Genetic analysis reveal significant and consistent pairwise interactions between common non-coding LOXL1 polymorphisms.

Abstract

Purpose : Common genetic polymorphisms within LOXL1 show the strongest evidence of association to date. However, the associations with these common variants (marked by rs1048661 encoding for p. R141L and rs3825942 encoding for p.G153D) are reversed in Japanese (for rs1048661) and Africans (for rs3825942). To this end, we suspected there could be non-additive interactions between genetic variants which may explain some of the allele reversal.

Methods : We performed genome-wide interaction analysis using GWAS data, by assessing for interaction between classical common LOXL1 variants and other loci genome-wide. As we also have deep re-sequencing data of the LOXL1 locus, we performed within-locus LOXL1 interaction analysis using deep sequencing data. This exercise was performed in representative ethnic groups (East Asians, represented by Japanese, South Africans, and Russians) where the reversal is clearly documented. The sample size for genome-wide interaction analysis was 3,061 cases and 2,968 controls from Japan, 387 cases and 674 controls from Russia, and 109 cases and 269 controls from South Africa. For the within-locus LOXL1 interaction analysis, a total of 2,827 cases and 3,013 controls from Japan, 476 cases and 859 controls from Russia, and 95 cases and 250 controls from South Africa.

Results : No evidence of association surpassing multiple-testing thresholds were observed in interactive tests between classical LOXL1 common variants and other loci in a genome-wide search using GWAS data. However, using targeted LOXL1 deep re-sequencing data, we observed genotypic combinations with pairs of LOXL1 common intronic variants that were consistently enriched in controls as compared to cases.

Conclusions : Our results suggests that at least part of the LOXL1 allele reversal phenomenon between Japanese, Europeans, and Africans could be explained by non-additive interactive effects. Further effort to study the biological underpinnings of these consistent associations are warranted, and to exclude statistical artifacts.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.