A multitrait model of corneal parameters was effective in identifying new keratoconus risk loci, resulting in enhanced polygenic risk score models.
A study published in JAMA Ophthalmology found that a polygenic risk score (PRS) model based on corneal resistance factor (CRF) and central corneal thickness (CCT) was associated with improved accuracy in identifying keratoconus, an ophthamological condition in which the cornea thins and worsens.
The genome-wide association study (GWAS) used the UK Biobank (UKB) and Canadian Longitudinal Study on Aging (CLSA) databases for their participants, with a total of 129,655 individuals whose data were used across the 2 databases. Participants were aged 45 years and older.
There were 4 cohorts for the GWAS: 105,427 UKB Europeans (53.2% female, mean [SD] age of 57  years); 5029 UKB South Asians (47.1% female; mean [SD] age, 54  years); 902 UKB East Asians (68.9% female; mean [SD] age, 53  years); and 18,307 CLSA Europeans (50.6% female; mean [SD] age, 63  years).
A multitrait analysis of GWAS (MTAG) was used to combine the CRF and CCT GWASs in European participants, where a strong genetic correlation (0.69 [SE, 0.049]) was observed between CCT and CRF input data sets. There were 369 genome-wide significant independent variants for CRF and 233 for CCT; after excluding previously reported loci, 36 CRF variants and 114 CCT variants were identified as novel loci.
The effect sizes between European and Asian participants were compared to assess the cross-ethnic effects of CRF and CCT hits. The genetic correlation between CRF and CCT summary statistics was 0.40 (SE, 0.43); good concordance in effect sizes across ancestries (R for CRF, 0.70; 95% CI, 0.53-0.82; R for CCT, 0.76; 95% CI, 0.58-0.87]) was observed.
The genetic correlation between keratoconus and CRF was –0.52 (SE, 0.04) and between keratoconus and CCT was –0.36 (SE, 0.06); there was a negative correlation when comparing the effect size of CRF and CCT variants and keratoconus GWAS. The researchers identified 354 CRF variants and 221 CCT variants that were available in the keratoconus GWAS, with 24 novel keratoconus variants identified from CCT variants.
They also discovered 213 CRF and 134 CCT loci that were near genes associated with rare cornea or connective tissue diseases, with several loci harbored within 1 Mb of mendelian genes associated with rare corneal or connective tissue diseases.
A PRS was constructed to test the predictive ability of the PRS in 102 cases with keratoconus and 142,493 controls. The researchers found that the area under the curve (AUC) of the PRS for the CRF variant was 0.720 and the AUC of the PRS for the CCT variant was 0.715.
The predictive ability of 36 lead variants from a prior study was tested to assess whether CRF and CCT variants would help improve the prediction of keratoconus. The AUC of the PRS for 36 keratoconus variants was 0.705 but the AUC improved when adding the CRF-based PRS (AUC, 0.756 for the PRS CRF variants plus the PRS 36 keratoconus variants). A significant AUC improvement of PRS (AUC, 0.755) was observed from the PRS CCT variants plus the PRS 36 keratoconus variants.
There were some limitations to this study. There was incomplete replication in the Asian population, likely attributed to low power. Larger studies are needed to accurately characterize the cross-ethnic effect sizes for other variants. There was also an overlap between the UKB data set used to test the PRS and the previously published keratoconus meta-analysis used to identify the 7 discordant variants, which could bias the identification of the 7 variants.
The researchers concluded that this study identified the variants and genes associated with CRF and CCT, which enhances understanding of the biological mechanisms underpinning CRF, CCT, and keratoconus.
“These findings underscore the power of multitrait GWAS in identifying new disease-related variants using quantitative traits,” the authors wrote.
He W, Han X, Ong JS, et al. Association of novel loci with keratoconus susceptibility in a multitrait genome-wide association study of the UK biobank database and Canadian longitudinal study on aging. JAMA Ophthalmol. Published online April 21, 2022. doi:10.1001/jamaophthalmol.2022.0891