Mixed Effects Models Indicate Several SNPs within the CNTNAP2 Gene Increase Rate of Glaucoma Related Visual Field Loss

Abstract

Purpose : Interrogate the CNTNAP2 gene’s contribution to rate of visual field loss in patients with glaucoma from individuals of African descent (AD) and European descent (ER).

Methods : Previous meta-GWAS analysis on 1,120 glaucoma patients in two different ancestries (ER: 557, AD: 563) highlighted the CNTNAP2 gene as potential contributor to the rate of visual field (VF) loss. This gene has already been associated with pseudoexfoliation syndrome, but not previously associated with glaucoma. Using whole-genome sequence data from these same individuals and clinical longitudinal covariates over a period of 2 – 19 years, mixed effects models (MEM) were used to supply additional evidence for the effect of SNPs within the CNTNAP2 gene. Models were built for suspected SNPs identified in the previous analyses, across 14 different visual field metrics. The power to detect signal is boosted by including time series data. MEM allow for within subject correlation between eyes, and adjustments for serial correlation. In addition to MEM, combined SNP effects were addressed using polygenic risk scores. Because most SNPs fell within the intergenic regions of the CNTNAP2 gene, potential effects were evaluated using various deleterious scoring metrics (CADD, PolyPhen2) and their predicted involvement in promotor and enhancer regions.

Results : Probabilistic comparison of the fit of MEM with and without SNP effect terms was used to determine the significance of each SNP on the rate of visual field loss. In the ER population of the 40 SNPs tested 38 were below P=5.5e-4. Six of these SNPs were previously identified enhancer / promotor regions for CNTNAP2. In the AD population only 9 of the 35 SNPs tested were statistically significant, and 2 of these 9 were within enhancer / promotor regions. Among the covariates included in the model, age, and sex were significant. IOP, medication, and procedures have yet to be included. These models were able to include 4.7 times as many data points for each SNP compared to previous GWAS analysis.

Conclusions : Previous analysis provided suggestive evidence that CNTNAP2 may play an influential role in the rate of decline of visual fields for patients with glaucoma. Our analysis provides additional support by including longitudinal modeling, a greater number of data points, important covariate effects, and greater biological annotation and evidence than previous studies.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.