Episodic migraine is associated with a more atherogenic lipid profile rather than independent insulin resistance: a case-control study Authors: Simona Szabóová, Oľga Duraníková, Balázs Örzsik, Monika Glevická, Igor Straka, Peter Valkovič Published: 2026 ABSTRACT AbstractBackgroundCardiometabolic factors may influence migraine biology through metabolic, inflammatory, and vascular mechanisms, but studies examining insulin resistance and dyslipidaemia in migraine have yielded inconsistent results, partly due to heterogeneity in study design, fasting status, and timing of blood sampling relative to migraine attacks. We evaluated insulin-resistance indices and lipid profiles, including derived lipid ratios, in episodic migraine compared with healthy controls under standardised interictal fasting conditions.MethodsIn this cross-sectional case-control study, 45 adults with episodic migraine and 45 healthy controls aged 18–50 years underwent 12-hour fasting blood sampling during the interictal period. Co-primary endpoints were homeostatic model assessment of insulin resistance (HOMA-IR), total cholesterol to high-density lipoprotein cholesterol ratio (TC/HDL-C ratio), and atherogenic index of plasma (AIP). Multivariable linear regression models were adjusted for age, sex, body mass index, and smoking.ResultsAfter adjustment for age, sex, BMI, and smoking, episodic migraine remained associated with a higher TC/HDL-C ratio (B = 0.956, p < 0.001), higher AIP (B = 0.122, p = 0.008), higher total cholesterol, LDL-C, non-HDL-C, and fasting glucose, and lower HDL-C, whereas no adjusted association was found for HOMA-IR or fasting insulin. In unadjusted analyses, triglycerides were also higher in the migraine group. Monthly migraine days and attack intensity were not associated with the co-primary metabolic indices. Exploratory subgroup analyses showed nominal lipid-related differences by aura status, which did not remain significant after false discovery rate correction.ConclusionsUnder standardised interictal fasting conditions, episodic migraine was associated with a more atherogenic lipid profile, including higher TC/HDL-C ratio and AIP. In contrast, fasting surrogate markers of insulin resistance were not associated with episodic migraine after adjustment for selected covariates, although small effects cannot be excluded. Larger longitudinal studies with more detailed metabolic assessment are needed to confirm these findings.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12944-026-02997-0.Keywords: Episodic migraine, Insulin resistance, Dyslipidaemia, Lipid profile, Cardiometabolic risk Background Migraine is a highly prevalent and disabling primary headache disorder characterised by recurrent attacks of moderate to severe headache, typically pulsatile, unilateral, aggravated by physical activity, and accompanied by symptoms such as photophobia, phonophobia, nausea and/or vomiting [1, 2]. According to recent Global Burden of Disease analyses, migraine affects approximately 1.2 billion people worldwide and remains one of the leading causes of disability globally, with women bearing a disproportionate share of that burden [3–5]. Migraine is a complex neurobiological disorder that involves both neuronal and vascular mechanisms and is influenced by genetic susceptibility and environmental factors [1, 6]. As no validated diagnostic biomarker is currently available, diagnosis remains based on clinical criteria defined in the International Classification of Headache Disorders, 3rd Edition (ICHD-3) [2]. Despite advances in understanding its pathophysiology, migraine remains a major global health burden, underscoring the need for effective prevention and management strategies [5]. Beyond the established trigeminovascular framework, increasing evidence suggests that cardiometabolic factors may influence migraine biology by affecting attack threshold and clinical phenotype, rather than being primary causes of attacks [7, 8]. This is consistent with growing recognition of migraine as a systemic disorder with vascular and metabolic components [8, 9]. In particular, insulin resistance (IR) and dyslipidaemia have been proposed as potential contributors to migraine pathophysiology through interconnected metabolic and vascular mechanisms, including altered energy metabolism, chronic low-grade inflammation, endothelial dysfunction, and oxidative stress [8, 10, 11]. Migraine is also associated with several cardiometabolic comorbidities. Epidemiological studies have reported links between migraine and obesity, metabolic syndrome, and IR, suggesting shared biological pathways involving inflammation, metabolic dysregulation, and vascular dysfunction [9, 12–14]. In addition to observational associations, emerging genetic evidence suggests a shared biological architecture between migraine and lipid metabolism [15–18]. Large-scale cross-trait analyses have demonstrated genetic overlap between migraine and several lipid traits, particularly lipoprotein measures related to triglyceride-rich particles and high-density lipoprotein cholesterol metabolism [16, 17]. Drug-target Mendelian randomization studies have provided additional support for a role of lipid-related pathways, including 3-hydroxy-3-methylglutaryl-coenzyme A reductase and lipoprotein lipase, in migraine biology [18]. However, prior observational studies examining IR and dyslipidaemia in migraine have yielded inconsistent results, likely reflecting heterogeneity in study design and phenotyping, including differences in population characteristics, fasting status, ictal versus interictal sampling, age distribution, comorbidity burden, and