reference (1.0)	1.36 (0.93–1.99)	1.29 (0.84–1.99)
Age and BMI-adjusted	Reference (1.0)	1.61 (1.07–2.41)**	1.28 (0.80–2.03)
Multivariable adjusted	Reference (1.0)	1.44 (0.95–2.19)	1.15 (0.72–1.86)

p-value < 0.001 compared to controls (non-migraine). **p-value < 0.05 compared to controls.

†

First tertile of total cholesterol (113–201 mg/dl), LDL real-C (21–98 mg/dl), VLDL-C (11–18 mg/dl), triglycerides (30–79 mg/dl), non-HDL–C (52–141 mg/dl), VLDL₃-C (7–11 mg/dl), triglycerides rich remnant (10–24 mg/dl), Logarithmic LDL density ratio (LLDR) (−2.44; −0.18), T-Chol/HDL ratio (1.63–3.30), Triglycerides/HDL ratio (0.32–1.28), and third tertiles of total HDL-C (66–133 mg/dl), HDL₂-C (20–67 mg/dl), HDL₃-C (46–83 mg/dl) and were considered as reference group, respectively.

All Multivariate analyses were adjusted by age, educational level, smoking, hypertension, diabetes, and body mass index. OR: Odds ratio; CI: Confidence interval; Triglyceride-rich lipoprotein. Logarithmic LDL density ratio (LLDR) = ln [(LDL₃-C + LDL4-C)/LDL₁-C + LDL₂-C)]

For men, we found a positive association with the highest tertile of VLDL₃-C (OR, 3.87; 95% CI, 1.23–12.19) and MO after multivariate adjustment (Table 4).

Table 4.

Odds ratio (95% CI) for the association with migraine, triglycerides and HDL-cholesterol lipid profile of 1,595 men in the ELSA-Brasil.

