Association between liver enzyme levels and prevalence of migraine: the atherosclerosis risk in communities study

Abstract

Background:

Cumulative research data indicate that migraine is characterized by a glutamatergic imbalance, particularly an excessive glutamatergic signal. Increases in glutamate levels in the brain and plasma of migraine patients have been reported, but less is known about the association between liver enzymes, such as aspartate aminotransferase (AST), alanine aminotransferase (ALT), and gamma-glutamyl transpeptidase (GGT) that regulate blood glutamate levels and migraine.

Objectives:

We evaluated associations between AST, ALT, and GGT levels across the quartiles and a history of probable/defined migraine in the Atherosclerosis Risk in Communities Study cohort.

Design:

We included 11,718 participants with measured liver enzyme levels and self-reported data on migraine with and without aura. Multiple logistic regression models were used to assess associations of sex-specific quartiles of liver enzymes with probable/definite migraine.

Results:

A total of 1626 probable/definite migraine events were reported in 1993–1995. After adjustment for age, race-center, and sex, higher levels of AST, ALT, and GGT were associated with a lower prevalence of migraine. The adjusted odds ratios (95% CIs) for migraine for Q1 versus Q4 levels of AST, ALT, and GGT were 1.24 (1.06–1.45), 1.17 (1.00–1.37) and 1.21 (1.03–1.41), respectively. Analysis by yes/no aura showed higher odds of migraine without aura for lower (Q1) compared with higher (Q4) levels of ALT (adjusted OR 1.38, 95% CI 1.05–1.82), while no significant association was observed between enzyme levels and prevalence of migraine with aura.

Conclusion:

Our findings suggest that higher AST, ALT, and GGT levels are associated with a lower prevalence of migraine. Although the exact mechanisms linking lower blood levels of AST, ALT, and GGT to migraines remain unclear, their association may be explained by inefficient plasma glutamate regulation, which could play a role in migraine pathology. This finding is important for patients as it sheds light on potential metabolic contributions to migraines, supporting the hypothesis that factors beyond traditional neurovascular theories are involved.

Introduction

Migraine affects over 1 billion people worldwide and is one of the most difficult headache disorders to treat. ¹ The condition affects women approximately three times more often than men, likely due to a combination of biological and psychosocial factors such as hormonal fluctuations, differences in the stress response, and genetic predisposition; however, its primary pathophysiological mechanisms are still not fully understood. ² Migraine was traditionally attributed to vascular dysfunction, but this view has since evolved to reflect a more complex pathophysiology involving neuronal hyperexcitability and excessive glutamatergic activity. These may contribute to cortical spreading depolarization/depression (CSD), a process considered to be the electrophysiological basis of migraine aura.^3–5 Moreover, glutamate is known to play a crucial role in nociception and central sensitization, which, along with CSD, are important processes in migraine biology. ⁶

Given that glutamate contributes to CSD, elevated cortical glutamate levels would be expected in individuals with migraine. However, most previous studies employing Proton Magnetic Resonance Spectroscopy (¹H-MRS) at 1T or 3T yielded conflicting results, largely due to the technical limitations in reliably distinguishing glutamate from glutamine using standard MRS protocols. ⁷ Nevertheless, a meta-analysis by Peek et al. ⁸ reported increased brain levels of glutamate, glutamine, their combined measure (Glx), and gamma-aminobutyric acid (GABA) during the interictal period in individuals with migraine; findings not observed in other pain conditions. Further stratification by migraine subtype revealed that glutamate elevations were more consistently observed in individuals with migraine without aura,^6,9 while results regarding GABA levels were variable and inconclusive. ¹⁰ More recent advances using high-field 7T ¹H-MRS have now allowed for the separate quantification of cerebral glutamate and glutamine. With this removal of previous technical difficulties, elevated cortical glutamate has been confirmed in individuals with migraine without aura, reinforcing the hypothesis of heightened cerebral excitability in migraine and raising important questions about the distinct neuro-molecular mechanisms underlying aura phenomena. ⁹

Elevated glutamate levels have been detected in the cerebrospinal fluid (CSF) and plasma of migraine patients, frequently alongside increased concentrations of calcitonin gene-related peptide (CGRP), in both chronic and interictal episodic migraine cases. These findings provide further evidence for systemic glutamatergic dysregulation in migraine pathophysiology.^11–13 Moreover, previous studies have demonstrated that heightened glutamatergic neurotransmission can stimulate the release of both CGRP and substance P (SP), two neuropeptides critically involved in nociceptive signaling and neurogenic inflammation associated with migraine.^14,15 In addition, meta-analytic evidence supports persistent interictal elevation of plasma glutamate in migraine sufferers. ¹² A positive correlation between the time elapsed since migraine attack onset and elevated plasma glutamate levels during the ictal phase suggests a dynamic relationship between glutamate accumulation and migraine progression. ¹⁶ Elevated salivary glutamate levels, particularly in chronic migraine, offer an additional, noninvasive biomarker supporting this pattern. ¹⁷