medication exposure [8, 14, 19, 20]. In addition, relatively few studies have assessed whether IR and atherogenic lipid patterns cluster together in episodic migraine (EM) after accounting for adiposity and key confounders. To address these gaps, we conducted a cross-sectional case-control study comparing fasting insulin-resistance indices and a detailed lipid profile, including derived atherogenic ratios, between adults with EM and healthy controls (HC) in a relatively young cohort free of overt cardiometabolic disease. Methods We conducted a cross-sectional case-control study comparing metabolic profiles in patients with EM and HC. Eligible participants were adults aged 18–50 years at enrolment. Baseline characteristics recorded in both groups included age, sex, weight, height, body mass index (BMI), blood pressure and current smoking status. All participants were of Caucasian ethnicity. Participants in both groups were screened and excluded if any of the following were present: known diabetes mellitus (type 1 or 2), prior diagnosis of IR, dyslipidaemia, arterial hypertension, other significant cardiovascular disease, systemic inflammatory or autoimmune disease, renal or hepatic disease, malignancy, acute infection within 2–4 weeks before enrolment, or pregnancy/breastfeeding. We also excluded participants currently receiving corticosteroid therapy or medications affecting glucose metabolism that included any glucose-lowering agents (such as glucagon-like peptide-1 receptor agonist medication). Participants with postmenopause (≥ 12 months amenorrhea not attributable to pregnancy/lactation/hormonal contraception) and surgical menopause and/or current hormone replacement therapy were excluded to minimize hormonal confounding of metabolic outcomes [21]. Postmenopausal status was assessed by structured history. All participants provided written informed consent, and the study was approved by the local Ethics Committee of the Derer’s University Hospital Bratislava (protocol number 23/2025), and all study procedures were conducted in accordance with the Declaration of Helsinki. Migraine diagnosis and clinical assessment Migraine was diagnosed according to ICHD-3 criteria [2]. Participants with EM were recruited consecutively from the outpatient headache clinic of the 2nd Department of Neurology, Comenius University Bratislava, Faculty of Medicine and Derer’s University Hospital Bratislava, Slovakia, between October 2025 and February 2026. Healthy controls were recruited from hospital staff and community volunteers during the same period. Controls were frequency matched at group level to achieve similar age and sex distribution and had no history of migraine or other primary headache disorders, as well as comorbidities, confirmed by structured interview. All participants underwent a standardised clinical evaluation by a headache specialist, including medical history, medication review, and physical examination. Key migraine clinical variables were recorded, including migraine subtype, migraine without aura (MO) and migraine with aura (MA), frequency measured as monthly migraine days (MMDs) over the last 3 months, attack duration (hours), and headache intensity measured using a Visual Analogue Scale (VAS) [22]. Episodic migraine was further classified as low-frequency episodic migraine (LFEM) with 4–7 MMDs or high-frequency episodic migraine (HFEM) with 8–14 MMDs. Patients with fewer than 4 MMDs were not included; therefore, the cohort does not represent the full spectrum of EM, particularly very-low-frequency EM. Current acute and preventive migraine treatments were recorded and are summarized in Additional file 1: Supplementary Table S1. Body weight and height were measured, and BMI was calculated as kg/m2. Resting blood pressure was measured after 5 min of seated rest. Blood sampling and laboratory variables All participants were instructed to fast for at least 12 h prior to blood sampling. To minimise potential acute-phase effects, blood sampling was performed during the interictal period, defined as the absence of a migraine attack for at least 48 h before blood draw. Participants were instructed to avoid strenuous exercise and alcohol intake 24 h before sampling and to refrain from smoking and caffeine intake on the morning of sampling. Fasting blood samples were analysed for glucose, insulin, and lipid parameters, including total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG) and non-high-density lipoprotein cholesterol (non-HDL-C; calculated by subtracting HDL-C from TC). Glucose and lipid fractions were assessed by spectrophotometry, and insulin was assessed by chemiluminescent magnetic microparticle immunoassay (CMIA) in the central laboratory of Derer’s University Hospital. In addition to individual lipid fractions, we calculated derived lipid indices, including total cholesterol to high-density lipoprotein cholesterol (TC/HDL-C) ratio and the atherogenic index of plasma (AIP). The AIP was calculated as log10(TG/HDL-C) using molar concentrations (mmol/L) [23]. Insulin resistance was assessed using the homeostatic model assessment of insulin resistance (HOMA-IR), calculated from fasting insulin and glucose as HOMA-IR = [insulin (µU/mL) × glucose (mmol/L)] / 22.5 [24]. Statistical analysis Statistical analyses were performed using IBM SPSS Statistics version 29 (IBM Corp., Armonk, NY, USA). Continuous variables are presented as mean ± standard deviation (SD), and categorical variables as counts and percentages. Between-group comparisons were performed using two-tailed independent-samples t tests for continuous variables and chi-square tests for categorical