Instrumenting Carotid Sonography Biomarkers and Polygenic Risk Score As A Novel Screening Approach For Retinal Detachment

Ching Yun Wang

Taichung Veterans General Hospital, Taichung, Taiwan

Abstract :

Background: Retinal detachment (RD) is a blind-threatening disease without effective screen­ing protocols. Thus, current medical practice deployed RD surveillance only after the first at­tack.

Aim: To foster a risk stratification framework attributing RD risk before disease onset, we in­strumented the hemodynamic biomarkers of the carotid ultrasonography (CUS) and the RD polygenic risk score (PRSRD) from single nucleotide polymorphism (SNP) profiling.

Methods: For 21,441 Taiwan biobank participants, a backpropagation logistic regression model was built and visualized by nomogram to identify RD-associated CUS biomarkers. A PRSRD model was built by charting the expression of SNP functional genes from retina scRNA datasets. Last, a two-component RD prediction model (CUS and PRSRD) was assembled by logistic cumulative analysis.

Results: Hypertension (HTN) status was significantly associated with RD risk (OR=1.32). The CUS model (AUCHTN+=0.632, AUCHTN-=0.630) showed RD risk increased with the min­imum flow of the right internal carotid artery and the timed average max velocity of the right common carotid artery (ICA-Qmin OR=1.04 and CCA-TAMAX OR=1.03 with p-val­ue<0.05). The genome-wide association study (GWAS) identified three outstanding RD SNPs (IGFBPL1 rs117248428 OR=1.63, CELF2 rs56168975 OR=1.72, and PAX6 rs11825821 OR=1.61 with p-value<5.00x10-6) with coded genes highly-expressed in retinal pigment epithelium (RPE) and choroid. Notably, the two-component model achieved state-of-the-art prediction (AUCHTN+=0.95, AUCHTN-=0.93).

Conclusions: Through instrumenting CUS images and genetic PRSRD, we proposed a screen­ing method for RD at-risk patients.