The Estrogen Hypothesis of Schizophrenia Implicates Glucose Metabolism: Association Study in Three Independent Samples

Background: Schizophrenia is a highly heritable complex psychiatric disorder with an underlying pathophysiology that is still not well understood. Metaanalyses of schizophrenia linkage studies indicate numerous but rather large disease-associated genomic regions, whereas accumulating gene- and protein expression studies have indicated an equally large set of candidate genes that only partially overlap linkage genes. A thorough assessment, beyond the resolution of current GWA studies, of the disease risk conferred by the numerous schizophrenia candidate genes is a daunting and presently not feasible task. We undertook these challenges by using an established clinical paradigm, the estrogen hypothesis of schizophrenia, as the criterion to select candidates among the numerous genes experimentally implicated in schizophrenia. Bioinformatic tools were used to build and priorities the signaling networks implicated by the candidate genes resulting from the estrogen selection. We identified ten candidate genes using this approach that are all active in glucose metabolism and particularly in the glycolysis. Thus, we tested the hypothesis that variants of the glycolytic genes are associated with schizophrenia or at least with gender-associated aspects of the illness.

Results: We genotyped 185 SNPs in three independent case-control samples of Scandinavian origin (a total of 765 patients and 1274 control subjects). Variants of the mitogen-activated protein kinase 14 gene (MAPK14) and the phosphoenolpyruvate carboxykinase 1 (PCK1) and fructose-1,6-biphosphatase (FBP1) were nominal significantly associated with schizophrenia, and several haplotypes within enolase 2 gene (ENO2) consist of the same SNP allele having elevated risk of schizophrenia. Importantly, we find no evidence of stratification due to nationality or gender.

Conclusion: Several gene variants in the Glycolysis were associated with schizophrenia in three independent samples. However, the findings are weak and not resistant to correction for multiple testing, which may indicate that they are either spurious or may relate to a particular subtype or aspect of the illness.