variables. Because controls were matched to the migraine group at the group level (frequency matching) rather than by individual 1:1 matching, standard unpaired statistical methods were used throughout. Levene’s test was used to assess equality of variances; when this assumption was violated, the Welch correction was applied. To evaluate adjusted differences in metabolic parameters between the migraine and control groups, multivariable linear regression analyses were performed with each metabolic marker (HOMA-IR, TC/HDL-C ratio, AIP, TC, LDL-C, HDL-C, TG, non-HDL-C, fasting glucose, and fasting insulin) entered as the dependent variable and group status as the main independent variable. All models were adjusted for age, sex, BMI, and smoking status, which were selected a priori because of their established associations with insulin sensitivity, lipid metabolism, and overall cardiometabolic risk, and because they could confound the relationship between migraine status and metabolic outcomes. Results are reported as unstandardised regression coefficients (B) with 95% confidence interval (CI) and adjusted R² values. Sensitivity analyses were performed for the three co-primary endpoints by additionally including systolic blood pressure as a covariate in the multivariable regression models. Migraine medication exposure was summarized descriptively and was not included in adjusted models or sensitivity analyses because individual preventive medication classes were sparsely represented. Within the migraine group, Spearman rank correlation coefficients (ρ) were calculated to assess associations between co-primary metabolic markers (HOMA-IR, TC/HDL-C ratio, and AIP) and migraine clinical characteristics, including MMDs and attack intensity measured by VAS. Exploratory comparisons of metabolic parameters according to migraine aura status, MO and MA, and migraine attack frequency, LFEM versus HFEM, were performed using two-tailed independent-samples t tests. Assumptions of parametric tests were evaluated prior to analysis. Residual plots from regression models were visually inspected to assess linearity, homoscedasticity, and approximate normality of residuals. Multicollinearity among predictors was evaluated using variance inflation factors (VIF). All p values were two-sided. The three co-primary endpoints, HOMA-IR, TC/HDL-C ratio, and AIP, were prespecified a priori. Primary analyses are reported with unadjusted p values; however, the lipid-related co-primary endpoints, TC/HDL-C ratio and AIP, remained statistically significant even under a conservative Bonferroni threshold for three co-primary endpoints, corresponding to p < 0.0167. Secondary metabolic endpoints were considered exploratory and interpreted descriptively. For exploratory aura-status subgroup analyses, Benjamini–Hochberg false discovery rate correction was applied across the eight aura-related comparisons. With 45 participants per group, the study had approximately 80% power to detect a medium-to-large effect size, Cohen’s d ≥ 0.60, at a two-sided alpha of 0.05 for the primary between-group comparisons. Results Baseline demographic, clinical, and metabolic characteristics The study included 45 participants with EM and 45 HC. The groups were comparable in age, sex distribution, BMI, smoking status, and blood pressure (all p > 0.05). In unadjusted analyses, the migraine group showed a less favourable metabolic profile, with higher fasting glucose (p = 0.005), total cholesterol (p < 0.001), LDL-C (p < 0.001), triglycerides (p = 0.029), non-HDL cholesterol (p < 0.001), TC/HDL-C ratio (p < 0.001), and AIP (p = 0.005), and lower HDL-C (p < 0.001) than the control group. Fasting insulin and HOMA-IR were numerically higher in the migraine group, but these differences were not statistically significant. Baseline characteristics are summarized in Table 1. Table 1Baseline demographic, clinical, and metabolic characteristics of study participants according to disease statusVariableControls (n = 45)Migraine (n = 45)p valueAge, years35.47 ± 8.7037.82 ± 8.510.198Female sex, n (%)42 (93.3)42 (93.3)1.000BMI, kg/m²23.49 ± 3.4724.11 ± 3.440.402Current smoking, n (%)6 (13.3)7 (15.6)0.764Systolic blood pressure, mmHg116.58 ± 8.10114.16 ± 12.320.274Diastolic blood pressure, mmHg74.36 ± 7.7275.76 ± 10.170.464Fasting glucose, mmol/L4.66 ± 0.414.90 ± 0.36 0.005 Fasting insulin, mIU/L5.68 ± 1.746.31 ± 2.580.179HOMA-IR1.19 ± 0.431.37 ± 0.560.098Total cholesterol, mmol/L4.41 ± 0.665.05 ± 0.89 < 0.001 HDL-C, mmol/L1.55 ± 0.261.33 ± 0.28 < 0.001 LDL-C, mmol/L2.49 ± 0.783.23 ± 0.71 < 0.001 Triglycerides, mmol/L0.81 ± 0.260.98 ± 0.45 0.029 Non-HDL-C, mmol/L2.86 ± 0.773.72 ± 0.89 < 0.001 TC/HDL-C ratio2.93 ± 0.703.95 ± 1.04 < 0.001 AIP−0.30 ± 0.19−0.17 ± 0.25 0.005 Values are presented as mean ± Standard Deviation (SD) for continuous variables and number (percentage) for categorical variables. Between-group comparisons were performed using (two-tailed) independent-samples t tests for continuous variables and chi-square tests for categorical variables. For variables with unequal variances according to Levene’s test, the Welch correction was applied. Statistical significance was defined as p < 0.05Bold is used to highlight statistical significance Baseline demographic, clinical, and metabolic characteristics of study participants according to disease status Values are presented as mean ± Standard Deviation (SD) for continuous variables and number (percentage) for categorical variables. Between-group comparisons were performed using (two-tailed) independent-samples t tests for continuous variables and chi-square tests for categorical variables. For