In light of glutamate’s role in migraine, several studies have explored therapeutic strategies targeting its systemic regulation.^18,19 Both repetitive transcranial magnetic stimulation (rTMS) and amitriptyline (AMT), a tricyclic antidepressant commonly used to treat major depressive disorder and various pain syndromes, have been shown to reduce plasma glutamate concentrations and the relative expression of the NR2B subunit of the NMDA receptor.^18,19 These effects were accompanied by significant improvements in headache severity and associated symptoms, suggesting a modulatory impact on glutamatergic signaling in migraine patients, including possible downregulation of glutamate receptors. ¹⁸ Prophylactic therapy with propranolol, a beta-blocker commonly used to treat hypertension, has similarly been associated with reduced migraine frequency and lower plasma glutamate levels in patients without aura. ²⁰ Neuroimaging studies further support the role of glutamate in mediating therapeutic responses to rTMS, showing activation of neural networks involved in pain signal integration, particularly the thalamus and brainstem, regions known to be rich in glutamate receptors. Nevertheless, the mechanism that makes most antimigraine prophylactic drugs effective is not completely understood.

Although the central and peripheral glutamatergic systems were historically considered distinct, recent evidence supports their interconnection via glutamate transport across the blood–brain barrier (BBB) and their mutual influence as an important factor in brain pathophysiology.^21–23 Excitatory amino-acid transporters 1–3 (EAAT1–3) previously believed to be localized to neurons and astrocytes, are also expressed on brain capillary endothelial cells and are main glutamate reuptake transporters actively participating in glutamate clearance.^24,25 Passive transporters have also been identified on the blood-facing side of brain capillaries, facilitating glutamate efflux from the brain to the blood in response to concentration gradients.^22,26,27 Studies in various animal models support the concept of glutamate efflux across the BBB as a protective response to elevated brain glutamate levels.^28–31 Our previous work demonstrated that patients with aneurysmal subarachnoid hemorrhage exhibit elevated glutamate levels in both the plasma and CSF, with plasma glutamate showing a strong correlation with neurological outcomes at 3 months postinjury. ³² Thus, collectively, these findings underscore the importance of glutamate homeostasis in CNS pathology, including migraine.

Blood glutamate is mainly regulated by blood-resident enzymes such as glutamate-oxaloacetate transaminase (GOT1 or AST) and glutamate-pyruvate transaminase (GPT1 or ALT), and their levels are slightly different between men and women. ³³ In migraine patients, low blood GOT1 levels have been shown to correlate with high blood glutamate concentration. ¹⁶ GOT1 and GPT1 transform glutamate into 2-ketoglutarate and aspartate or alanine, respectively.^21,27,34,35 While gamma-glutamyl transpeptidase (GGT) is a cell-surface enzyme also involved in glutamate synthesis and regulation. ³⁶ Thus, the role of these liver enzymes in glutamate metabolism suggests a potential role in migraine attacks. Indeed, in patients with migraine, decreased levels of GOT1 and GPT1 have been inversely correlated with elevated plasma glutamate concentrations.^16,34,35 While our previous work and others have shown that exogenous administration of GOT1 reduces blood glutamate levels in both animal models and clinical contexts.^28,29,31,36

Despite accumulating evidence linking glutamate dysregulation to migraine, large-scale population-based studies evaluating the relationship between systemic glutamate-regulatory enzymes and migraine prevalence are lacking. The current study addresses this gap by investigating associations between plasma levels of GOT1 (AST), GPT1 (ALT), and GGT and the prevalence of migraine (with and without aura) using data from the well-characterized Atherosclerosis Risk in Communities (ARIC) cohort. This study is the first, to our knowledge, to examine liver enzyme levels as surrogate markers of glutamate metabolism in relation to migraine subtypes in a large epidemiological sample. By leveraging a systemic perspective on glutamate regulation, our findings may provide new insights into migraine pathophysiology and help identify accessible, blood-based biomarkers for disease stratification and therapeutic monitoring.

Research design and methods

Research population

The ARIC study involved a community-based cohort of 15,792 individuals aged 45–64 at baseline (1987–1989). Participants were recruited from four US communities (Forsyth County, NC; Jackson, MS; the suburbs of Minneapolis, MN; and Washington County, MD). For the present analysis, we used the liver enzyme levels and sociodemographic data that were collected at ARIC Visit 2 (1990–1992). Migraine diagnosis was reported using a validated headache questionnaire in participants who attended ARIC Visit 3 (1993–1995). ³⁷ Among the 14,348 participants at Visit 2, we excluded participants with missing data on liver enzymes (AST, ALT, or GGT; n = 976), as well as those with missing data about migraine at Visit 3 (n = 1528). Participants whose race was not White or Black were excluded, as were Black participants in Minnesota and Washington Country (n = 80) to minimize potential confounding effects and ensure the robustness of the statistical analyses. Individuals with missing information on BMI, smoking status, alcohol use, education level, and women with missing data on estrogen use were also excluded (n = 46). The final analytic sample thus included 11,718 participants with data on liver enzyme levels at Visit 2 and without a history of liver disease who completed the headache questionnaire at Visit 3.

