Abstract
Purpose:
Age of spectacle wear (AOSW) has recently been used as a proxy for myopia severity in a successful genome-wide association study (GWAS) of 43,000 people from the 23andMe personal genomics company. This contradicts our expectation that careful phenotyping of refractive error is needed for genetic studies. Given most myopia heritability is currently unexplained, genetic effect sizes are small, and large population studies may not have refractive error data but could ask AOSW, these data challenge future research strategies. We set out to examine the reliability of using AOSW as a proxy for myopia severity using a large British twin cohort with autorefraction, and to examine how well AOSW identifies polymorphisms associated with myopia compared to spherical equivalent.
Methods:
Volunteers from the TwinsUK cohort were asked AOSW via questionnaires in 2003 and 2008, and non-cycloplegic autorefraction was performed on over 6,000 twins between 1998 and 2010, with the refractive error defined as the mean spherical equivalent (SEq) of both eyes. Myopia was defined as <= -0.5 D. Separate analyses were performed on 120 SNPs from 16 loci, from the 23andMe GWAS for AOSW and SEq, to compare significance levels. 2173 participants, who reported AOSW before 33 years were included in the analysis, using Merlin to account for the family structure.
Results:
Data on AOSW and refraction was available for 4280 twins, at a mean age of 54 years (SD 11.42, range 18 to 83). 1782 subjects (42%) were myopic with a median AOSW of 16 years (mean 21.0 years, SD 12.0, range 1 to 74) and a mean refractive error of -3.1 D (SD 2.54, range -25.13 to -0.50). AOSW for the myopes and SEq were correlated (pearson correlation coefficient 0.49, p < 0.0001), equating to AOSW explaining approximately 24% of the variance of SEq. SNPs were generally more significantly associated with SEq than AOSW (for example, rs745480 in RGR associated with SEq p = 0.0001 vs AOSW p = 0.846, rs524952 in the GJD2/GOLGA8B locus associated with SEq p = 0.0001 vs AOSW p = 0.994).
Conclusions:
AOSW reflects severity of myopia but only explains approximately 24% of SEq variance. Associations with SNPs previously identified with myopia were generally of greater significance when using SEq rather than AOSW. However, these data suggest that AOSW may be useful in large-scale population studies where measurement of refractive error is not possible.
Keywords: 605 myopia •
539 genetics