Implications for the migraine SNP rs1835740 in a Swedish cluster headache population

Migraine and cluster headache (CH) are two primary headache disorders that share pathological features and a majority of CH patients are successfully treated with drugs used for migraine, such as triptans [1‐4]. Both disorders are characterized by activation of the trigeminovascular system and neurogenic inflammation [5]. There are also phenotypic similarities between CH and migraine, such as recurring attacks of headache, lateralized pain and associated autonomic symptoms [3, 4]. However in migraine these symptoms are typically nausea and vomiting while in CH autonomic symptoms are e.g. rhinorrhea, lacrimation and ptosis. Sometimes tearing of the eye and nasal congestion at the site of pain as seen in CH can occur also in migraine. However, there are marked differences in the clinical manifestations and associated symptoms between migraine and CH. In particular, the diurnal and seasonal rhythmicity with which headache attacks can be observed in many CH patients is unparalleled in other primary headaches [6, 7].

Genetic factors are likely to influence the risk of developing both migraine and CH [8]. The heritability of CH is unclear, but having a first- or second-degree relative with CH greatly increases the risk of developing the disorder [9]. One study reports heritability (h²) for CH of 0.26, but this number is probably an underestimation since only 2.3% of the patients in the cohort reported a familial history [10]. In our Swedish CH material, 11.2% of the patients report that they have a first-, second- or third degree relative with CH [11]. Several candidate genes such as HCRTR2 (hypocretin receptor 2), ADH4 (alcohol dehydrogenase 4), NOS (nitric oxide synthase) and CLOCK (Circadian Locomotor Output Cycles Kaput) have been suggested to be associated with the disease, although results are conflicting [12‐19]. In contrast the heritability of migraine has been estimated to be around 50% (h² = 0.44) [8, 20]. The contribution of genetic factors to migraine has further been confirmed in several genome-wide association studies (GWAS) [21‐25].

We have previously replicated the finding of an association for two of the established risk alleles for migraine in a Swedish migraine case-control population, rs2651899 in PRDM16 (PR/SET domain 16) and rs1835740 located between MTDH (metadherin) and PGCP (carboxypeptidase Q) on chromosome 8q22.1 [26]. Association of the PRDM16 gene to migraine has been confirmed in several replication studies and meta-analyses and it is now considered a well-established migraine candidate gene [27‐32]. Replication studies on rs1835740 are less affirmative [33‐36]. The MTDH gene was found to associate with migraine in a gene-based multimarker analysis, but the rs1835740 variant did not reach genome-wide significance in a large GWAS meta-analysis [24, 31].

There are indications of CH being more common in families with migraine [37]. Moreover there are genes that have been suggested as candidate genes for both disorders, for example the MTHFR (methylenetetrahydrofolate reductase), HCRTR2 and iNOS (inducible NOS) genes, but there are no conclusive reports on shared genetic risk factors [13, 16, 38‐43].

The aim of this study was to investigate whether migraine and CH might share genetic determinants since similar pathophysiological events occur in both disorders. Analysis of the distribution of the genetic migraine markers rs2651899 and rs1853740 has not previously been performed in CH. We therefore screened a large Swedish CH case-control population and performed an association analysis of these two markers that have been shown to increase the risk for migraine in Sweden.

We genotyped 541 CH patients and 581 controls for the rs2651899 and rs1835740 SNPs, and tested for association with CH; results are represented in Table 1. For rs1835740, we found a weak association between the rare allele (T) and an increased risk for CH (Chi-square (Χ²) =4.1, OR 1.25, 95% CI 1.01–1.53, p-value = 0.043). Both the TT and the CT genotypes were more common in patients than controls, but there was no significant genotypic association. There was no association between rs2651899 in PRDM16 and CH in our material (Table 1). Both SNPs were in HWE in patients and controls (data not shown).

Since these two SNPs are known to influence the risk of migraine, and many CH patients in our material also suffer from migraine, we performed a stratified analysis with respect to migraine. Patients were separated into two groups; one consisting of patients suffering from CH only, and one consisting of CH patients with migraine. The allelic association of rs1835740 became stronger when analyzing CH patients with migraine, CH patients without migraine, and controls (Χ² = 6.96, p-value = 0.031) (Table 2). When comparing the allele and genotype frequencies, the mutated allele became increasingly more common in the group of patients suffering from both CH and migraine; 21.1% in CH patients and 26.2% in CH patients with migraine, as compared to the control group where the MAF was 18.5%. The stratified analysis did not affect the results for rs2651899.