Standard protocol approvals, registrations, and patient consent

The institutional review boards of all institutions participating in the ARIC approved the study, and all participants provided written informed consent at each research visit.

Assessment of migraine and headache characteristics

Migraine was characterized using the criteria of the modified International Classification of Headache Disorders (ICHD), second edition, ³⁸ based on a questionnaire on headache administered to participants who attended ARIC Visit 3 (1993–1995), and participants with definite/probable migraine were defined as individuals with prevalent migraine. Definite migraine was reported for participants fulfilling four criteria: (1) headaches lasting ⩾4 h; (2) headaches with at least two of the following three features: (a) throbbing, pulsing, or pounding quality, (b) unilateral location, (c) desire to go to a dark room and lie down; (3) nausea or vomiting or both photophobia and phono-phobia; and (4) ⩾1 year history of headaches (at any point in their life). Participants with a history of headaches lasting ⩾4 h and satisfying two of the three features in the second criterion were categorized as probable migraine. These criteria are consistent with previous studies. ³⁹ In addition, visual aura was recorded if participants reported seeing spots, jagged lines, or heat waves in one or both eyes before the onset of headache.

Measurement of liver enzymes

ALT, AST, and GGT were measured in Visit 2 (1990–1992) from serum samples stored at −80°C since collection, using Roche AST, ALT, and GGT reagents and the Roche Modular P Chemistry analyzer. The laboratory interassay coefficient of variability (CV) was 6.5% at a value of 29 U/L and 1.6% at a value of 137 U/L for AST; 5.6% at a value of 40 U/L and 3.5% at a value of 135 U/L for ALT; and 5.1% at a value of 39 U/L and 2.9% at a value of 171 U/L for GGT.

Additional variables

Weight and height were measured using standardized protocols, and body mass index (BMI) was calculated as weight (kg)/height (m²). Smoking status and alcohol use were self-reported and categorized as never, former, or current. Prevalent hypertension was defined as an average of the last two pressure measurements, systolic blood pressure ⩾ 140 mm Hg or average diastolic blood pressure ⩾ 90 mmHg or use of hypertension medication.

Statistical analysis

Participants’ characteristics were categorized into sex-specific quartiles for each enzyme. Median (25th and 75th percentile) levels were reported for variables with a skewed distribution. For all other continuous variables, the mean and standard deviation were reported. Associations between differences in baseline characteristics, as well as migraine status and type of migraine by quartiles of liver enzymes, were tested with chi-square tests for categorical parameters and ANOVA for continuous parameters. Estrogen use among women was reported according to their medication history, and liver enzyme levels were studied by migraine status.

Multiple logistic regression models were used to assess associations between sex-specific quartiles of ALT, AST, and GGT levels with probable/definite migraine. We produced two models: Model 1 adjusted for age, sex, and race-center (Whites, Washington County; Whites, Minneapolis; Blacks, Jackson; Blacks, Forsyth County; or Whites, Forsyth County), and Model 2 adjusted for all the variables in Model 1 with the addition of BMI, drinking status, smoking status, education level, and estrogen use in women. We tested for linear trends across the sex-specific quartiles of liver enzymes. Restricted cubic spline models with knots at the 5th, 35th, 65th, and 95th percentiles of each liver enzyme were used to characterize the shape of the associations of ALT, AST, and GGT levels with probable/definite migraine, and models were centered at the 10th percentile of each liver enzyme. We also assessed associations between levels of liver enzymes and migraine with and without aura. Since nonalcoholic fatty liver disease (NAFLD), now termed metabolic dysfunction-associated steatotic liver disease (MASLD), is usually accompanied by changes in liver enzyme levels, we evaluated the abovementioned associations in a subsample excluding 659 participants with MASLD, defined as levels of AST ⩾ 31U/L for females or ⩾37 U/L for males, or ALT levels of ⩾31 U/L for women or ⩾41 U/L for men together with self-reported none or moderate alcohol consumption. ⁴⁰ All reported p-values were two-sided, and p < 0.05 was considered statistically significant. Analyses were conducted using Stata/SE 17.0 software (StataCorp, College Station, TX, USA).

Results

Participants’ characteristics

Our final analytic sample included 11,718 ARIC participants (56.5% women and 22.4% Blacks). Mean (SD) age was 56.8 (5.7) years at ARIC Visit 2 (1990–1992). Table 1 presents the characteristics of the participants by migraine status, and Table 2 presents the migraine characteristics. Among the 1626 (13.8%) participants with migraine, 742 were defined as probable migraine, and 884 as definite migraine. Median (25%–75%) levels of GGT, AST, and ALT were significantly higher in participants without migraine versus participants with probable/definite migraine, with mean values of GGT 21.0 (15.0–33.0) versus18.0 (12.0–30.0), AST 20.0 (17.0–23.0) versus 18.0 (16.0–22.0), and ALT 15.0 (11.0–20.0) versus 13.0 (10.0–17.0) (Table 1). Other risk factors, including gender and smoking, were distributed differently between those with and without diagnosed migraine (Table 1). Participants with definite migraine reported more severe symptoms like nausea and/or vomiting, lights, and sounds compared to those with probable migraine (Table 2). Characteristics of the study population by sex-specific quartiles of AST, ALT, and GGT levels are presented in Supplemental Tables 1–3.

