Abstract
Purpose :
The onset and progression of aortic aneurysms are asymptomatic, which eventually becomes a life-threatening event when the vessel wall is dissected or ruptured. The screening and diagnosing of aneurysms are expensive and sometimes invasive. Retinal vascular features (RVFs) are measured easily and non-invasively, they reflect the systemic vascular properties as well as the clinical risk factors for aneurysms, hence they can be the potential biomarkers for aortic aneurysms and dissections. In this study, we used the genetic information of aneurysms as the genotypes and RVFs as phenotypes and conducted a phenome-wide association study (PheWAS), in order to identify novel imaging biomarkers which can assist the early diagnosis of aortic aneurysms.
Methods :
The study was conducted within the UK Biobank participants. Digital retinal fundus images from the right eyes were used. A machine learning method was used to extract the RVFs, which resulted in 91 quantitative features. Genetic risk scores (GRS) for abdominal, thoracic and intracranial aneurysms were generated by PLINK and SNPs representing 4 different Marfan genes (FBN1, COL16A1, LOC728241 andLOC441376 ) were extracted. The R package 'PheWAS' was used and age, sex, blood pressure and BMI were adjusted as covariates.
Results :
A total number of 50380 participants were included. In the PheWAS, after Bonferroni correction, 6 RVFs which represent the vessel caliber, complexity or tortuosity were associated with abdominal aneurysm, and 'artery_calibre' and 'number_of_vascular_tree' (ntree_a) were associated with thoracic (β=0.06, P=4.41e-05) and intracranial (β=-0.09, P=9.92e-05) aneurysms respectively. The 'ntree_a' was shared between abdominal and intracranial aneurysms. For Marfan genes, 'mean_artery_curve_angle' (CurveAngle) was common for all 4 genes (FBN1:β=-0.09, P=4.30e-10; COL16A1:β=-0.07, P=4.28e-08; LOC728241:β=-0.05, P=1.26e-04; LOC441376:β=0.07, P=1.75e-05 ), and 'ntree_a' was shared by 3 genes (FBN1:β=-1.52, P=3.53e-14; COL16A1:β=-0.04, P=2.71e-04; LOC728241:β=-0.04, P=4.27e-04) (Figure 1). Among the overall traits, 'ntree_a' and 'CurveAngle' were the most common RVFs (Figure 2).
Conclusions :
The PheWAS identified several RVFs that are associated with different types of aneurysms. Among them, 'ntree_a' and 'CurveAngle' are most commonly identified by different aneurysms, suggesting they are potential vascular biomarkers for aneurysms.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.