Retinal microvascular phenotypes can track small vessel disease burden and CPAP treatment effectiveness in obstructive sleep apnea

Abstract

Optical coherence tomography angiography (OCT-A) retinal imaging enables in vivo visualization of the retinal microvasculature that is developmentally related to the brain and can offer insight on cerebrovascular health. We investigated retinal phenotypes and neuroimaging markers of small vessel disease (SVD) in individuals with obstructive sleep apnoea (OSA). We enrolled 44 participants (mean age 50.1 ± SD 9.1 years) and performed OCT-A imaging before and after continuous positive airway pressure (CPAP) therapy. Pre-treatment analyses using a generalized estimating equations model adjusted for relevant covariates, revealed perivascular spaces (PVS) volume in basal ganglia associated with greater foveal vessel density (fVD) (p-value < 0.001), and smaller foveal avascular zone area (p-value = 0.01), whereas PVS count in centrum semiovale associated with lower retinal vessel radius (p-value = 0.02) and higher vessel tortuosity (p-value = 0.01). A reduction in retinal vessel radius was also observed with increased OSA severity (p-value = 0.05). Post-treatment analyses showed greater CPAP usage was associated with a decrease in fVD (p-value = 0.02), and increased retinal vessel radius (p-value = 0.01). The findings demonstrate for the first time the potential use of OCT-A to monitor CPAP treatment and its possible impact on both retinal and brain vascular health.

Keywords: CPAP treatment; OCT-angiography; OSA; Retinal imaging; SVD burden.

Figure 1.

OCT-A processing. (a) OCT-A scan; (b) Binary image; (c) Parafoveal ring and foveal circle (blue); (d) Vessels with caliber greater than 20 µm; (e) Vessels with caliber smaller than 20 µm and (f) FAZ detection.

Figure 2.

Standardized betas with 95% confidence interval (CI) of retinal and brain variables using GEE model adjusted for age, sex, systolic blood pressure, self-reported diabetes status, BMI, hypercholesterolemia, smoking status, and image quality. PVS volumes were corrected for the volume of the relevant region of interest. PVS: perivascular spaces; CSO: centrum semiovale; BG: basal ganglia. CI containing the value zero (dashed line) indicates an unadjusted p-value > 0.05, while *: p-value ≤ 0.05, **: p-value ≤ 0.01, ***: p-value ≤ 0.001. VD: vessel density, FAZ: foveal avascular zone, KS: Kolmogorov-Smirnov, L: large vasculature, S: small vessels.

Figure 3.

Standardized coefficients with 95% confidence interval (CI) obtained using the GEE model adjusted for age, sex, SBP, self-reported diabetes status, BMI, hypercholesterolemia, smoking status, and image quality. CI containing the value zero (dashed line) indicates a p-value greater than 0.05. *: p-value ≤0.05, **: p-value ≤ 0.01, ***: p-value ≤0.001. pAHI: WatchPAT derived apnoea-hypopnoea index, SpO2: oxygen saturation; VD: vessel density; FAZ: foveal avascular zone; KS: Kolmogorov-Smirnov; L: large vasculature; S: small vessels.

Figure 4.

(a) Boxplots of the amount of change in FAZ area pre- and post-treatment in the right eye (OD) between participants with optimal and sub-optimal CPAP adherence. (b) FAZ area value pre- and post- treatment according to CPAP average usage (hr/day) and (c) Boxplots of the amount of change in large vessel radius. (d) Radius value in large vessels pre- and post- treatment. Quartile (Q) Q1: lowest CPAP mean usage; Q4: greatest CPAP mean usage.