	Non-migraine	Migraine without aura	Migraine with aura
Total cholesterol
Second tertile (198–232 mg/dl)^†
Crude	Reference (1.0)	1.23 (0.48–3.15)	1.09 (0.46–2.58)
Age and BMI adjusted	Reference (1.0)	1.30 (0.51–3.33)	1.20 (0.50–2.85)
Multivariable adjusted	Reference (1.0)	1.29 (0.50–3.33)	1.22 (0.51–2.91)
Third tertile (233–415 mg/dl)^†
Crude	Reference (1.0)	1.50 (0.61–3.70)	0.70 (0.26–1.85)
Age-adjusted	Reference (1.0)	1.58 (0.64–3.91)	0.80 (0.30–2.14)
Age and BMI adjusted	Reference (1.0)	1.64 (0.66–4.07)	0.83 (0.31–2.21)
Multivariable adjusted	Reference (1.0)	1.66 (0.67–4.14)	0.84 (0.31–2.24)
LDL real-cholesterol
Second tertile (101–126 mg/dl)^†
Crude	Reference (1.0)	0.48 (0.18–1.29)	0.97 (0.38–2.45)
Age and BMI adjusted	Reference (1.0)	0.50 (0.19–1.35)	1.05 (0.41–2.68)
Multivariable adjusted	Reference (1.0)	0.49 (0.18–1.32)	1.04 (0.41–2.67)
Third tertile (127–270 mg/dl)^†
Crude	Reference (1.0)	1.01 (0.45–2.27)	1.12 (0.45–2.79)
Age-adjusted	Reference (1.0)	1.05 (0.46–2.35)	1.24 (0.50–3.10)
Age and BMI adjusted	Reference (1.0)	1.10 (0.49–2.47)	1.29 (0.52–3.23)
Multivariable adjusted	Reference (1.0)	1.07 (0.47–2.44)	1.30 (0.52–3.24)
VLDL-cholesterol
Second tertile (23–31 mg/dl)^†
Crude	Reference (1.0)	1.99 (0.74–5.36)	0.46 (0.17–1.22)
Age and BMI adjusted	Reference (1.0)	2.33 (0.86–6.33)	0.51 (0.19–1.38)
Multivariable adjusted	Reference (1.0)	2.35 (0.86–6.40)	0.51 (0.19–1.38)
Third tertile (32–208 mg/dl)^†
Crude	Reference (1.0)	2.00 (0.75–5.37)	0.69 (0.29–1.63)
Age and BMI adjusted	Reference (1.0)	2.49 (0.90–6.85)	0.77 (0.31–1.90)
Multivariable adjusted	Reference (1.0)	2.69 (0.97–7.48)	0.80 (0.32–2.00)
Total HDL-cholesterol
First tertile (23–43 mg/dl)^†
Crude	Reference (1.0)	1.02 (0.42–2.48)	1.15 (0.44–3.01)
Age and BMI adjusted	Reference (1.0)	1.17 (0.47–2.92)	1.23 (0.46–3.33)
Multivariable adjusted	Reference (1.0)	1.19 (0.48–2.96)	1.26 (0.47–3.41)
Second tertile (44–52 mg/dl)^†
Crude	Reference (1.0)	1.07 (0.44–2.58)	1.46 (0.58–3.67)
Age and BMI adjusted	Reference (1.0)	1.13 (0.46–2.75)	1.48 (0.58–3.75)
Multivariable adjusted	Reference (1.0)	1.18 (0.48–2.89)	1.50 (0.59–3.83)
Triglycerides
Second tertile (99–148 mg/dl)^†
Crude	Reference (1.0)	0.38 (0.14–1.07)	0.82 (0.35–1.92)
Age and BMI-adjusted	Reference (1.0)	0.43 (0.15–1.23)	0.91 (0.38–2.18)
Multivariable adjusted	Reference (1.0)	0.43 (0.15–1.25)	0.93 (0.39–2.23)
Third tertile (149–3,930 mg/dl)
Crude	Reference (1.0)	0.92 (0.41–2.03)	0.49 (0.19–1.33)
Age and BMI-adjusted	Reference (1.0)	1.11 (0.48–2.57)	0.56 (0.19–1.59)
Multivariable adjusted	Reference (1.0)	1.23 (0.52–2.91)	0.59 (0.20–1.70)
Non-HDL-cholesterol
Second tertile (149–182 mg/dl)^†
Crude	Reference (1.0)	0.78 (0.31–2.00)	0.80 (0.33–1.95)
Age and BMI adjusted	Reference (1.0)	0.85 (0.33–2.17)	0.90 (0.37–2.22)
Multivariable adjusted	Reference (1.0)	0.82 (0.32–2.12)	0.90 (0.37–2.21)
Third tertile (183–375 mg/dl)^†
Crude	Reference (1.0)	1.20 (0.51–2.79)	0.72 (0.29–1.82)
Age and BMI adjusted	Reference (1.0)	1.36 (0.58–3.22)	0.86 (0.34–2.19)
Multivariable adjusted	Reference (1.0)	1.40 (0.59–3.31)	0.88 (0.34–2.24)
HDL₂-cholesterol
First tertile (3–9 mg/dl)^†
Crude	Reference (1.0)	1.33 (0.54–3.28)	1.44 (0.59–3.50)
Age and BMI-adjusted	Reference (1.0)	1.53 (0.60–3.90)	1.48 (0.59–3.75)