Table 1.

Characteristics of the study population by migraine status, the Atherosclerosis Risk in Communities study, visit 2, 1990–1992, N = 11,718.

Participants	Total	No migraine	Probable migraine	Definite migraine
N (%)	11,718	10,092 (86.1%)	742 (6.3%)	884 (7.5%)
Age (SD), years	56.8 (5.7)	57.1 (5.7)	55.4 (5.6)	55.0 (5.3)
Black, %	22.4	23.5	21.6	10.5
Women, %	56.5	53.3	69.3	82.9
Race-center, %
Minneapolis Whites	27.6	27.3	28	29.5
Jackson Blacks	19.8	20.8	18.7	9.7
Washington Co. Whites	26.5	25.7	27.9	35.1
Forsyth Blacks	2.5	2.7	2.8	0.8
Forsyth Whites	23.6	23.5	22.5	24.9
Educational level, %		19.8	20.1	17.2
Less than high school	19.6	19.8	20.1	17.2
High school or equivalent	42.1	41.4	43.9	48.6
College or above	38.2	38.7	36	34.2
BMI (kg/m2)	28.0 (5.3)	28.0 (5.3)	27.9 (5.5)	27.5 (5.7)
Estrogen use, women %	26.3	24.2	35.2	43.9
Hypertension, %	38.7	39.3	36.0	34.5
Smoking status, %
Current	20.3	20.4	22.0	17.3
Former	38.3	38.9	35.6	34.4
Never	41.4	40.7	42.5	48.3
Alcohol drinking, %
Current	57.6	58.1	54.3	54.4
Former	19.8	19.6	22.9	19.6
Never	22.6	22.3	22.8	26.0
AST (U/L), median (25%–75%)	20.0 (17.0–23.0)	20.0 (17.0–23.0)	19.0 (16.0–22.0)	18.0 (16.0–22.0)
ALT (U/L), median (25%–75%)	14.0 (11.0–19.0)	15.0 (11.0–20.0)	14.0 (10.0–18.0)	13.0 (10.0–17.0)
GGT (U/L), median (25%–75%)	21.0 (14.0–33.0)	21.0 (15.0–33.0)	19.0 (13.0–29.0)	18.0 (12.0–30.0)

Data are presented as percentages or as specified. Migraine was characterized using the modified International Classification of Headache Disorders (ICHD), second edition.

ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transpeptidase.

Table 2.

Headache characteristics by migraine status, the Atherosclerosis Risk in Communities study, visit 2, 1990–1992, N = 11,718.

Categories	Total N = 11,718	No migraine n = 10,092 (86.1%)	Probable migraine n = 742 (6.3%)	Definite migraine n = 884 (7.5%)
Headaches lasting more than 4 h, %
Yes	21.5	8.8	100	100
No	78.4	91.0	0	0
Missing	0.2	0.2	0	0
Pain mostly on one side, %
Yes	10.9	2.3	62.8	65.5
No	10.4	6.3	37.2	34.5
Missing	78.7	91.4	0	0
Throbbing, pulsing, or pounding, %
Yes	15.5	3.6	82.9	94.7
No	5.8	5.0	17.1	5.3
Missing	78.8	91.4	0	0
With nausea and/or vomiting, %
Yes	6.2	0.2	12.3	69
No	15.3	8.5	87.7	31.0
Missing	78.6	91.2	0	0
Lights bother, make it worse, %
Yes	8.9	0.9	26.1	86.1
No	12.4	7.8	73.9	13.7
Missing	78.7	91.3	0	0.2
Sounds bother, make it worse, %
Yes	10.4	1.6	37.6	87.6
No	11.0	7.1	62.3	12.3
Missing	78.7	91.3	0.1	0.1
Want to go into dark room, lie down, %
Yes	12.5	1.5	64.7	94.3
No	8.9	7.2	35.3	5.7
Missing	78.6	91.3	0	0
Notice jagged lines before headache, %
Yes	4.7	1.1	17.9	34.8
No	16.6	7.6	81.8	65
Missing	78.7	91.3	0.3	0.1
Physician said you have migraines, %
Yes	10.6	5.5	24.9	56.2
No	89.3	94.4	74.8	43.7
Missing	0.1	0.1	0.3	0.1
Parent suffers from migraines, %
Yes	11.1	8.4	20.9	33.4
No	88.0	90.8	77.9	64.4
Missing	0.9	0.8	1.2	2.3

Migraine was characterized using the modified International Classification of Headache Disorders (ICHD), second edition.

Association of AST, ALT, and GGT plasma levels with probable/definite migraine

Overall

As shown in Table 3, after adjusting for age, race-center, sex and other risk factors (Model 2), the odds ratios (95% CI) for migraine in the lowest compared to the highest quartile were 1.24 (1.06–1.45) for AST, 1.17 (1.00–1.37) for ALT, and 1.21 (1.03–1.41) for GGT. For AST, increased odds of migraine were observed also for Q2 levels compared with Q4 levels (adjusted OR, 95% CI 1.22, 1.04–1.43). Lastly, restricted cubic spline models showed negative associations between the levels of all three enzymes (GGT, AST, and ALT) and probable/definite migraine (shown in Figure 1).

