Genome-wide association study of adolescent idiopathic scoliosis in Southern Chinese and significant improvement in detection of missing heritability and genetic risk prediction in genome-wide association studies by 1000 genomes-based imputation

Abstract

Adolescent idiopathic scoliosis (AIS) is a lateral curvature of the spine affect school age children. Although many studies have been conducted to investigate the candidate loci and genes for AIS, the underlying mechanisms of the pathogenesis is still poorly understood. In this study, a genome-wide association study on a cohort consisting of 189 AIS patients and 313 controls was performed to investigate the genetic variants associated with AIS in the southern Chinese population. The most promising association region is 12p13.32, in which a significant linkage signal was found in previous study. The three most significant SNPs (rs10849195, P = 6.84 × 10-7, OR = 0.51, 95% CI 0.39-0.67; rs11063476, P = 9.82 × 10-6, OR = 0.55, 95% CI 0.42-0.72; rs2109185, P = 3.83 × 10-5, OR = 0.57, 95% CI 0.44-0.75) in this region were in the same linkage disequilibrium (LD) block of about 2 kb. The most significant SNPs rs10849195 on chromosome 12 was successfully replicated in two independent replication studies. One replication study with 1271 individuals consist of 568 cases and 703 controls (P = 0.037, OR = 0.84, 95% CI 0.72-0.99), and the other replication study with 536 individuals consist of 286 cases and 250 controls (P = 0.026, OR = 0.76, 95% CI 0.60-0.97). The combined two stage P-value is 9.1 × 10-7, OR = 0.74.

SNP rs11190870 near LBX1 (ladybird homeobox 1), which was associated with Japanese adolescent idiopathic scoliosis patients by a genome-wide association study, was also validated in an independent southern Chinese population. The results suggested that SNP rs11190870 is statistically significant associated with adolescent idiopathic scoliosis in southern Chinese (P = 9.1 10-10; odds ratio = 1.85; 95% confidence interval = 1.52 - 2.25).

Quality control (QC) by removing individuals or markers with low genotyping quality is of utmost importance to minimize potential false positive associations. IPGWAS was developed to facilitate the identification of the rational thresholds in QC of GWAS datasets, association analysis, Manhattan plot, quantile-quantile (QQ) plot, and format conversion for genetic analyses, such as meta-analysis, genotype phasing, and imputation. IPGWAS is freely available at http://sourceforge.net/projects/ipgwas/.

Missing heritability is a common problem in genome-wide association study. The Wellcome Trust Case Control Consortium Phase I genotype data were imputed using 1000 genomes as reference. Then phenotypic variance explained by SNPs passed certain p-value thresholds was estimated. The results suggested that the proportions of phenotypic variance explained by genetic variants increased significantly when the new association variants identified through 1000 genome based imputation were included for most complex diseases. The obtained results were consistent with the larger number of variants that are yet to be identified as potential sources of missing heritability.

Most GWAS results cannot achieve a reliable genetic risk prediction performance. After genotype imputation, both prediction sensitivity and specificity were increased to a relative reliable level (>0.7). It is promising to transfer the results from basic genetic research to clinical utility.