variables with unequal variances according to Levene’s test, the Welch correction was applied. Statistical significance was defined as p < 0.05 Bold is used to highlight statistical significance Adjusted differences in co-primary endpoints between the migraine and control groups Multivariable linear regression models were used to assess adjusted differences in the co-primary metabolic endpoints between the migraine and control groups after adjustment for age, sex, BMI, and smoking status. For HOMA-IR, the overall model was statistically significant (F(5,84) = 3.14, p = 0.012) and explained 10.7% of the variance (adjusted R²=0.107). Among the covariates, BMI showed a significant positive association with HOMA-IR (B = 0.052, 95% CI 0.020–0.083, p = 0.002), whereas age, sex, and smoking status were not significant predictors. HOMA-IR did not differ significantly between groups after adjustment (B = 0.167, 95% CI − 0.034 to 0.368, p = 0.102). By contrast, the migraine group had a significantly higher TC/HDL-C ratio than the control group in the adjusted model (B = 0.956, 95% CI 0.604 to 1.308, p < 0.001), with the model explaining 33.6% of the variance (F(5,84) = 10.00, p < 0.001; adjusted R²=0.336). Sex was also a significant predictor of TC/HDL-C ratio (B = 0.919, p = 0.012). Atherogenic index of plasma was likewise higher in the migraine group after adjustment (B = 0.122, 95% CI 0.033 to 0.211, p = 0.008), with the model explaining 14.6% of the variance (F(5,84) = 4.04, p = 0.002; adjusted R²=0.146). No covariate showed an independent association with AIP. Overall, compared with controls, participants with EM had higher adjusted values of the co-primary lipid-related endpoints TC/HDL-C ratio and AIP, whereas between-group difference for HOMA-IR was not statistically significant. Sensitivity analyses additionally adjusting for systolic blood pressure produced materially similar results to the primary models. Migraine status remained significantly associated with TC/HDL-C ratio (B = 1.035, p < 0.001) and AIP (B = 0.146, p = 0.001), whereas the association with HOMA-IR remained non-significant (B = 0.180, p = 0.086). Exploratory adjusted differences in secondary lipid and glucose-metabolism endpoints between the migraine and control groups Exploratory additional multivariable models examined adjusted differences in secondary lipid and glucose-metabolism endpoints between the migraine and control groups. Among secondary lipid measures, total cholesterol was higher in the migraine group than in the control group (B = 0.568, 95% CI 0.255 to 0.882, p < 0.001; F(5,84) = 7.89, p < 0.001; adjusted R²=0.236). Age was also significantly associated with total cholesterol (B = 0.031, p = 0.001), whereas sex, BMI, and smoking status were not significant predictors. LDL-C was higher in the migraine group than in HC (B = 0.682, 95% CI 0.374 to 0.989, p < 0.001; F(5,84) = 6.52, p < 0.001; adjusted R²=0.237), whereas HDL-C was lower (B = − 0.220, 95% CI − 0.328 to − 0.112, p < 0.001; F(5,84) = 6.47, p < 0.001; adjusted R²=0.235). Sex was also significantly associated with HDL-C (B = − 0.287, p = 0.011). Non-HDL cholesterol was higher in the migraine group (B = 0.786, 95% CI 0.452 to 1.120, p < 0.001; F(5,84) = 8.23, p < 0.001; adjusted R²=0.289), and age was positively associated with this outcome (B = 0.024, p = 0.019). Triglycerides showed a small positive association with group status (B = 0.154, 95% CI 0.001 to 0.307, p = 0.048), suggesting slightly higher levels in the migraine group. Age, sex, BMI, and smoking were not significant predictors. However, because the overall triglyceride model was not statistically significant (F(5,84) = 2.21, p = 0.060) and explained only a small proportion of the variance (adjusted R²=0.064), this finding should be interpreted cautiously. Among secondary glucose-metabolism measures, fasting glucose was higher in the migraine group after adjustment (B = 0.210, 95% CI 0.047 to 0.372, p = 0.012; F(5,84) = 2.95, p = 0.017; adjusted R²=0.099). By contrast, fasting insulin did not differ significantly between groups (B = 0.632, 95% CI − 0.255 to 1.520, p = 0.160), although the overall model was statistically significant (F(5,84) = 3.01, p = 0.015; adjusted R²=0.101). In that model, BMI was positively associated with fasting insulin (B = 0.220, p = 0.002), whereas age showed a modest inverse association (B = − 0.063, p = 0.024). A summary of all adjusted multivariable linear regression analyses is presented in Table 2. Overall, the adjusted analyses showed a less favourable lipid profile in the migraine group, whereas insulin-resistance markers were not significantly different between groups after adjustment for covariates. Table 2Adjusted differences in metabolic markers between the migraine and control groupsOutcome variableB (migraine vs. control)95% CIp valueAdjusted R²HOMA-IR0.167−0.034 to 0.3680.1020.107TC/HDL-C ratio0.9560.604 to 1.308 < 0.001 0.336AIP0.1220.033 to 0.211 0.008 0.146Total cholesterol mmol/L0.5680.255 to 0.882 < 0.001 0.236LDL-C, mmol/L0.6820.374 to 0.989 < 0.001 0.237HDL-C, mmol/L−0.220−0.328 to − 0.112 < 0.001 0.235Triglycerides, mmol/L0.1540.001 to 0.307 0.048 0.064Non-HDL-C, mmol/L0.7860.452 to 1.120 < 0.001 0.289Fasting glucose, mmol/L0.2100.047 to 0.372 0.012 0.099Fasting insulin, mIU/L0.632−0.255 to 1.5200.1600.101Unstandardised regression coefficients (B) represent the difference between migraine patients and healthy controls. All models were adjusted for age, sex, BMI, and smoking status. Adjusted R² values represent the proportion of variance in the outcome explained by each modelBold is used to highlight statistical