Depression is one of the most common neuropsychiatric and disabling disorders with an approximate 10% prevalence worldwide (1). The etiology of depression is multifactorial with a complicated neurobiological and psychosocial framework (2). Meanwhile, migraine is a common neurological disorder characterized by chronic and recurrent headache events predominantly in females, affecting 10–20% of adults worldwide (3). Analysis of the 2019 Global Burden of Disease data revealed that migraine was the second most disabling disease worldwide and the first among young women (4). Of note, migraine comorbid with depression has been observed in clinical studies (5, 6) far more commonly than expected by chance. In fact, the two diseases have a reported bidirectional association; that is, migraine increases the risk of major depression, and major depression increases the risk of migraine (7). Indeed, in migraine patients with depression, the clinical symptoms may be exacerbated and are more difficult to treat (8).

Previous research has demonstrated that epidemiological co-occurrence of diseases may not be random and can indicate possible shared factors, including genetic, environmental, or interactions between the two (9). Meanwhile, large health data banks with detailed medical information provides a significant opportunity for the exploration of the etiologies of complex disease comorbidities. A UK Biobank cohort study with 117,392 subjects supported the relationship between migraine and depression comorbidity (10). In particular, characterization of a shared genetic background may direct further personalized diagnosis and treatment. Indeed, the genetic association of migraine and depression comorbidity is well-documented in genome-wide association studies (GWASs) within European populations (11–13). Genetic risk score analysis conducted by Ligthart et al. (11) indicated that migraines with or without depression comorbidity are genetically distinct disorders and the subgroup of patients with depression and migraine comorbidity were genetically much similar to patients with depression. A genetic association between migraine and depression comorbidity was also observed in the Netherlands Twin Registry. By investigating discordant monozygotic twins, the study found that only the twin that had depression had increased risk for migraine (12). Since most studies addressing this issue were conducted in non-Asian populations, whether these GWASs finding are applicable in an Asian population remains questionable. Considering that migraine is a complicated neurovascular disease with a strong genetic component that differs for each subtype (14), we hypothesize that migraine patients with different subtypes may develop depression via different mechanisms.

To the best of our knowledge, no previous study has reported on the genetic basis of overall and subtype migraine, nor on depression comorbidity in an Asian population. Therefore, in this study, we perform a GWAS on a migraine cohort in a tertiary hospital in Taiwan, to identify genetic loci associated with comorbid depression and migraine. Furthermore, we divide the migraine cohort based on migraine subtypes: episodic migraine, chronic migraine, migraine with aura, and migraine without aura. We then analyze possible susceptible genetic loci by comparing the genotype of various migraine subtypes with and without depression.

We analyzed 966 patients with migraine and stratified them into groups with or without depression. SNP variants detected in the intron region of CDH4 (rs78063755), downstream region of NTRK3-AS1 (rs57729223), the intergenic region between NTRK3-AS1 and MRPL46 (rs16941601), and the intergenic region between LINC01918 and GPR45 (rs2679891) were found to be suggestively associated with depression in the Han Chinese population in Taiwan. However, the distance between rs57729223 and rs16941601 was close, and the r2 measure of linkage disequilibrium (LD) of rs57729223 and rs16941601 was around 0.94 using LDlink (24), indicating that the two SNPs have non-random association and that rs57729223 may be the only one with significance.

Cadherin 4 (CDH4) is located on chromosome 20q13.33 and encodes CDH4, a calcium-dependent cell-cell adhesion glycoprotein. CDH4 is mainly expressed in the brain but also in other tissues, including the spleen, ovary, testis, adrenal gland, and thyroid gland (25). CDH4 plays a key role in brain segmentation and neuronal outgrowth (26). A GWAS of brain aging on magnetic resonance imaging data showed that the SNP variant in CDH4 is significantly related to reduced total cerebral brain volume, potentially caused by impairment of the neural tract and synaptic development (27). Another GWAS of depression phenotypes in the UK Biobank also identified SNP variants in CDH4 as related variants (28). Although CDH4 polymorphism has been reported as a susceptible locus with respect to migraine (29), the associated mechanism remains unclear. Our study showed that the CDH4 intronic polymorphism rs78063755 is a candidate for further studies on migraine with depression in the Han Chinese population, possibly due to its impact on brain segmentation, neuronal growth, and synaptic development.