Multivariable adjusted	Reference (1.0)	1.57 (0.62–3.99)	1.53 (0.60–3.89)
Second tertile (10–13 mg/dl)^†
Crude	Reference (1.0)	1.04 (0.41–2.65)	0.73 (0.26–2.02)
Age and BMI-adjusted	Reference (1.0)	1.13 (0.44–2.93)	0.75 (0.26–2.10)
Multivariable adjusted	Reference (1.0)	1.15 (0.44–2.98)	0.74 (0.26–2.10)
HDL₃-cholesterol
First tertile (16–33 mg/dl)^†
Crude	Reference (1.0)	0.95 (0.40–2.26)	1.65 (0.63–4.28)
Age and BMI-adjusted	Reference (1.0)	1.08 (0.44–2.62)	1.78 (0.67–4.76)
Multivariable adjusted	Reference (1.0)	1.09 (0.45–2.66)	1.79 (0.67–4.79)
Second tertile (34–39 mg/dl)^†
Crude	Reference (1.0)	0.81 (0.33–1.98)	1.42 (0.54–3.75)
Age and BMI-adjusted	Reference (1.0)	0.85 (0.35–2.07)	1.44 (0.54–3.83)
Multivariable adjusted	Reference (1.0)	0.86 (0.35–2.10)	1.43 (0.54–3.81)
TRL-cholesterol
Second tertile (27–35 mg/dl)^†
Crude	Reference (1.0)	1.98 (0.74–5.31)	0.84 (0.37–1.88)
Age, BMI-adjusted	Reference (1.0)	2.17 (0.80–5.86)	0.90 (0.39–2.05)
Multivariable adjusted	Reference (1.0)	2.22 (0.82–6.01)	0.90 (0.39–2.05)
Third tertile (36–112 mg/dl)
Crude	Reference (1.0)	1.88 (0.70–5.01)	0.29 (0.09–0.89)**
Age and BMI-adjusted	Reference (1.0)	2.20 (0.81–5.99)	0.33 (0.11–1.03)
Multivariable adjusted	Reference (1.0)	2.32 (0.85–6.32)	0.33 (0.10–1.03)
VLDL₃-cholesterol
Second tertile (13–16 mg/dl)^†
Crude	Reference (1.0)	2.26 (0.72–7.05)	0.41 (0.16–1.03)
Age and BMI adjusted	Reference (1.0)	2.59 (0.82–8.15)	0.44 (0.17–1.12)
Multivariable adjusted	Reference (1.0)	2.59 (0.82–8.19)	0.44 (0.17–1.12)
Third tertile (17–63 mg/dl)^†
Crude	Reference (1.0)	2.87 (0.94–8.77)	0.50 (0.21–1.23)
Age and BMI adjusted	Reference (1.0)	3.61 (1.15–11.27)**	0.57 (0.23–1.45)
Multivariable adjusted	Reference (1.0)	3.87 (1.23–12.19)**	0.58 (0.23–1.47)
Logarithmic LDL density ratio (LLDR)
Second tertile (0.32–0.71)
Crude	Reference (1.0)	0.35 (0.13–0.98)*	1.09 (0.44–2.71)
Age, BMI-adjusted	Reference (1.0)	0.36 (0.13–1.02)	1.09 (0.44–2.74)
Multivariable adjusted	Reference (1.0)	0.37 (0.13–1.05)	1.12 (0.45–2.82)
Third tertile (0.71–2.78)^†
Crude	Reference (1.0)	0.78 (0.35–1.73)	0.99 (0.39–2.52)
Age, BMI-adjusted	Reference (1.0)	0.82 (0.37–1.84)	1.02 (0.40–2.62)
Multivariable adjusted	Reference (1.0)	0.86 (0.38–1.93)	1.10 (0.43–2.85)
T-Chol/HDL-C ratio
Second tertile (3.97–4.98)^†
Crude	Reference (1.0)	1.23 (0.51–2.99)	1.12 (0.45–2.78)
Age and BMI-adjusted	Reference (1.0)	1.38 (0.56–3.41)	1.24 (0.49–3.12)
Multivariable adjusted	Reference (1.0)	1.37 (0.56–3.40)	1.25 (0.49–3.15)
Third tertile (4.98–10.38)^†
Crude	Reference (1.0)	1.11 (0.45–2.76)	1.00 (0.40–2.55)
Age, BMI-adjusted	Reference (1.0)	1.31 (0.52–3.32)	1.13 (0.43–2.96)
Multivariable adjusted	Reference (1.0)	1.35 (0.53–3.44)	1.13 (0.43–2.98)
Triglycerides/HDL-C ratio
Second tertile (1.86–3.40)^†
Crude	Reference (1.0)	0.99 (0.41–2.40)	0.66 (0.27–1.63)
Age, BMI-adjusted	Reference (1.0)	1.13 (0.46–2.82)	0.70 (0.28–1.78)
Multivariable adjusted	Reference (1.0)	1.16 (0.47–2.89)	0.69 (0.27–1.78)
Third tertile (3.40–83.62)^†
Crude	Reference (1.0)	0.99 (0.41–2.40)	0.66 (0.27–1.63)
Age and BMI-adjusted	Reference (1.0)	1.23 (0.49–3.13)	0.74 (0.28–1.94)
Multivariable adjusted	Reference (1.0)	1.35 (0.52–3.48)	0.76 (0.28–2.03)