significance Adjusted differences in metabolic markers between the migraine and control groups Unstandardised regression coefficients (B) represent the difference between migraine patients and healthy controls. All models were adjusted for age, sex, BMI, and smoking status. Adjusted R² values represent the proportion of variance in the outcome explained by each model Bold is used to highlight statistical significance Correlations of co-primary metabolic markers with migraine frequency and intensity Within the migraine group, Spearman correlation analyses were performed to examine associations between the co-primary metabolic markers and MMDs and attack intensity measured by the VAS. No significant correlations were observed between MMDs and HOMA-IR, TC/HDL-C ratio, or AIP (all p > 0.55). Similarly, attack intensity was not associated with HOMA-IR (ρ=−0.183, p = 0.230), TC/HDL-C ratio (ρ = 0.080, p = 0.603), or AIP (ρ=−0.064, p = 0.677). These results indicate that within the migraine group, variation in the co-primary metabolic markers was not significantly related to migraine frequency or attack intensity. Results are summarized in Table 3. Table 3Associations of migraine severity with co-primary metabolic endpointsClinical variableHOMA-IR ρ (p)TC/HDL-C ratio ρ (p)AIP ρ (p)Monthly migraine days (MMDs)−0.073 (0.633)0.091 (0.554)0.041 (0.789)Attack intensity (VAS)−0.183 (0.230)0.080 (0.603)−0.064 (0.677)Spearman rank correlation coefficients (ρ) are presented with p values in parentheses. Reported p values are unadjusted. Statistical significance was defined as p < 0.05 Associations of migraine severity with co-primary metabolic endpoints Spearman rank correlation coefficients (ρ) are presented with p values in parentheses. Reported p values are unadjusted. Statistical significance was defined as p < 0.05 Exploratory subgroup analyses within the migraine group Metabolic profile according to migraine aura status In exploratory subgroup analyses we evaluated 15 participants with MA and 30 participants with MO. Participants with MA showed nominally less favourable lipid-related measures than those with MO. Specifically, TC/HDL-C ratio (4.46 ± 0.99 vs. 3.69 ± 0.98, p = 0.017), AIP (− 0.06 ± 0.23 vs. − 0.22 ± 0.24, p = 0.033), and LDL-C levels (3.57 ± 0.70 vs. 3.06 ± 0.66 mmol/L, p = 0.022) were higher in the MA group, whereas HDL-C levels were lower (1.21 ± 0.24 vs. 1.40 ± 0.28 mmol/L, p = 0.029). Total cholesterol levels were numerically higher in the MA group (5.22 ± 0.89 vs. 4.97 ± 0.90 mmol/L), although this difference did not reach statistical significance (p = 0.385). HOMA-IR did not differ significantly between migraine subtypes (p = 0.216). Triglycerides and non-HDL cholesterol were numerically higher in the MA subgroup; these differences did not reach statistical significance. Results are presented in Table 4. However, after Benjamini–Hochberg false discovery rate correction across the eight aura-related comparisons, none of these differences remained statistically significant (all q = 0.066), although the direction of the nominal differences was consistently towards a less favourable lipid profile in the MA group. Table 4Metabolic profile of episodic migraine according to aura statusVariableMO (n = 30)MA (n = 15)p valueHOMA-IR1.29 ± 0.471.51 ± 0.690.216TC/HDL-C ratio3.69 ± 0.984.46 ± 0.99 0.017 AIP−0.22 ± 0.24−0.06 ± 0.23 0.033 Total cholesterol, mmol/L4.97 ± 0.905.22 ± 0.890.385LDL-C, mmol/L3.06 ± 0.663.57 ± 0.70 0.022 HDL-C, mmol/L1.40 ± 0.281.21 ± 0.24 0.029 Triglycerides, mmol/L0.90 ± 0.411.14 ± 0.490.093Non-HDL-C, mmol/L3.57 ± 0.884.02 ± 0.860.118Values are presented as mean ± Standard Deviation (SD). Comparisons between migraine without aura (MO) and migraine with aura (MA) were performed using two-tailed independent-samples t tests. Reported p values are unadjusted. Benjamini–Hochberg false discovery rate correction was applied across the eight aura-related comparisons; after correction, no comparison remained statistically significant (TC/HDL-C ratio, AIP, LDL-C, and HDL-C: all q = 0.066)Bold is used to highlight statistical significance Metabolic profile of episodic migraine according to aura status Values are presented as mean ± Standard Deviation (SD). Comparisons between migraine without aura (MO) and migraine with aura (MA) were performed using two-tailed independent-samples t tests. Reported p values are unadjusted. Benjamini–Hochberg false discovery rate correction was applied across the eight aura-related comparisons; after correction, no comparison remained statistically significant (TC/HDL-C ratio, AIP, LDL-C, and HDL-C: all q = 0.066) Bold is used to highlight statistical significance Metabolic profile according to migraine attack frequency Metabolic parameters were compared between patients with LFEM and HFEM. No significant differences were observed between the groups for HOMA-IR, TC/HDL-C ratio, AIP, total cholesterol, LDL-C, HDL-C, triglycerides, or non-HDL cholesterol (Table 5). Although TC/HDL-C ratio and LDL-C levels were numerically higher in the HFEM group, these differences did not reach statistical significance. Table 5Metabolic profile of episodic migraine according to attack frequencyVariableLFEM (n = 29)HFEM (n = 16)p valueHOMA-IR1.45 ± 0.591.22 ± 0.470.189TC/HDL-C ratio3.86 ± 0.994.11 ± 1.150.450AIP−0.18 ± 0.25−0.14 ± 0.240.622Total cholesterol, mmol/L5.08 ± 0.965.01 ± 0.790.791LDL-C, mmol/L3.19 ± 0.673.29 ± 0.790.644HDL-C, mmol/L1.36 ± 0.291.28 ± 0.270.328Triglycerides, mmol/L0.98 ± 0.460.98 ± 0.440.982Non-HDL-C, mmol/L3.72 ± 0.923.73 ± 0.860.967Comparisons between low-frequency episodic migraine (LFEM, 4–7 monthly migraine days) and