Age-related macular degeneration (AMD), cardiovascular disease and stroke

Abstract

Purpose : The relationship between age-related macular degeneration (AMD), cardiovascular disease (CVD) and stroke has been long sought, with decades of inconsistent results. We demonstrate new, specific strong associations by first separating AMD into its two major intermediate phenotypes, soft drusen (SD) and subretinal drusenoid deposits (SDD), and comparing their associations with certain common high risk vascular diseases.

Methods : Two hundred AMD patients, ages 51 to 100, 121 females were recruited. Retinal imaging, health history questionnaires, and lipid profiles were obtained. Patients were assigned by volume spectral-domain optical coherence tomography (SD-OCT) into two groups, SDD (with or without SD) and SD (SD only). Fundus autofluorescence (AF) and near-infrared reflectance (NIR) imaging were used to confirm the presence of SDD. Self-reported health history questionnaires further classified patients into those with and without these high risk vascular disorders (HRV): 1) cardiac valve defect; 2) cardiac pump defect (myocardial infarction, congestive heart failure, etc); 3) carotid stroke or transient ischemic attack (TIA) with carotid stenosis. Chi-square testing was used for categorical variables. ‘Waikato 222 Environment for Knowledge Analysis (WEKA) Version 3.8.5’ was used to build a multivariate risk model for presence of HRV as a function of SDD status and other covariates.

Results : The prevalence of HRV was 41.24% (40/97) and 6.80% (7/103) in the SDD and SD groups respectively (p =0.000000009). Differences in the prevalence of HRV sub-types in these groups were also significant: valve defects 14/97 in SDD vs. 3/103 in SD (p = 0.0035), pump defects 16/97 vs. 3/103 (p = 0.0011), stroke/TIA 10/97 vs.1/103 (p = 0.0038). A serum HDL < 62 mg/dL was a strong predictor in our total AMD cohort of HRV (p = 0.0046). Data modeling showed that the joint presence of SDD and an HDL<62 predicted the presence of HRV with an accuracy of 85%, 95% CI, 77.5 %-90.7 %.

Conclusions : HRV were significantly more prevalent in subjects with SDD than those with SD only. This suggests that SDD specifically are a biomarker of these life-threatening systemic vasculopathies, all of which are likely to compromise ocular perfusion. This in turn suggests a direct vascular mechanism for SDD that merits investigation. Risk modeling shows a potential for saving lives by detecting occult HRV with non-invasive SD-OCT imaging and serum risk factors.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.