p-value < 0.001 compared to controls (non-migraine).

p-value < 0.05 compared to controls.

†

First tertiles of total cholesterol (82–197 mg/dl), LDL real –C (18–100 mg/dl), VLDL-C (11–22 mg/dl), triglycerides (30–79 mg/dl), non-HDL-C (44–148 mg/dl), VLDL₃-C (7–12 mg/dl), triglycerides-rich remnant (10–26 mg/dl), Logarithmic LDL density ratio (LLDR) (−1.94; 0.32), T-Chol/HDL ratio (1.39–3.97), Triglycerides/HDL ratio (0.39–1.86), and third tertiles of total HDL-C (66–133 mg/dl), HDL₂-C (20–67 mg/dl), HDL₃-C (46–83 mg/dl) were considered as the reference group, respectively.

All Multivariate analyses were adjusted by age, educational level, smoking, hypertension, diabetes, and body mass index. OR: Odds ratio; CI: Confidence interval; TRL: Triglyceride-rich lipoprotein. Logarithmic LDL density ratio (LLDR) = ln [(LDL3-C + LDL4-C)/LDL1-C + LDL2-C)].

Additional adjustments for HS-CRP or migraine prophylaxis did not change the directions and significance of our findings (Data not shown).

Data on all samples regarding lipid distribution and logistic regression are demonstrated in Supplemental Tables 1 and 2.

Discussion

Among middle-aged participants without previous CVD and not taking lipid lowering medications from the ELSA-Brasil cohort, elevated levels of TRL-C and their cholesterol-rich remnants were observed mainly in MO. Gender related differences were observed with regards to VLDL-C and TRL-C levels that were positively associated with MO, but only in women. In addition, VLDL₃-C was positively associated to MO only in men. No associations were observed in the total of HDL-C and LDL_r-C and their respective subfractions with migraine after multivariate adjustments.

Our main results do not support prior evidence that suggested a more atherogenic lipid profile among MA (1 –3,10,11). In fact, most of our relevant findings regarding abnormalities in the lipid profile were observed in MO, which also had a higher frequency of diabetes and hypertension compared to MA. We found that migraineurs were three years younger than non-migraineurs (47 vs. 50 years old), which could justify differences in some CVRF frequencies such as hypertension and diabetes that in fact were more often found in the oldest group (non-migraineurs) (see Table 1). However, after multivariate analyses, we found positive associations between lipids subfractions and migraine headaches. Although we cannot rule out the effect of residual confounders influencing this relationship, in previous publication of ELSA-Brasil (4) we also found a positive association between migraine and dyslipidemia among women and, in the opposite direction, we found an inverse relationship between migraine and hypertension in men.

Most previous studies have reported higher frequencies of T-Chol > 200 mg/dl in migraineurs (1,2,10,11). Our results were similar to the Genetic Epidemiology of Migraine (GEM) study for the association between MA and the second tertile of T-Chol, but no significant association were found with the third tertile, particularly in women. In the GEM study, a cohort consisting of 5,755 middle-aged individuals, a positive association with T-Chol ≥ 240 mg/dl and MA (OR, 1.43; 95% CI, 1.0–2.1) was demonstrated (1). In the Epidemiology of Vascular Ageing (EVA) study, the OR was even higher for the association between MA and the highest tertile of T-Chol in the elderly population. However, the estimate was very imprecise (OR, 5.97; 95% CI, 1.29–27.61) (9). In the Nord-Trøndelag Health Study (HUNT 2), T-Chol was not associated to MA (3).

Our most consistent findings were the positive associations with MO, VLDL-C and VLDL₃-C in women and men, respectively. The VLDL₃, which is the smallest and densest VLDL subfraction, has been previously associated with the presence of subclinical atherosclerosis particularly in diabetics, and might be associated with higher CVD risk (24 –26). VLDL is fractioned into remnant VLDL particles (VLDL₃) and IDL, which contains proportionally less triglycerides but more cholesterol than larger, less dense VLDL, and, consequently, is implicated in atherogenesis due to its direct deposition into the sub-intimal space in the arterial wall (27,28). Significantly, no previous study has reported the association between VLDL₃-C and migraine headache.

