Abstract
Migraine, and particularly migraine with aura (MA), has a strong genetic component. The best evidence for this are the significantly increased familial risk of migraine, the significantly higher concordance rate of migraine in monozygotic than in dizygotic twin pairs (1,2) and the three identified genes, CACNA1A, ATP1A2 and SCNA1, that can cause familial hemiplegic migraine (FHM), a rare monogenic subtype of migraine associated with hemiparesis during aura (3).
Migraine is also a prominent part of the phenotype of several mendelian diseases, including Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL) and other genetic vasculopathies (4). CADASIL is a monogenic form of small-vessel disease of the brain, characterized by recurrent ischemic stroke in 60–85% of patients and cognitive decline in nearly all patients after the age of 50 years. MA is the third cardinal manifestation of the disease, present in up to 40% of patients and often the first clinical manifestation, presenting some 10 years ahead of the other symptoms (5). CADASIL is caused by dominant mutations in the NOTCH3 gene, which encodes a transmembrane receptor primarily expressed in the vasculature and important for the structural and functional integrity of small arteries (6). NOTCH3 has 33 exons but all CADASIL mutations occur in the exons 2–24 encoding the 34 epidermal growth factor–like repeats, with strong clustering in exons 3 and 4, and, lead to an odd number of cysteine residues causing the NOTCH3 protein to abnormally accumulate in the vessel. The dominant view is that the primary determinant of the CADASIL disease involves novel pathogenic roles for mutant Notch3 receptor rather than compromised NOTCH3 function (5). Genetically modified mice that recapitulate the pre-symptomatic stage of the CADASIL disease have been recently obtained. Of interest, recent work showed that transgenic mice expressing a Notch3 protein with an archetypal CADASIL mutation (Arg90Cys) in smooth muscle cells have a lower threshold for evoking cortical spreading depression (CSD) than wild-type mice, consistent with a MA phenotype (7).
Menon et al. (8) analyzed the NOTCH3 gene for its involvement with common migraine. They conducted an association study, which tests for significant differences in allele frequencies between case and controls, with two common polymorphisms located in exons 3 and 4 of NOTCH3, in a community-based setting. They elegantly demonstrate that rs1043994 (684A > G), located in exon 4 of the gene, is associated with an increased risk of developing MA. As is now a requirement for publication of genetic association studies, in order to distinguish true findings from spurious associations, Menon et al. have used a pre-planned replication sample to confirm or refute findings in the discovery cohort. While the association with rs1043994 was replicated, the second polymorphism (rs3815188, [381C > T], in exon 3) that was associated with migraine in the discovery cohort could not be confirmed in the second sample. Interestingly, rs1043994 increased the risk of MA but not of migraine without aura (MO), similarly to the CADASIL mutation in NOTCH3 (5). This is in line with heritability studies suggesting that MA and MO may have at least partly distinct genetic backgrounds (3).
Of note, however, both the discovery and the replication cohort had a limited sample size, with 260 cases or less, and the power to detect an association with both polymorphisms, which have a minor allele frequency between 8% and 12%, was relatively low. Indeed, as for other complex genetic diseases, the odds ratio (OR) of developing migraine associated with individual genetic variants is expected to be below 1.5 (9). If the frequency of the allele being tested is 12%, to detect an OR of 1.5 with 80% power would require over 300 cases and control subjects at p < .05, and this number rises dramatically if the allele frequency or the OR drops (10). Thus, studies on larger sample sizes are needed to formally exclude an association with rs3815188.
The association of a common variant in NOTCH3 with migraine in community persons is intriguing, and raises a number of questions. The risk variant rs1043994 is a synonymous polymorphism, that is, does not lead to an amino acid change. Although some functionality via transcriptional regulation cannot be formally excluded, this polymorphism is probably merely a marker in linkage disequilibrium with a neighboring functional variant. Thus, more comprehensive analysis of the genetic variation within and in close proximity of the NOTCH3 gene, on a larger sample, is required to look for a possible causal variant. Also, could the association of migraine with rs1043994 be a synthetic association reflecting the effect of one or multiple rare variants in that region, which could include CADASIL-type mutations (11)? Or does another type of qualitative or quantitative modification in the expression of the NOTCH3 gene, different from the modification characteristic of CADASIL (i.e. loss or gain of a cysteine residue [12]), explain the increased risk of migraine in carriers of the rs1043994 variant? From a clinical point of view, it might be interesting to look for an interaction between rs1043994 and CADASIL mutations, by testing whether rs1043994 has a modifying effect on the migraine phenotype in CADASIL patients. There is some evidence that genetic variants distinct from the disease-causing mutation in NOTCH3 influence the white-matter lesion load in CADASIL patients (13), and this could well be the case also for migraine. From a fundamental standpoint, further studies are required to understand the mechanistic basis linking Notch3 to the CSD phenotype.
Taking a broader perspective, identifying the genetic determinants of the common forms of migraine remains yet a great and difficult challenge. The original approach presented by Menon et al. is an incentive to explore further the links between mendelian and complex forms of migraine. Whereas monogenic forms of migraine are classically considered as entirely distinct from multifactorial migraine, the findings of Menon et al. (8) and evidence from other diseases (14) suggest that they may just be the extremes of the same continuum. Some of the genes involved in the complex or multifactorial form probably overlap with the genes leading to mendelian forms, as suggested by the data presented (8) and as previously demonstrated with other phenotypes such as lipid levels or neurodegenerative disorders (14). Expanding on this concept, studies looking at the association of migraine with genetic variants in ion transport genes involved in FHM at the community level could also be of interest, although recent attempts have been rather disappointing (15).
Further insight can be gained in the near future from hypothesis-free genome-wide association studies, (GWAS), a powerful and unbiased tool for identifying genes related to complex diseases. The first published GWAS of migraine recently revealed a novel genetic susceptibility variant on chromosome 8q22.1, between the genes encoding astrocyte elevated gene 11 and plasma glutamate carboxypeptidase (16). Additional susceptibility variants will probably be detected by more GWAS on increasingly large samples, and perhaps also through large-scale deep sequencing, which could reveal rare susceptibility variants with large effect size. These new variants could then be tested for a modifying effect on the migraine phenotype in mendelian forms of the disease. Thus, the journey toward bridging the gap between mendelian and complex causes of migraine is just beginning. The expected impact of this type of research is currently expected to be mostly an improved understanding of the pathophysiology underlying migraine, thus indirectly opening new avenues for treatments, via the identification of novel targets or possibly the tailoring of treatments to genetic background.