Skip to main content

  • AI Retinal Vasculature Analysis May Predict Risk of NTG Progression

    By Lynda Seminara
    Selected by Richard K. Parrish II, MD
    Add to My Bookmarks

    Journal Highlights

    American Journal of Ophthalmology, March 2023

    Download PDF

    The pathogenesis of normal-tension glaucoma (NTG) is unknown, and the condition can be challenging to manage. The mounting evidence in support of a “vascular theory” of glaucoma comes from cross-sectional studies, which limits the utility of findings. Lin et al. used a deep learning system (DLS) to assess the relation­ship between DLS-computed baseline measurements of retinal vessel calibers and the risk of subsequent glaucoma progression in patients with NTG. They found, as hypothesized, that narrow retinal vessel calibers are an independent prognostic factor for NTG progression.

    For this prospective study, the authors included 390 eyes with NTG that received follow-up for at least 24 months. The authors’ previously vali­dated DLS was used to compute central retinal arteriolar equivalent (CRAE) and central retinal venular equivalent (CRVE) from baseline fundus pho­tographs. The visual field (VF) and retinal nerve fiber layer (RNFL) were evaluated semiannually, and the Cox proportional-hazards model was used to explore the potential relationship between baseline retinal calibers and the risk of NTG progression.

    During the mean follow-up period of 34.36 months, progressive RNFL thinning developed in 69 study eyes, while VF deteriorated in 22. Accord­ing to multivariable Cox regression analysis (adjusted for demographics, IOP, ocular perfusion pressure, systolic blood pressure, axial length, standard automated perimetry mean deviation, and RNFL thickness), narrower base­line CRAE and CRVE were each linked to progressive RNFL thinning (hazard ratio [HR] per SD decrease, 1.36 and 1.35, respectively). Narrower CRAE also was associated with VF deterioration (HR, 1.98).

    Each SD decline from baseline in CRAE or CRVE raised the risk of RNFL thinning by more than 30% and the likelihood of VF deterioration by more than 90% during the two years of follow-up. Therefore, “high-throughput deep-learning-based retinal vasculature analysis demonstrated its clinical utility for NTG risk assessment,” said the authors. They added that this research corroborates the vascular theory of glaucoma.

    The original article can be found here.