Table 3.

Odds ratios (95% confidence intervals) for probable/definite migraine by sex-specific quartiles of liver enzymes, the Atherosclerosis Risk in Communities study, 1990–1992.

Categories	n	Model 1	Model 2
AST
Q1	507	1.25 (1.07–1.46)	1.24 (1.06–1.45)
Q2	468	1.22 (1.05–1.43)	1.22 (1.04–1.43)
Q3	339	1.18 (0.99–1.39)	1.17 (0.99–1.38)
Q4	312	1 (Reference)	1 (Reference)
p for linear trend*		0.006	0.007
ALT
Q1	488	1.22 (1.05–1.42)	1.17 (1.00–1.37)
Q2	391	1.05 (0.90–1.23)	1.02 (0.87–1.19)
Q3	391	1.1 (0.94–1.29)	1.09 (0.93–1.27)
Q4	356	1 (Reference)	1 (Reference)
p for linear trend*		0.021	0.089
GGT
Q1	562	1.28 (1.10–1.48)	1.21 (1.03–1.41)
Q2	346	1.05 (0.89–1.23)	1.02 (0.86–1.20)
Q3	381	1.1 (0.93–1.29)	1.1 (0.93–1.29)
Q4	337	1 (Reference)	1 (Reference)
p for linear trend*		0.002	0.038

Model 1: Adjusted for age, sex, race-center. Model 2: Adjusted for covariates in Model 1 plus BMI, smoking status, drinking status, education level, and estrogen in women. AST levels (U/L): MEN Q1 <12, Q2 12–15, Q3 16–21, Q4 ⩾25; WOMEN Q1 <16, Q2 16–18, Q3 19–21, Q4 ⩾22. ALT levels (U/L): MEN Q1 <18, Q2 18–21, Q3 21–25, Q4 ⩾22; WOMEN Q1 <12, Q2 10–12, Q3 13–16, Q4 ⩾17. GGT levels (U/L): MEN Q1 <18, Q2 18–24, Q3 25–37, Q4 ⩾38; WOMEN Q1 <13; Q2 13–17; Q3 18–28; Q4 ⩾29.

^*Cochran–Armitage test.

ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transpeptidase.

Figure 1.

Adjusted odds ratios (OR; 95% confidence intervals (CIs) for probable/definite migraine by levels of GGT, AST, and ALT.

Adjusted odds ratios (95% CI) for migraine by baseline GGT (panel a), AST (panel b), and ALT (panel c). Gray shading represents the 95% confidence interval. Baseline biomarkers were modeled using restricted cubic splines (solid lines) with knots at the 5th, 27.5th, 50th, 72.5th, and 95th percentiles of each liver enzyme, and models were centered at the 10th percentile of each enzyme and adjusted for age, sex, race-center, education level, smoking status, BMI, drinking status, and use of estrogen (in women only).

ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transpeptidase.

Comparing with and without aura

Of the 1626 participants with probable/definite migraine, data on the presence or absence of aura were available for 1182. In total, 741 (72.7%) reported migraine without aura, and 441 (27.3%) reported migraine with aura. In the analysis by aura presence or absence (Table 4), the lowest levels of ALT were associated with increased odds of migraine without aura (adjusted OR, 95% CI 1.38, 1.05–1.83 for ALT Q1 compared with Q4). A similar association was observed for AST but only in Model 1 (OR, 95% CI 1.36, 1.04–1.79 for AST Q1 compared with Q4). GGT levels were not associated with migraine without aura. Conversely, no significant associations were observed between individual liver enzyme levels and the presence of migraine with aura.

Table 4.

Odds ratios (95% confidence intervals) for probable/definite migraine with and without aura, by sex-specific quartiles of liver enzymes stratified by aura status in the participants with migraine, the Atherosclerosis Risk in Communities study, visit 2 1990–1992, N = 1182.