We found positive associations between triglycerides and MO for the total sample (Supplemental material), but not in analyses stratified by sex. Few studies have reported the relationship between triglycerides and migraine (10). In a cross-sectional evaluation in the EVA study, a positive relationship between MA and the highest levels of triglycerides was encountered after multivariate adjustment (RR, 4.42; 95% CI, 1.32–14.77) (10). In the EVA study, this direct relationship was stronger than in ELSA-Brasil and it was evidenced among MA. Possible explanations for these dissimilarities in both studies could be due to differences in study populations, such as age and sample size of migraineurs. In the EVA study, participants were much older and the sample of MA was much smaller (n = 23) than ours (n = 164). Indeed, this small number of observations in the EVA study probably did not allow analyses stratified by sex (10).

With the use of VAP to separate the lipoproteins, we found a positive association between MO and TRL-C, particularly in women. Although the evidence of an association between migraine and CVD is stronger for MA rather than for MO according to the literature (5 –8); in the present analyses we only found some associations between MA and lipid subfractions in the second tertiles of TC, LDL real and non-HDL-cholesterol. It may be possible that we did not find statistical significance in the third tertiles of these lipids subfractions due to the small number of observations of participants with MA. TRL composed of VLDL particles (VLDL₃-C + VLDL₁₊₂-C) and IDL in fasting, or by both lipoproteins with chylomicron remnants in the fed state, are risk factors for CVD (17). Recently, epidemiological and Mendelian randomization studies have associated increased non-fasting triglycerides, a surrogate marker of elevated remnant cholesterol, to increased CHD risk (17,28,29). Not only can these cholesterol-rich lipoproteins, which are small particles, penetrate the artery wall and generate foam cells, but they can also produce a heightened systemic inflammatory state that induces endothelial dysfunction and contributes to atherogenesis (30). Inflammation and endothelial dysfunction could function as a causal link between migraine and CVD (31,32).

The inverse relationship between total HDL-C concentrations and migraine has been described by other authors (1,3,9,11). Among those previous studies, two reported positive associations for migraine and low levels of total HDL-C (1,3), while two others did not (9,11). In the HUNT 2 study, individuals with HDL-C <42 mg/dl had the highest odds of having MA after multivariate adjustment (3). In the GEM study, an increased OR for the relationship between MA and T-Chol/HDL-C ratio ≥ 5 (multivariate OR, 1.64; 95% CI; 1.1–1.64) was found (1). This risk was increased for women compared to men (1).

In our study, we did not find any association with HDL-C and its subfractions (29) after multivariate analyses. In fact, we found progressively higher OR for the associations with T-Chol/HDL-C ratio > 3.57 and MO in all samples. Unlike previous studies (1), we did not confirm these findings after multivariate analyses according to gender. Previous large observational studies have reported T-Chol/HDL-C ratio as a better predictor of coronary events compared to total-Chol, LDL-C or HDL-C individually (33,34). Similarly, we observed high ORs for the associations with the third tertile of triglycerides/HDL-C ratio and MO in all samples, but not by sex. The ratio of triglycerides to HDL-C, which is inversely associated with dense LDL-C particles, was linked with a more atherogenic profile, suggesting it is an even more powerful independent predictor of CHD than T-Chol/HDL-c (35).

We also demonstrated a positive association between MO and the second tertile of non-HDL-C for all samples, but similar to T-Chol, we found no association with the third tertile of non-HDL-C. Non-HDL-C quantifies the cholesterol content of all atherogenic lipoproteins, including LDL-C. Among all cholesterol measures, non-HDL-C has been shown to be the best predictor for CHD and stroke (24). A previous cross-sectional observation from ELSA-Brasil demonstrated a slightly increased OR of 1.25 (95% CI, 1.13–1.38) for the relationship between migraine and dyslipidemia at baseline (LDL-C concentrations > 130 mg/dl and/or the use of lipid-lowering medications) in women (4). In the present study, more robust analysis with lipid subfractions demonstrated some positive associations with real LDL-C and MA in age and BMI-adjusted models. However, they lost significance after multivariate adjustment in women and were not significant in men. Using the VAP approach (15), it is possible to evaluate LDL-C particles into two distinct LDL phenotypes: Pattern B, with a predominance of small, dense LDL particles, and pattern A, with a predominance of large, more buoyant LDL-C particles. Both small and large LDL-C fractions are associated with CVD outcomes (33 –35). In our analysis, an inverse relationship between small dense LDL-C and migraine, regardless of aura, was suggested for the entire sample, but was not confirmed in multivariate analyses stratified by sex.