high-frequency episodic migraine (HFEM, 8–14 monthly migraine days) were performed using two-tailed independent-samples t tests. Values are presented as mean ± Standard Deviation (SD). Reported p values are unadjusted Metabolic profile of episodic migraine according to attack frequency Comparisons between low-frequency episodic migraine (LFEM, 4–7 monthly migraine days) and high-frequency episodic migraine (HFEM, 8–14 monthly migraine days) were performed using two-tailed independent-samples t tests. Values are presented as mean ± Standard Deviation (SD). Reported p values are unadjusted Discussion In this cross-sectional case–control study, EM was associated with a more atherogenic lipid profile than that observed in HC. The most consistent findings were lipid-related. Compared with controls, the migraine group showed higher TC/HDL-C ratio, AIP, total cholesterol, LDL-C, and non-HDL-C together with lower HDL-C, and these associations persisted after adjustment for age, sex, BMI, and smoking. By contrast, the associations with insulin-resistance markers were weaker. Although fasting glucose was modestly higher in the migraine group, neither HOMA-IR nor fasting insulin was associated with migraine after adjustment for selected covariates. Within the migraine group, metabolic markers were not related to MMDs or attack intensity. Exploratory subgroup analyses showed nominally less favourable lipid-related measures in MA than in MO, although these findings did not remain significant after false discovery rate correction. Taken together, these findings support an association between EM and a more atherogenic lipid profile after adjustment for selected cardiometabolic covariates, while providing less consistent evidence for a direct relationship between EM and IR in this relatively young cohort [14, 19, 20, 25, 26]. These findings are relevant in the broader context of migraine as a disorder increasingly linked not only to neuronal hyperexcitability and trigeminovascular activation, but also to systemic vascular and metabolic alterations [7, 8, 11, 27]. Large epidemiological studies have shown that migraine, particularly MA, is associated with increased risk of ischemic stroke and, in several cohorts, also myocardial infarction, haemorrhagic stroke, venous thromboembolism, and atrial fibrillation or flutter [28, 29]. Consistently, the Women’s Health Study reported that MA was associated with a higher incidence of major cardiovascular disease, with an absolute risk increase comparable to that of several traditional vascular risk factors [30]. Even in our relatively young cohort without overt cardiometabolic disease, migraine remained associated with an atherogenic lipid pattern. Together, these findings support the relevance of considering cardiometabolic features when studying EM, while not implying a causal relationship between migraine and the observed lipid pattern [28–31]. From that perspective, the rationale for evaluating both insulin-resistance indices and lipid-derived markers becomes clearer. We considered these parameters not simply as accompanying comorbid traits, but as plausible markers of pathways increasingly implicated in migraine pathophysiology, including endothelial dysfunction, oxidative stress, chronic low-grade inflammation, altered nitric oxide signalling, impaired cerebrovascular reactivity, and impaired energy metabolism [8, 10, 11, 27]. We selected HOMA-IR as the primary fasting index of IR because it is widely used and provides a practical, clinically interpretable estimate of basal glucose-insulin homeostasis [24]. In parallel, we assessed not only conventional lipid fractions but also derived indices, TC/HDL-C ratio and AIP, because they are thought to provide a more integrated reflection of the balance between atherogenic and protective lipoprotein components than single fractions alone, especially in relatively young adults without overt dyslipidaemia [23, 32–35]. One of the most consistent findings of our study was that lipid ratios distinguished migraine from controls more clearly than individual lipid fractions. The TC/HDL-C ratio showed the strongest and most robust association, and AIP also remained higher in the migraine group after multivariable adjustment. This pattern is potentially informative, as TC/HDL-C ratio reflects the balance between cholesterol carried in potentially atherogenic particles and HDL-mediated protective transport [33], whereas AIP integrates the triglyceride-HDL axis and has been linked to smaller, denser LDL particles and broader cardiometabolic risk [23, 32, 34, 35]. In our cohort, triglycerides alone showed only a modest association, and the adjusted TG model had limited explanatory power, suggesting that TG were not the dominant abnormality. Instead, the more consistent finding was a broader atherogenic pattern characterised by higher LDL-C and non-HDL-C, lower HDL-C, and unfavourable lipid-ratio indices. Our lipid findings align with prior observational evidence, while also reflecting the heterogeneity reported across studies [19, 25, 26]. In a meta-analysis, Liampas et al. reported modestly higher LDL-C, TC, and TG in adults with migraine compared with HC, although some associations were attenuated after accounting for BMI [19]. Population-based studies have also linked migraine to a less favourable lipid profile, including higher TC/HDL-C ratio in MA in the GEM study and a more atherogenic lipoprotein pattern in the ELSA-Brasil cohort [25, 26]. Our findings add to this literature by suggesting that, in EM, lipid alterations are better characterised as an overall atherogenic pattern than as an isolated