Genome-wide association study of recurrent early-onset major depressive disorder

Abstract

A genome-wide association study was carried out in 1020 case subjects with recurrent early-onset major depressive disorder (MDD) (onset before age 31) and 1636 control subjects screened to exclude lifetime MDD. Subjects were genotyped with the Affymetrix 6.0 platform. After extensive quality control procedures, 671 424 autosomal single nucleotide polymorphisms (SNPs) and 25 068 X chromosome SNPs with minor allele frequency greater than 1% were available for analysis. An additional 1 892 186 HapMap II SNPs were analyzed based on imputed genotypic data. Single-SNP logistic regression trend tests were computed, with correction for ancestry-informative principal component scores. No genome-wide significant evidence for association was observed, assuming that nominal P<5 × 10(-8) approximates a 5% genome-wide significance threshold. The strongest evidence for association was observed on chromosome 18q22.1 (rs17077540, P=1.83 × 10(-7)) in a region that has produced some evidence for linkage to bipolar-I or -II disorder in several studies, within an mRNA detected in human brain tissue (BC053410) and approximately 75 kb upstream of DSEL. Comparing these results with those of a meta-analysis of three MDD GWAS data sets reported in a companion article, we note that among the strongest signals observed in the GenRED sample, the meta-analysis provided the greatest support (although not at a genome-wide significant level) for association of MDD to SNPs within SP4, a brain-specific transcription factor. Larger samples will be required to confirm the hypothesis of association between MDD (and particularly the recurrent early-onset subtype) and common SNPs.

Figure 1. Genome-wide results

Panel A shows the association test result for each SNP (genotyped and imputed) as -log₁₀(P-value), for the primary analysis of all subjects. The largest symbols represent P<10⁻⁶, and the intermediate-size symbols represent P<10⁻⁵. For the same SNPs, Panel B shows the quantile-quantile plot of observed vs. expected Χ² (1df) statistics. The genomic control λ value was 1.032. Manhattan and QQ plots are shown for males and for females separately in online Figures S7 and S14.

Figure 2. Results for selected chromosomal regions

Shown from top to bottom of each plot are chromosome ideogram (plotted region marked with vertical red bar); genomic information (RefSeq genes with direction of transcription; and in Panel A, mRNAs); association test results (−log₁₀[P]) in the primary analysis of all subjects; and HapMap linkage disequilibrium (r²) information.