In summary, when we consider all these findings together our study suggests a lack of association between LDL-C and migraine. In fact, an unfavorable lipid profile that conferred positive associations with migraine was more linked to the highest levels of triglyceride rich-lipoprotein than total triglycerides, LDL-C and HDL-C subfractions in both sexes.

This study has some strength. ELSA-Brasil brings important information about lipoproteins on migraine risk according to aura, which were previously reported by few authors (9). While our study population is relatively young, the frequencies of cardiovascular risk factors are high, including lipid disorders. Therefore, it is possible to study this association using our sample. Further, we were able to exclude individuals who were receiving lipid-lowering treatment and/or with previous CVD history from our analyses.

For the first time, using the VAP technique, we report unique data about VLDL-C, HDL-C subfractions, TRL-C and migraine according to aura symptoms in both sexes. The exhaustive evaluation of these lipid subfractions may represent an important tool for assessing emerging lipid risk factors for the association of migraine and incident CVD. Some studies suggested that with the VAP cholesterol methodology, it is possible to early and accurately identifying people under CVD risk at a detection rate about more than twice that of conventional cholesterol panels (37).

The limitations of our study are mainly due to its cross-sectional design, which does not allow us to make inferences about causal relationships. Since ELSA-Brasil is an observational study, we did not interfere in personal medications previously prescribed by the participants’ doctors. However, prospective analyses foreseen in ELSA-Brasil will allow us to better evaluate this complex relationship between migraine headache, lipids, lipoprotein subfractions, lipid-lowering drugs and genetic associations possibly involved in this process and that can corroborate to incidental CVD outcomes. Another possible limitation of our study could be the inclusion of participants with other headaches in the control group. However, after performing a post-roc sensitivity analysis excluding participants with other headaches (n = 1,665) from the control group, previous associations between lipid subfractions and migraine headache according to sex were kept. Regarding the misdiagnosis of migraine among participants with headache, we cannot rule out some kind of information bias. However, this possibility is unlikely because we used a very robust questionnaire of headache based on the International Headache Society (IHS-200), previously validated and widely used in Brazil.

Moreover, we found some random associations with the second tertile of CT, probably due to non-HDL-cholesterol, mostly in women, but not in the third tertiles of these lipid subfractions. We could associate a worse cardiovascular profile in migraineurs in the second tertile. However, in a secondary analysis we did not find higher frequencies of hypertension, diabetes or obesity in the second tertiles of T-Chol and Non-HDL-cholesterol compared to other tertiles.

There are no previous data in the literature with VAP methodology in migraineurs to allow comparisons with more specific lipid sub-fractions as we have done here. It is also possible that we do not have powerful analyses in MA because of the small number of such patients, although they are superior to previous studies (2,9). Thus, our findings yield important new information about the role of triglyceride-rich lipoproteins as possible predictors or markers of CVD outcomes in migraineurs . Moreover, working on prospective data, we will be able to evaluate the complex causal relationship between migraine and incident CVD outcomes.

Conclusions

In middle-aged participants without CVD or lipid disorders requiring medication, the worst lipid profiles in migraineurs was determined by the highest levels of TRL-C and their cholesterol-rich remnants, particularly among those without aura symptoms. Gender-related differences were observed with regard to VLDL-C and TRL-C levels that were positively associated with MO, but only in women. In addition, VLDL3-C was positively associated to MO only in men.

Article highlights

ELSA-Brasil brings unique information based on the VAP technique about VLDL-C, HDL-C subfractions, triglyceride-rich lipoprotein cholesterol (TRL-C) and migraine according to aura symptoms in both sexes.

In middle-aged participants without CVD or lipid disorders requiring medication the worst lipid profiles in migraineurs was determined by the highest levels of VLDL-C, VLDL3-C and TRL-C, particularly among those without aura symptoms.

Outcoming data in ELSA-Brasil will be very useful to elucidate the possible role of lipoproteins as potential risk factors for CVD outcomes in the future.

