Genetics analysis may predict myopia in children

April 15, 2020

Article

Clinicians may soon be able to use genetic information alone to determine the risk of myopia in very young children, researchers believe.

“A personalised medicine approach to detecting children at risk of myopia is now feasible, although currently the accuracy of [polygenic risk scores] is not yet good enough to warrant their use in clinical practice,” wrote Neema Ghorbani Mojarrad and colleagues at the University of Bradford in Bradford, UK. They published their findings in JAMA Ophthalmology.

Genome-wide studies

At present, the best predictor of the risk of myopia in children is a low hyperopic refractive error at an age before myopia typically manifests, so clinicians seeking to evaluate the risk in young children typically rely on a screening regimen of cycloplegic autorefraction. Successfully predicting the likelihood of children developing myopia can help clinicians recommend lifestyle changes such as spending more time outdoors, which can help mitigate or prevent the development of the refractive error.

However, the use of cycloplegic autorefraction is time and resource-intensive, so Mr Mojarrad and his team wanted to see if they could develop a polygenic risk score (PRS) for evaluating risk based purely on genetic information. To develop their PRS, the researchers conducted three genome-wide association studies (GWASs). Data used for the GWASs included 287,448 UK Biobank participants who had not undergone autorefraction, and 95,619 who had undergone noncycloplegic autorefraction.

In addition, researchers collected data on 328,917 individuals whose information had originally been compiled to study the relationship between genes and years of educational attainment, and had been downloaded from the Social Science Genetic Association Consortium (SSGAC) website. They then combined their data in a meta-analysis in order to increase the effective sample size, before generating a PRS for each participant using genotypes for 1.1 million variants.

The investigators validated their PRS by using it to predict rates of myopia in 1,516 women whose genome data and known rates of myopia were obtained from the Avon Longitudinal Study of Parents and Children (ALSPAC).
The PRS derived from the SSGAC GWAS provided limited predictive accuracy (R2 = 0.14%, where R2 = 100% would be a completely accurate prediction). The autorefraction and non-autorefraction GWASs were also relatively inaccurate (R2 = 7.1% versus R2 = 6.9%).

The researchers were able to improve the accuracy of the PRS by combining sets of traits. The best result came from combining the autorefraction and non-autorefraction GWASs (R2 = 10.8%). Combining all three traits yielded even greater accuracy (R2 = 11.2%).

The GWAS was unable to produce a PRS that was as reliable as the results obtainable using cycloplegic autorefraction. Measured in terms of the under the receiver operating characteristic (AUROC), where 0.5 is a useless model and 1 is a perfect model, the PRS had an AUROC of 0.67 while the cycloplegic autorefraction scored 0.87.

Three categories of risk

While they were unable to develop a suitably accurate alternative to cycloplegic autorefraction for predicting rates of myopia, Mr Mojarrad and his team were able to conclude from their study that children can be categorised into three groups based on their risk of developing myopia. Those with a PRS in the top 25% were 3-5 fold more likely to develop the condition, those in the top 10% had a 3.5-6 fold increased risk, whilst the likelyhood was 4.5-6.5 fold higher for those in the top 5%.

The researchers are hopeful that this information will aid clinicians in recommending lifestyle interventions or prescribing orthokeratologic or atropine treatment. They predict that future research may lead to the development of a PRS that will enable doctors to accurately predict myopia in children, obviating the need for cycloplegic autorefraction.