abnormality of a single fraction. Findings related to insulin resistance were more nuanced. Although HOMA-IR was modestly higher in participants with EM in unadjusted analyses, this association attenuated after adjustment for age, BMI, sex, and smoking. Fasting insulin showed a similar pattern, whereas fasting glucose remained modestly higher in the migraine group after adjustment. This suggests that, in a relatively young cohort with EM and low cardiometabolic burden, subtle abnormalities of glucose regulation may be present even without clear evidence of fasting insulin resistance. This interpretation is broadly consistent with previous literature [14, 20, 36–39]. Sacco et al. reported higher fasting glucose across studied migraine subgroups, with no difference in insulin-resistance indices, including HOMA-IR [14]. In contrast, stronger associations with IR have been reported in cohorts enriched for more severe migraine phenotypes or greater metabolic burden [20, 36]. Fava et al. reported a higher prevalence of IR in chronic migraine than in EM and HC, supporting the possibility that insulin resistance-related mechanisms may be more evident in higher-frequency phenotypes [36]. Similarly, Ali et al. studied newly diagnosed migraine patients, predominantly MO, and found higher fasting insulin and HOMA-IR than in controls, together with greater migraine severity and migraine-related impact among those with IR [20]. Recent studies using alternative insulin resistance surrogates such as the triglyceride-glucose index have also yielded mixed and population-dependent results [37–39]. Collectively, these findings support the view that the relationship between migraine and insulin resistance is unlikely to be uniform and may depend on migraine frequency, episodic versus chronic status, the cohort’s metabolic profile, and the marker used to assess IR [14, 20, 36–39]. The observed difference between lipid-related markers and fasting insulin-resistance indices should be interpreted cautiously. In this cohort, lipid-related indices showed more consistent between-group differences than HOMA-IR or fasting insulin. However, the absence of an independent association with HOMA-IR should not be interpreted as evidence that insulin resistance is unrelated to EM. Previous studies have reported mixed findings regarding insulin resistance in migraine, with stronger associations observed in some cohorts with chronic migraine, higher migraine burden, or greater metabolic risk [14, 20, 36–39]. Moreover, HOMA-IR is a fasting surrogate marker of basal glucose-insulin homeostasis rather than a dynamic measure of insulin sensitivity [24]. The present study was powered to detect medium-to-large effects, included a relatively young and metabolically healthy population, and relied on fasting surrogate markers rather than dynamic metabolic testing. Therefore, a small or context-dependent association between EM and insulin resistance cannot be excluded. Another notable observation was the lack of association between metabolic abnormalities and MMDs or attack intensity, together with the absence of significant metabolic differences across migraine-frequency subgroups. This suggests that the metabolic changes observed in our cohort were not simply markers of current migraine burden. Because samples were collected during the interictal period, they are less likely to reflect only short-term changes related to an individual migraine attack. In this context, the observed lipid alterations may be consistent with a broader biological profile linked to migraine, rather than only to short-term attack burden [7, 8, 10, 11]. Participants with EM with aura showed higher TC/HDL-C ratio, AIP, LDL-C, and lower HDL-C than those without aura, whereas HOMA-IR did not differ significantly between subgroups. However, these differences did not remain statistically significant after false discovery rate correction and should be interpreted with caution. Although these analyses were limited by small subgroup size, they are broadly in line with prior literature suggesting a less favourable vascular and lipid profile in migraine with aura, as well as with evidence that this phenotype is more strongly linked to ischemic stroke and major cardiovascular events [25, 28–30, 40]. Our findings therefore raise the possibility that a more atherogenic lipid profile may be more common in the aura phenotype, but this observation should be considered hypothesis-generating and requires confirmation in larger studies. Although our cohort included patients with EM rather than medication-overuse headache, our findings may also be considered in the broader context of emerging biomarker research in headache disorders. In an exploratory serum proteomic study, Pellesi et al. suggested that medication-overuse headache may share serum biochemical markers with chronic pain conditions, supporting the relevance of systemic biochemical profiling in selected headache phenotypes [41]. However, because medication-overuse headache differs from EM in chronicity, medication exposure, and clinical context, these findings should be considered complementary rather than directly comparable to our lipid-focused results. The present findings should be interpreted as a step toward better metabolic characterization of migraine, rather than as evidence of a causal relationship. Future longitudinal studies should assess the stability of lipid-related indices over time, their potential value in identifying patients with EM at higher long-term vascular risk, and their variation across migraine subgroups, including very-low-frequency EM, high-frequency EM, chronic