Categories	Model 1	Model 2
Migraine without aura, n = 741
AST
Q1	1.36 (1.04–1.79)	1.3 (0.99–1.71)
Q2	1.11 (0.84–1.44)	1.09 (0.83–1.42)
Q3	0.99 (0.75–1.31)	0.98 (0.74–1.30)
Q4	1 (Reference)	1 (Reference)
p for linear trend*	0.013	0.035
ALT
Q1	1.37 (1.05–1.78)	1.38 (1.05–1.82)
Q2	1.08 (0.83–1.41)	1.11 (0.83–1.44)
Q3	1.05 (0.81–1.37)	1.08 (0.83–1.42)
Q4	1 (Reference)	1 (Reference)
p for linear trend*	0.02	0.024
GGT
Q1	1.15 (0.88–1.49)	1.25 (0.95–1.65)
Q2	1.06 (0.80–1.41)	1.09 (0.82–1.47)
Q3	1.01 (0.76–1.34)	1.02 (0.76–1.36)
Q4	1 (Reference)	1 (Reference)
p for linear trend*	0.237	0.078
Migraine with aura, n = 441
AST
Q1	1.01 (0.56–1.77)	0.97 (0.54–1.75)
Q2	1.29 (0.70–2.38)	1.3 (0.70–2.43)
Q3	1.13 (0.60–2.14)	1.16 (0.61–2.23)
Q4	1 (Reference)	1 (Reference)
p for linear trend*	0.989	0.916
ALT
Q1	1.04 (0.56–1.93)	1.1 (0.57–2.10)
Q2	0.58 (0.32–1.04)	0.61 (0.33–1.13)
Q3	0.95 (0.51–1.78)	0.97 (0.51–1.85)
Q4	1 (Reference)	1 (Reference)
p for linear trend*	0.702	0.879
GGT
Q1	0.65 (0.37–1.14)	0.74 (0.40–1.36)
Q2	0.62 (0.33–1.15)	0.67 (0.35–1.27)
Q3	0.99 (0.53–1.83)	1.07 (0.57–2.00)
Q4	1 (Reference)	1 (Reference)
p for linear trend*	0.063	0.191

Data on aura presence/absence were available for 1,623/1,626 participants with migraine. Model 1: Adjusted for age, sex, race-center. Model 2: Adjusted for covariates in Model 1 plus BMI, smoking status, drinking status, education level, and estrogen in women. AST levels (U/L): MEN Q1<12, Q2 12–15, Q3 16–21, Q4 ⩾25; WOMEN Q1<16, Q2 16–18, Q3 19–21, Q4 ⩾22. ALT levels (U/L): MEN Q1<18, Q2 18–21, Q3 21–25, Q4 ⩾22; WOMEN Q1<12, Q2 10–12, Q3 13–16, Q4 ⩾17. GGT levels (U/L): MEN Q1<18, Q2 18–24, Q3 25–37, Q4 ⩾38; WOMEN Q1<13; Q2 13–17; Q3 18–28; Q4 ⩾29.

^*Cochran–Armitage test.

ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyl transpeptidase.

Without MASLD

Results of the analysis excluding individuals with MASLD (Table 5) indicated increased odds of migraine in participants with lower levels of AST, both for Q1 compared with Q4 (adjusted OR, 95% CI 1.22, 1.03–1.44) and Q2 compared with Q4 (adjusted OR, 95% CI 1.42, 1.01–1.43). Higher odds of migraine were also found for GGT lower levels (adjusted OR, 95% CI 1.19, 1.01–1.40 for GGT Q1 levels compared with Q4). Finally, no significant associations between migraine and ALT levels were found in this subsample.

Table 5.

Odds ratios (95% confidence intervals) for probable/definite migraine by sex-specific quartiles of liver enzymes among participants without NAFLD, the Atherosclerosis Risk in Communities study, visit 2, 1990–1992, N = 11,059.

Categories	n	Model 1	Model 2
AST
Q1	507	1.21 (1.02–1.43)	1.22 (1.03–1.44)
Q2	468	1.19 (1.00–1.42)	1.42 (1.01–1.43)
Q3	336	1.14 (0.95–1.37)	1.14 (0.95–1.37)
Q4	235	1 (Reference)	1 (Reference)
p for linear trend*		0.032	0.028
ALT
Q1	487	1.17 (1.00–1.38)	1.13 (0.96–1.34)
Q2	389	1.01 (0.86–1.20)	0.98 (0.83–1.17)
Q3	387	1.07 (0.90–1.26)	1.05 (0.89–1.25)
Q4	283	1 (Reference)	1 (Reference)
p for linear trend*		0.077	0.201
GGT
Q1	558	1.26 (1.08–1.48)	1.19 (1.01–1.40)
Q2	340	1.03 (0.87–1.23)	1.00 (0.84–1.19)
Q3	367	1.09 (0.92–1.29)	1.09 (0.92–1.29)
Q4	281	1 (Reference)	1 (Reference)
p for linear trend*		0.005	0.066

Model 1: Adjusted for age, sex, race-center. Model 2: Adjusted for covariates in Model 1 plus BMI, smoking status, drinking status, education level, and estrogen in women. AST levels (U/L): MEN Q1 <12, Q2 12–15, Q3 16–21, Q4 ⩾25; WOMEN Q1 <16, Q2 16–18, Q3 19–21, Q4 ⩾22. ALT levels (U/L): MEN Q1 <18, Q2 18–21, Q3 21–25, Q4 ⩾22; WOMEN Q1 <12, Q2 10–12, Q3 13–16, Q4 ⩾17. GGT levels (U/L): MEN Q1 <18, Q2 18–24, Q3 25–37, Q4 ⩾38; WOMEN Q1 <13; Q2 13–17; Q3 18–28; Q4 ⩾29.

^*Cochran–Armitage test.