NTRK3-AS1 is an RNA gene located on chromosome 15q25.3 that is expressed in the testis (25). According to the GWAS catalog, NTRK3-AS1 polymorphisms are associated with the levels of growth-regulated protein alpha and 3-hydroxypropyl mercapturic acid in smokers (30). However, to the best of our knowledge, no previous study has reported the association between NTRK3-AS1 polymorphism and migraine or depression. Our analysis also identified another variant, rs57729223, in the downstream region of NTRK3-AS1, and two intergenic variants, rs16941601 and rs2679891, which have not previously been reported as associated with migraine and depression. As discussed earlier, because the distance between rs57729223 and rs16941601 was close and the two SNPs are not independent, rs57729223 is considered the one with significance. In addition to all migraine groups, rs57729223 was genetically suggestively associated with comorbid depression in the EM and MoA subgroups. However, the mechanism is unclear and warrants further investigation, particularly in the Han Chinese population.

To investigate whether different SNPs are associated with depression in different migraine types, we analyzed SNP differences between patients with or without depression in different subgroups. Eight SNPs in the EM group and ten in the MoA group were suggestively associated with comorbid depression (OR > 1). One SNP each in EM and CM had OR < 1. In the EM group, we identified nine SNPs suggestively associated with depression (eight with OR > 1; one with OR < 1). Two variants with OR > 1 (rs57729223 and rs16941601) also demonstrated an association with depression in all migraine groups, since rs57729223 and rs16941601 are not independent, the association may be only in rs57729223. Six other SNP variants were suggestively associated with depression in the EM group, including the intronic regions of CHD2L, GMO5: rs2297754, intronic region of BUB3: rs75155690, intronic region of PDZRN4: rs79135118, 3’UTR region of ATOH8: rs1055770, the intergenic region between MIR197 and FAM241A: rs5897874, and the intergenic region between OR2F2 and OR2F1: rs41376753.

BUB3 is located on chromosome 10q26.13 and encodes BUB3, a mitotic checkpoint protein belonging to the Bub protein family that plays an important role in the spindle assembly checkpoint (31). Abnormal BUB3 expression causes defective mitotic function, impaired spindle gate function, and chromosome instability, leading to aneuploidy and possibly tumorigenesis (31). The GWAS catalog documented its association with exacerbations in children despite using long-acting beta 2-agonists and plasma metabolite kynurenic acid measurement (30). However, the association between BUB3 SNP and the pathophysiology of migraine or depression remains unclear and warrants further analysis.

PDZ domain-containing ring finger 4 (PDZRN4) is located on chromosome 12q12 and encodes the PDZ domain-containing ring finger 4 (PDZRN4), which belongs to the LNX family and acts as a suppressor in multiple cancers (32, 33) and in association with multiple sclerosis in a GWAS (34). However, to date, no study has discussed its relationship with migraine or depression, and hence may serve as a candidate for further research.

ATOH8 is located on chromosome 2p11.2 and encodes protein ATOH8, which belongs to a family of transcriptional regulators: basic helix-loop-helix proteins (35). ATOH8 is involved in embryonic development, disease initiation, and disease progression (35). The GWAS catalog reports that SNPs in the ATOH8 are suggestively associated with optic disk size, abdominal aortic aneurysm, total blood homocysteine level, total cerebrospinal fluid paired helical filament-tau level, and metastasis in low/stable colorectal cancer (30). Meanwhile, the association of ATOH8 with depression or migraine has not been reported, and further studies are needed to delineate the possible mechanisms.

One SNP variant, the intronic deletion in steroid receptor RNA activator 1 (SRA1) gene (rs60236323), had an OR < 1 with a variant allele frequency of 50.7% in non-depression EM patients and 37.5% in depression EM patients. The SRA1 gene is located on chromosome 5q31.3 and has a bi-functional role as both protein and non-coding RNAs (36). Moreover, it is involved in the development of multiple cancers (37, 38) and cardiomyopathy (39). However, its association with migraines and depression remains unclear. Whether this SNP variant has a possible protective role in Han Chinese patients with EM requires further investigation with larger samples.

In the MoA group, we identified ten variants suggestively associated with depression (OR > 1). Two of which (rs78063755 in CDH4 and rs57729223 in NTRK3-AS1) were also found to be associated with depression in the migraine group. Six ncRNA intronic variants (rs9268145, rs6910071, rs3763305, rs28361060, rs9268362, and rs3817964) were located in the same gene TSBP1-AS1, and two intronic variants in NRG3 (rs10748994) and MTSS1 (rs1017342).