migraine, migraine with aura, and medication-overuse headache. Larger studies with repeated interictal measurements, more detailed assessment of insulin sensitivity, lipoprotein profiling, lifestyle factors, and medication exposure may help clarify whether the observed lipid pattern is directly related to migraine, reflects shared risk factors, or is simply an associated finding. Strengths and limitations To our knowledge, this is one of the few studies to investigate both atherogenic lipid patterns, including derived lipid indices beyond standard lipid fractions, and insulin resistance in EM, using a strictly standardised interictal fasting design. Several limitations should be acknowledged. The cross-sectional case-control design cannot determine whether the observed metabolic differences precede migraine, result from migraine-related behavioural or biological factors, or represent an epiphenomenon related to shared risk factors. The sample size was modest, comprising 45 participants with EM and 45 HC, and the study was powered primarily to detect medium-to-large between-group effects. Therefore, smaller associations, particularly for HOMA-IR and fasting insulin, may have been missed. Insulin resistance was assessed using fasting surrogate markers rather than dynamic metabolic testing, and direct lipoprotein particle analysis was not available. Accordingly, the absence of an adjusted association with HOMA-IR should not be interpreted as evidence that insulin resistance is unrelated to EM. Our cohort was highly selected, participants were young, predominantly female, Caucasian and without overt cardiometabolic comorbidities such as hypertension, known dyslipidaemia, diabetes or insulin resistance; postmenopausal women were also excluded. This strengthened internal validity but limits generalizability to broader migraine populations, including men, older adults, patients with cardiometabolic comorbidities, individuals with chronic migraine or very-low-frequency EM. Participants with fewer than 4 MMDs were not included, and therefore the findings may not apply to the full spectrum of EM. Although we adjusted for age, sex, BMI, and smoking, residual confounding cannot be excluded. Detailed data on waist circumference, diet, physical activity, socioeconomic status, sleep, and stress were not collected and therefore could not be included in the adjusted analyses. Medication exposure was recorded, but medication-specific adjusted analyses were not performed because of the limited sample size and small numbers within individual treatment classes. Finally, analyses according to aura status and migraine frequency were exploratory and underpowered; therefore, these findings should be considered hypothesis-generating and require confirmation in larger cohorts designed to compare HC, MO, and MA groups directly. Future longitudinal studies should determine whether these metabolic patterns identify migraine subgroups at higher long-term vascular risk, track with disease progression, or are modifiable through targeted metabolic interventions [8, 42, 43]. Conclusion In this cross-sectional case-control study conducted under standardised interictal fasting conditions, EM was associated with a more atherogenic lipid profile, characterised by higher TC/HDL-C ratio, AIP, LDL-C, and non-HDL-C, and lower HDL-C. By contrast, fasting surrogate markers of IR were not associated with EM after adjustment for selected covariates, although small effects cannot be excluded. These findings suggest that lipid-related indices, particularly TC/HDL-C ratio and AIP, may capture cardiometabolic differences associated with EM, but larger longitudinal studies are needed to confirm their relevance and clinical implications. Supplementary Information Additional file 1: Table S1. Current acute and preventive migraine treatment exposure in the episodic migraine group. Acknowledgements We would like to thank all the participants for their contribution to the study. We would like to thank the civic association Neurologický spolok Kramáre for covering the article publication fee. Authors’ contributions SS: study conceptualization, study design, participant recruitment, data collection, data interpretation, manuscript writing; OD: study conceptualization, study design, participant recruitment, data collection, data interpretation, writing-review; BÖ: data analyses, data interpretation; MG: data collection; IS: writing-review, editing; PV: writing-review, editing, supervision. All authors critically reviewed the manuscript for important intellectual content and approved the final version for submission. Funding This research was funded by the Comenius University Bratislava, grant number UK/1217/2025. Data availability The summary data supporting the findings of this study are included within the article and its additional file. The full datasets used and analysed during the current study are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate This study was approved by the local Ethics Committee of the Derer’s University Hospital Bratislava (protocol number 23/2025), and all study procedures were conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants before enrolment. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Simona Szabóová, Email: simonah09@gmail.com. Peter Valkovič, Email: peter.valkovic@gmail.com. Reference list Associated Data Data Availability Statement The summary data supporting the findings of this study are included within the article and its additional file. The full datasets used and analysed during the current study are available from the corresponding author upon reasonable request.