Footnotes

Acknowledgements

The authors thank Atherotech (Birmingham, AL), who performed all lipids measurements with an unrestricted grant. Atherotech had no access to the study, or its analysis, and had no right to approve or disapprove this manuscript.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr Alessandra C Goulart, Dr Paulo A Lotufo, Itamar S Santos, Márcio S Bittencourt, Michael J Blaha, Steven Jones and Isabela M Benseñor report no disclosures. Raul D Santos has received honoraria related to consulting and or speaker activities from Amgen, Aegerion, Akcea, AstraZeneca, Biolab, Boehringer Ingelheim, Cerenis, Eli Lilly, Genzyme, Kowa, Merck, Pfizer (including those for services as a member of the Steering Committee for the SPIRE clinical trial program), Sanofi/Regeneron, Torrent, and Unilever. Peter Toth has received honoraria for consulting or being a member of the speaker’s bureau for Amarin, Amgen, AstraZeneca, Kowa, Merck, and Regeneron-Sanofi.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Scher

Terwindt

Picavet

et al.

Cardiovascular risk factors and migraine: The GEM population-based study. Neurology 2005; 64: 614–620.

Stang

Carson

Rose

et al.

Headache, cerebrovascular symptoms, and stroke: The Atherosclerosis Risk in Communities Study. Neurology 2005; 64: 1573–1577.

Winsvold

Hagen

Aamodt

et al.

Headache, migraine and cardiovascular risk factors: The HUNT study. Eur J Neurol 2011; 18: 504–511.

Goulart

Santos

Lotufo

et al.

Gender aspects of the relationship between migraine and cardiovascular risk factors: A cross-sectional evaluation in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Cephalalgia 2015; 35: 1103–1114.

Schürks

Rist

Bigal

et al.

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

Kurth

Gaziano

Cook

et al.

Migraine and risk of cardiovascular disease in women. JAMA 2006; 296: 283–291.

Kurth

Gaziano

Cook

et al.

Migraine and risk of cardiovascular disease in men. Arch Intern Med 2007; 167: 795–801.

Kurth

Winter

Eliassen

et al.

Migraine and risk of cardiovascular disease in women: Prospective cohort study. BMJ 2016; 353: i2610–i2610.

Mattiuzzi

Cervellin

Lippi

. Epidemiological associations between migraine and lipoprotein (a): A systematic review. J Thromb Thrombolysis 2015; 39: 113–117.

10.

Rist

Tzourio

Kurth

. Associations between lipid levels and migraine: Cross-sectional analysis in the EVA study. Cephalalgia 2011; 31: 1459–1465.

11.

Cirillo

Stellato

Lombardi

et al.

Headache and cardiovascular risk factors: Positive association with hypertension. Headache 1999; 39: 409–416.

12.

Kurth

Ridker

Buring

. Migraine and biomarkers of cardiovascular disease in women. Cephalalgia 2008; 28: 49–56.

13.

Teber

Bektas

Yılmaz

et al.

Lipoprotein a levels in pediatric migraine. Pediatr Neurol 2011; 45: 225–228.

14.

Cinzia

Daniela

Elena

et al.

Lipoprotein (a) [Lp(a)]: A possible link between migraine and stroke. Transl Res 2009; 153: 44–47.

15.

Kulkarni

. Cholesterol profile measurement by vertical auto profile method. Clin Lab Med 2006; 26: 787–802.

16.

Ahmed

Elshazly

Martin

et al.

Ratio of dense to buoyant LDL subclass is associated with LDL density phenotype (VLDL-5). Open Chem Biomed Methods J 2013; 6: 1–5.

17.

Nordestgaard BG and Varbo A. Triglycerides and cardiovascular disease. Lancet 2014; 16; 384: 626–635.

18.

Schmidt

Duncan

Mill

et al.

Cohort profile: Longitudinal Study of Adult Health (ELSA-Brasil). Int J Epidemiol 2015; 44: 68–75.

19.

Aquino

Barreto

Bensenor

et al.

Brazilian Longitudinal Study of Adult Health (ELSA-Brasil): Objectives and design. Am J Epidemiol 2012; 175: 315–324.

20.

Bensenor

Griep

Pinto

et al.