Discussion

In the ARIC community-based population, lower levels of AST, ALT, and GGT by sex-specific quartiles were associated with a higher prevalence of probable/definite migraine. The results were consistent after excluding individuals with presumed MASLD, which usually accompanies alterations in liver enzyme levels. Interestingly, in the analyses by type of migraine, the associations between higher levels of liver enzymes and migraine were significant only for migraine without aura, and only for ALT after controlling for risk factors, supporting previously reported abnormal brain glutamate levels in free-aura patients. ⁹ However, in this study, the number of participants reporting migraine with aura was small. Thus, the statistical power was limited in the analysis by aura status.

Migraine is a common but poorly understood nervous system disorder, and the reasons for the onset of a migraine attack remain unclear. While migraine is frequently considered a spontaneous disease and not related to specific risk factors, many patients report a linkage between exogenous and endogenous factors and an increased probability of attacks.^41,42 Among the modifiable reported risk factors, obesity, depression, and stressful life events are the most important; however, they mainly affect the progression from episodic to chronic migraine, but not the initial odds of migraine occurrence.^43,44 Accumulated evidence now supports the role of glutamate in the pathogenesis of migraine. Plasma glutamate concentrations have been reported to be higher in subjects suffering from migraine and in patients suffering from tension-type headaches compared to controls.^19,45 Moreover, a high plasma concentration of glutamate was reported in episodic and chronic migraine with and without aura.^12,16,46 Up to one-third of migraine patients experience aura. ⁴⁷ However, whether migraine with and without aura are different in neurotransmitter dysregulation remains unclear. Lowering plasma glutamate levels by preventive therapies has proven to be an effective prophylactic treatment for migraine without aura, supporting the potential role of blood glutamate regulation in the occurrence of this type of migraine.^18,20 Nevertheless, the mechanisms underlying the increased plasma levels in migraine patients are yet to be elucidated. Reduced platelet stimulus–response and impaired platelet function as a reflection of central serotonergic disturbances may contribute to increased glutamate plasma concentration in migraine patients, while differences in the glutamate release/reuptake ratio of platelets may be a key regulatory factor in free-aura versus with aura migraine.^45,48 Accordingly, there is a plausible hypothesis that the increase in glutamate levels in neurons and blood platelets may affect blood glutamate levels in migraine patients.^49–51

An alternative potential mechanism involved in migraine pathology, as well as a future treatment direction, concerns the blood glutamate balance by glutamate–regulatory enzymes such as AST and ALT. ¹⁶ Low-activity plasma-resident enzyme AST has been associated with higher plasma glutamate levels in migraine patients, as has a positive correlation between the time elapsed from attack onset and glutamate levels during the ictal period. ¹⁶ Our results showed a negative correlation between blood AST and ALT activity and the prevalence of probable/definite migraine. This is in line with previous observations that ischemic stroke patients with good functional outcomes have shown high AST activity and low glutamate levels in plasma at admission, demonstrating an association between blood glutamate and AST activity and reflecting brain glutamate alterations.^16,34 Although glutamate cannot freely pass through the BBB, the efflux of excess glutamate from the CNS to the plasma may occur when there is greater BBB permeability under increased CNS glutamate levels.^21,52 Our research, along with others, has shown the efflux of Glu from the brain to the blood through capillary transporters and an accelerated efflux following injection of AST or ALT. This causes a significant decrease in glutamate levels in the blood and consequently in the CNS, as measured by MRS and high-performance liquid chromatography (HPLC) methods in animal models of neurotrauma.^{26,28,29,36,53} These findings thus suggest that AST and ALT may also play a role in migraine pathology and could serve as potential targets for future migraine treatments. Another liver enzyme that is linked to glutamate synthesis is GGT. Extracellular GGT is mainly located on the extracellular membranes of the liver, kidney, blood vessels, brain, and heart, ⁵⁴ where it is responsible for transporting amino acids into the cells and is involved in the metabolism of glutathione, the most important intracellular antioxidant produced during normal metabolism. ⁵⁵

GGT activity is mainly genetically determined, and its heritability has been estimated to range from 50% to 77% in adults. ⁵⁶ Half of the genetic variance in GGT is thought to be shared by ALT and AST. ⁵⁷ Higher enzyme activity is independently associated with higher rates of classical vascular risk factors, including hypertension and smoking, stroke, and all-cause mortality.^58–60 In this study, there were nonlinear associations between GGT levels and rates of migraine, although no significant association with GGT levels was evident when we analyzed only individuals with migraine and evaluated associations in those with and without aura. Since one of the products of GGT activity is temporarily increased glutamate, we expected to find positive associations between GGT level and the prevalence of migraine; however, our results showed negative associations. Although the mechanisms by which lower serum GGT levels might be associated with a higher prevalence of migraine are unknown, a possible explanation for our findings might be related to the antioxidative activity of the enzyme. However, further studies are needed to explore the relationship between blood levels of glutamate, AST, ALT, and GGT, and the occurrence of migraines, both with and without aura.