The genetic association discovered here between CH and rs1835740, in combination with the discovery study reporting that rs1835740 potentially affects the expression levels of the MTDH gene lead us to also investigate the MTDH mRNA expression levels in a subset of patients and controls [21]. mRNA was obtained from primary fibroblast cell lines provided from 12 patients and 12 control individuals and analyzed using qRT-PCR. MTDH mRNA expression was normalized to PDHB, which was used as a reference gene, and then to a reference sample. One control individual was identified as an outlier using Grubb’s test and was removed from the analysis. We found a significant difference in the mRNA levels of MTDH when comparing individuals carrying two C alleles (wild-type) at rs1835740, with individuals having one or two T alleles, p = 0.0318, Fig. 1a. There was only one individual carrying the minor allele at both positions in our material, which is the reason we could not compare the three individual genotypes. We further analyzed the mRNA levels with respect to CH diagnosis, as can be observed in Fig. 1b, control subjects and CH patients had similar MTDH mRNA levels (p = 0.195).

We have analyzed a Swedish CH population for two SNPs that have previously been identified as genetic risk factors for migraine in Sweden and globally [21, 22]. One of the variants, rs1835740, was found to be associated with an increased risk for CH. The other SNP, rs2651899, was not associated with CH in Sweden.

A more detailed analysis of rs1835740 revealed that the association was even stronger in a subgroup of patients that have both CH and migraine. When comparing the MAF between the three groups, we found that the minor allele (T) was present in 26.2% of the individuals suffering from both disorders, which is 7.7% more than among controls, a significant difference (p = 0.016) with an OR of 1.56 (95% CI 1.10–2.21), and 5.1% more than in CH without migraine. This difference was not significant possibly because of the smaller sample sizes in the stratified analysis. Our findings thus appear to indicate that the T allele is enriched in the patient group having both CH and migraine. In a former GWAS suggesting MTDH as a candidate gene for migraine, the effect of the rs1835740 association was stronger in a subgroup of patients suffering from migraine with aura (MA) than in patients with migraine without aura (MO), which is also an indication of this SNP being associated with more complex headache phenotypes [21]. Interestingly, in a study on rs1835740 and migraine by Azimova et al. (2015) small groups of CH patients (n = 9) and chronic tension type headache patients (n = 20) were used as control groups. In their study, all chronic tension headache patients were wild type for rs1835740, while the MAF of CH patients was 44.4% [36]. One possible limitation of our study is the lack of clinical information related to our control material. We cannot exclude the occurrence of headache in the control group; statistically, the occurrence of CH should be extremely low, or non-existent, but we have to assume that around 15% of our controls also suffer from migraine, corresponding to the prevalence of migraine in the Swedish population. In view of our data, the occurrence of migraine in the control group would imply a higher MAF of rs1835740, and should therefore not introduce a risk of false positive results.

MTDH has been suggested to be a likely candidate gene for migraine because of its proximity to the intergenic SNP rs1835740. Moreover, rs1835740 was found to affect the gene expression of MTDH, thus providing an appealing pathophysiological mechanism for migraine: MTDH is known to downregulate EAAT2 (excitatory amino acid transporter 2), a major glutamate transporter, and could thus affect glutamate regulation at a synaptic level [52, 53]. We studied the MTDH mRNA levels in individuals with different rs1835740 genotype, and also analyzed expression with respect to CH diagnosis. Similarly to previous reports, the relative quantification showed a difference in expression between individuals with different genotypes for rs1835740 [21]. Individuals carrying one or two T alleles had significantly higher mRNA levels than individuals with the CC genotype. eQTLs (expression quantitative trait loci) may vary between different types of tissue and although gene expression in primary fibroblasts does not reveal how MTDH is expressed in nervous tissue, we were encouraged to be able to confirm the initial finding that associated the T allele of rs1835740 with increased MTDH expression in immortalized lymphocytes [21, 54]. There was no difference in mRNA levels between CH patients and controls.

The discovery of rs1835740 and MTDH provided a molecular link between migraine and familial hemiplegic migraine (FHM), a familial form of migraine caused by mutations in ion channels, since mutations associated with FHM subtypes one and two are known to cause increased levels of glutamate at the synaptic cleft [55, 56]. Similarly, our results suggest that MTDH and glutamatergic mechanisms may be involved also in CH pathophysiology. There are previous indications that glutamate might play a role in CH as well. A recent publication demonstrated a decrease in kynurenine metabolites, known glutamate receptor antagonists, in CH patients [57]. This could potentially affect the activity of the glutamate receptor NMDA and nociceptive sensitization [57, 58].

We conclude that rs1835740 in MTDH is associated not only with migraine but also with CH, whilst rs2651899 in PRDM16 seems to be specifically related to migraine in Sweden. In view of our stratified analysis showing that the T allele of rs1835740 was more common in patients with both disorders, we also suggest that rs1835740 might constitute a genetic risk factor for more complex headache phenotypes, for example a combination of different primary headache syndromes as illustrated in our study. The T allele of rs1835740 was further shown to affect MTDH mRNA expression in primary fibroblasts. In depth analyses of MTDH and the rs1835740 SNP in CH and in other primary headache disorders would be of interest in order to determine to what extent rs1835740 is related to severe headache.