Routines of organization of clinical tests and interviews in the ELSA-Brasil investigation center. Rev Saude Publica 2013; 47: 37–47. [Article in Portuguese].

21.

The Headache Classification Committee of the International Headache Society: The international classification of headache disorders, 3^rd edn. Cephalalgia 2013; 33: 629–708.

22.

Benseñor

Lotufo

Pereira

et al.

Validation of a questionnaire for the diagnosis of headache in an outpatient clinic at a university hospital [in Portuguese]. Arq Neuro-Psiquiatria 1997; 55: 364–369.

23.

Silberstein

Holland

Freitag

et al.

Evidence-based guideline update: Pharmacologic treatment for episodic migraine prevention in adults: Report of the Quality Standards Subcommittee of the American Academy of Neurology and the American Headache Society. Neurology 2012; 78: 1337–1345.

24.

Chor

Alves

Giatti

et al.

Questionnaire development in ELSA-Brasil: Challenges of a multidimensional instrument. Rev Saude Publica 2013; 47: 27–36.

25.

Fedeli

Vidigal

Leite

et al.

Logistics of collection and transportation of biological samples and the organization of the central laboratory in the ELSA-Brasil. Rev Saude Publica 2013; 47: 63–71.

26.

Emerging Risk Factors Collaboration

Di Angelantonio

Sarwar

Perry

et al.

Major lipids, apolipoproteins, and risk of vascular disease. JAMA 2009; 302: 1993–2000.

27.

Amor

Catalan

Pérez

et al.

Nuclear magnetic resonance lipoprotein abnormalities in newly-diagnosed type 2 diabetes and their association with preclinical carotid atherosclerosis. Atherosclerosis 2016; 247: 161–169.

28.

Ren

Grundy

Liu

et al.

Long-term coronary heart disease risk associated with very-low-density lipoprotein cholesterol in Chinese: The results of a 15-Year Chinese Multi-Provincial Cohort Study (CMCS). Atherosclerosis 2010; 211: 327–332.

29.

Hodis

. Triglyceride-rich lipoprotein remnant particles and risk of atherosclerosis (editorial). Circulation 1999; 99: 2852–2854.

30.

Grundy

Cleeman

Merz

et al.

Implications of recent clinical trials for the National Cholesterol Education Program Adult Treatment Panel III Guidelines. Circulation 2004; 110: 227–239.

31.

Langsted

Freiberg

Nordestgaard

. Fasting and nonfasting lipid levels: Influence of normal food intake on lipids, lipoproteins, apolipoproteins, and cardiovascular risk prediction. Circulation 2008; 118: 2047–2056.

32.

Varbo

Benn

Tybjærg-Hansen

et al.

Elevated remnant cholesterol causes both low-grade inflammation and ischemic heart disease, whereas elevated low-density lipoprotein cholesterol causes ischemic heart disease without inflammation. Circulation 2013; 128: 1298–1309.

33.

Larsen JS, Skaug EA, Wisløff U, et al. Migraine and endothelial function: The HUNT3 Study. Cephalalgia 2016; 36: 1341–1349.

34.

Liman

Bachelier-Walenta

Neeb

et al.

Circulating endothelial microparticles in female migraineurs with aura. Cephalalgia 2015; 35: 88–94.

35.

Castelli

Garrison

Wilson

PWF

et al.

Incidence of coronary heart disease and lipoprotein cholesterol levels. The Framingham Study. JAMA 1986; 256: 2835–2838.

36.

Grover

Palmer

Coupall

. Serum lipid screening to identify high risk individuals for coronary death. The results of the Lipid Research Clinics prevalence cohort. Arch Intern Med 1994; 154: 679–684.

37.

Gaziano

Hennekens

O’Donnell

et al.

Fasting triglycerides, high density lipoprotein, and risk of myocardial infarction. Circulation 1997; 96: 2520–2525.

38.

Superko

. Did grandma give you heart disease? The new battle against coronary artery disease. Am J Cardiol 1998; 82: 34Q–46Q.

39.

Segrest JP, Garber DW, Brouillette CG, et al. The Amphipathic α Helix: A Multifunctional Structural Motif in Plasma Apolipoproteins. Advances in Protein Chemistry 1994; 45: 303–369.

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

0.13 MB