While this study focuses primarily on the role of systemic glutamate-regulating enzymes in migraine, it is important to acknowledge the potential contribution of vascular mechanisms. Liver enzyme activity is not solely a reflection of hepatic metabolism but is also influenced by broader systemic and vascular factors. Given that migraine is recognized as a neurovascular disorder, it is plausible that the associations observed in our study result from a combination of metabolic and vascular influences. For example, endothelial glutamate transporters at the BBB regulate glutamate flux between the periphery and the central nervous system, and their function may be modulated by vascular integrity and systemic inflammation. In this context, altered enzyme levels may reflect a broader dysregulation of glutamatergic signaling and vascular homeostasis. This multifactorial perspective underscores the need for future research to explore how liver enzyme activity interacts with neurovascular function and glutamate metabolism in the pathophysiology of migraine. Future studies with longitudinal data and larger sample sizes that incorporate both the monitoring of glutamate-regulatory enzyme activity, and the measurement of plasma glutamate levels, are needed to better characterize causal relationships and evaluate the predictive value of these biomarkers in migraine pathophysiology.

Study strengths and limitations

This study has several notable strengths, including the use of a large community-based population with comprehensive data on sociodemographic and clinical variables. Nonetheless, there are some limitations. First, blood glutamate levels were not available, and therefore, we could not study the direct association between glutamate and its metabolites and migraine. Second, only a single measure of liver enzymes was available for ARIC participants. Given the large intraindividual variation in AST, ALT, and GGT, the use of a single measure of these enzymes may bias the results toward the null. ⁶¹ Third, the power in our analysis of migraine with/without aura was limited due to the small number of participants with migraine with aura. Fourth, migraine was self-reported, so information bias cannot be excluded. Fifth, the available data captured only visual auras, limiting the ability to identify other clinically relevant aura types such as sensory, language, or motor disturbances. This may have introduced classification bias in identifying cases of migraine with aura and potentially underestimates the true prevalence of this subtype in the study population.

Next, due to the fact that the available data on migraine were collected about 3 years after the measurement of liver enzymes, this did not allow for a true cross-sectional evaluation of the prevalent associations, nor for the longitudinal assessment of associations with migraine incidence. This is relevant since migraine is a long-term neurological condition, where changes in migraine status over a period of several years are not common. ⁶² Migraine onset peaks well before age 40 in the vast majority of people, so it is unlikely that significant numbers of newly diagnosed migraine would occur in the ARIC population after Visit 3. Moreover, a systematic review of primary headaches reported no substantial changes in types of headaches during the lifespan. ⁶³

Last, while this cohort does cover part of the age range where migraine remains prevalent, it does not fully capture the younger population in which migraine incidence peaks, particularly individuals aged 30–45. Additionally, many women in this cohort are in the perimenopausal or early postmenopausal period, during which hormonal fluctuations can influence migraine frequency, severity, and presentation. These factors may affect the generalizability of our findings and should be considered when interpreting the results. Future studies involving younger populations and stratification by hormonal status would help to further clarify the observed associations and validate these findings across different life stages.

Liver enzyme activity and its association with glutamate regulation represent more intrinsic metabolic pathways that are likely to remain stable over time, particularly when adjusted for confounders such as age, sex, and BMI. Nevertheless, future studies using more recent datasets and including direct assessments of diet, physical activity, and stress levels would be valuable to further elucidate the impact of these variables on the relationships we observed.

Statistical considerations: One limitation of our study is the absence of a formal sample size calculation. The analyses were conducted on a subset of ARIC participants with complete data, which may limit the power to detect small-to-moderate associations. While this study employed multiple logistic regression models to assess the associations between liver enzyme quartiles and migraine prevalence, we acknowledge that this approach does not involve formal model establishment or validation. Given the cross-sectional design and exploratory nature of our analysis, logistic regression was selected for its robustness and interpretability in large epidemiological datasets. However, more complex modeling techniques, such as predictive risk modeling or machine learning approaches, could provide additional insight into the mechanistic links between liver enzymes and migraine, particularly if validated in independent cohorts. Future studies with longitudinal data and larger sample sizes should explore these avenues to better characterize causal relationships and evaluate the predictive value of these biomarkers in migraine pathophysiology

Conclusion

To the best of our knowledge, this is the first report on associations between liver enzymes and migraine. Although the mechanisms by which lower serum AST, ALT, and GGT levels may be associated with migraine are not clear, inefficient plasma glutamate regulation might explain the role of AST and ALT in the pathology of migraine, especially in migraine without aura.

Appendix

Abbreviations

ALT alanine transaminase

ARIC Atherosclerosis Risk in Communities

AST aspartate transaminase

BBB blood-brain barrier

BMI body mass index

CGRP calcitonin gene-related peptide

CSD cortical spreading depolarization/depression

CSF cerebrospinal fluid

EAAT1–3 excitatory amino-acid transporters 1–3

GABA gamma-aminobutyric acid

GGT gamma-glutamyl transferase

GOT1 glutamate-oxaloacetate transaminase 1

GPT1 glutamate-pyruvate transaminase 1

HPLC high-performance liquid chromatography

ICHD International Classification of Headache Disorders

MASLD metabolic dysfunction-associated steatotic liver disease

NAFLD non-alcoholic fatty liver disease

NMDA N-methyl-D-aspartate

NR2B N-methyl-d-aspartate receptor subunit 2B

rTMS repetitive transcranial magnetic stimulation

¹H-MRS proton magnetic resonance spectroscopy