TSBP1-AS1, located on chromosome 6q21.32, is an RNA gene affiliated with the lncRNA class (30). The function of TSBP1-AS1 remains to be identified. Nevertheless, it is highly expressed in immune system cells (40) and genetically associated with several immune-related diseases, hepatitis, and dermatologic disorders, according to the GWAS catalog (30). Interestingly, TSBP1-AS1 is located in the major histocompatibility complex (MHC) region (41). Previous studies have reported that the clinical features of migraine may be influenced by SNPs located in the MHC region (42, 43). Huang et al. (43) showed that some alleles of HLA class I were associated with migraine and increased risks of chronic migraine and medication overuse. However, whether migraine is genetically related to immune disorders warrants further investigation. A previous GWAS identified a TSBP1-AS1 variant (rs140002913) in association with migraine disorder (44). In our study, we identified six TSBP1-AS1 variants in association with MoA and comorbid depression with variant allele frequencies 2.86–3 times higher than those without comorbid depression. Further studies are needed to elucidate the role and pathophysiology of TSBP1-AS1 in migraine and depression.

NRG3 is located on chromosome 10q23.1 and encodes neuregulin 3, a protein belonging to the neuregulin family that plays important roles in neuronal development, differentiation, proliferation, and plasticity (45, 46). According to the GWAS catalog, NRG3 polymorphisms are genetically associated with various diseases, including cancer, insulin resistance, and tau protein formation (30). However, the association between migraine and depression has not been reported. We observed that the frequency of the variant allele (rs10748994) was higher in MoA patients with comorbid depression. The underlying mechanism remains unknown, and NRG3 may serve as a suitable candidate for further studies.

MTSS1 is located on chromosome 8q24.13 and encodes metastasis suppressor 1, a protein involved in cell morphology, motility, development, and metastasis (47). According to the GWAS catalog, its SNPs were reported to be genetically associated with multiple traits, including cardiovascular disease, tau protein formation, and cognitive function (30). No association between MTSS1 SNPs and migraine or depression has been reported. We identified a MTSS1 variant (rs1017432) that may be genetically associated with MoA and comorbid depression. Further studies are needed to validate our findings and explore the underlying mechanisms.

A multivariate association study demonstrated that the SNP variant in the intron region of CDH4 rs78063755 was associated with MIDAS and BDI scores. Although the association between migraine and the variant rs78063755 has not been widely discussed, it was suggestively associated with multiple migraine-related indices and may serve as a candidate for further studies. We employed Affymetrix Axiom Genome-Wide TWB 2.0, which covers a representative sample of the gene pool in Taiwan and contains approximately 752,921 probes for a total of 686,463 SNPs in this study, whereas only 6 out of 101 SNPs reported in European cohorts (23) were found in the TWB array and none of the SNPs were significantly related to depression and migraine comorbidity in our cohort. This might have been due to all the participants in our study belonging to the Han Chinese population, whereas the previous study focused primarily on a European population.

Our study has several strengths. For instance, the strict diagnostic criteria and designed standard operation protocol allowed us to record detailed patient data, including history and medical evaluation, correct diagnosis made by a headache expert, and evaluations using validated questionnaires. Further statistical analysis, including stratified analysis of migraine subgroups and multivariate regression analysis, allowed us to investigate the genetic association between migraine and comorbid depression. Moreover, we employed Affymetrix’s Axiom Genome-wide TWB 2.0 array, which covers a representative sample of the gene pool in Taiwan (19), and identified several SNPs suggestively associated with migraine or depression. Our study had some limitations. We did not perform genotype imputation prior to performing GWAS because using data from European populations may not necessarily increase the number of matched SNPs and the results obtained may not be applicable in the Asian population. However, further exploration of this study with genotype imputation will be performed when more cross-population whole exome sequencing data are available. We did not perform replication on all the SNPs found in the European population due to limited SNPs in the TWB 2.0 array. Our cohort sample size was relatively modest, and the sub-groups results could not meet a traditionally accepted P-value threshold of <5E-08 in GWAS. Hopefully, our exploratory genetic association results may shed light on the genetic links for depression and migraine comorbidity; however, further validation is warranted.

This study showed that migraine patients with SNP variants in the intron region of CDH4: rs78063755, downstream region of NTRK3-AS1: rs57729223, and the intergenic region between LINC01918 and GPR45: rs2679891 are suggestively associated with depression in the Han Chinese population in Taiwan. We also identified several SNPs suggestively associated with depression in migraine subgroups (EM, CM, and MoA). These results provide insights into the possible genetic basis of both overall migraine, and its subtypes, and comorbid depression in the Han Chinese population.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: European Variation Archive, PRJEB57851.

M-CT and F-CY: conceptualization and writing—original draft. F-CY, C-LT, C-SL, Y-KL, C-KT, G-YL, YL, and K-SH: data curation. F-CY, M-CT, C-KT, YL, and K-SH: formal analysis. F-CY: methodology and writing—review and editing. All authors have read and approved the final version of the manuscript.