Predicting Diabetic Vitreous Hemorrhage Using Neovascularization Features on Widefield Swept-Source Optical Coherence Tomography Angiography (WF SS-OCTA)

Abstract

Purpose : Risk factors for the development of vitreous hemorrhage (VH) in patients with proliferative diabetic retinopathy (PDR) are not completely understood. We performed a prospective, observational clinical study to investigate the association between neovascularization features on WF SS-OCTA and the development of VH in eyes with PDR.

Methods : Patients with type 1 or type 2 diabetes mellitus and PDR without VH at baseline were included. All patients were imaged with WF SS-OCTA (Montage 15mm×15mm and HD-51 Line scan) at baseline. Images were independently evaluated by two graders for WF SS-OCTA metrics defined a priori. Mixed effects logistic regression models (outcome: occurrence of VH) and mixed effects Cox proportional-hazards models (outcome: time to occurrence of VH) were used for statistical analyses.

Results : Fifty-five eyes of 45 subjects were included. Over a median follow-up of 280 days (range: 28-534 days), 7 of 55 PDR eyes (12.73%) developed VH during the follow-up period. Presence of forward neovascularization (NV), defined as NV that traverse the posterior hyaloid face into the vitreous (odds ratio [OR]=2.56, P=0.007), larger flow area of NV (OR=1.36, P=0.039), and intraocular pressure greater than 21 mmHg (OR=31.20, P<0.001) were significantly associated with the occurrence of VH. Similarly, number of forward NV (hazard ratio [HR]=2.71 per 1 forward NV increase, P=0.01) and larger flow area of NV (HR=1.39 per mm² increase, P=0.02) were associated with time to occurrence of VH.

Conclusions : The number of forward NVs and NV flow area as measured by WF SS-OCTA were associated with the development of diabetic vitreous hemorrhage. Larger sample sizes with a greater duration of follow-up along with an examination of systemic factors and ophthalmic interventions (e.g. PRP, anti-VEGF) are needed to validate imaging biomarkers for the prediction of diabetic VH.

This is a 2021 ARVO Annual Meeting abstract.