Abstract
Purpose :
Retinal vascular caliber is a potential biomarker of numerous ophthalmic and non-ophthalmic diseases. However, normative values obtained from population studies might not be directly comparable, due to the varying wavelength and optical properties of different instrumentation. This study aimed to test the hypothesis that the choice of imaging technique is a major determinant of vessel diameter estimation.
Methods :
71 eyes from 71 ophthalmically healthy individuals (50-65y) were imaged twice with an SLO (Optos 200Tx) and twice with a FC (TRC-NW400), after pupil dilation. Central retinal artery and vein equivalents (CRAE, CRVE) and the arterio-venous ratio (AVR) were measured with the Knudtson-Parr-Hubbard algorithm and the Automated Retinal Image Analyzer (ARIA, Peter Bankhead). Bland-Altman analysis was performed; intra-device coefficients of repeatability (CR) and inter-device limits of agreement (LoA) were reported. Two-way mixed intraclass correlation coefficients (ICCs) for absolute agreement were also calculated. Paired t-tests were performed to assess discrepancies between the SLO and FC measurements.
Results :
Average (SD) CRAE, CRVE, and AVR values for the SLO were 158 (16) μm, 235 (19) μm, and 0.68 (0.06), respectively, versus 163 (15) μm, 230 (17) μm, and 0.71 (0.06) for the FC. CR for the intra-device SLO measurements were 9 μm (CRAE), 12 μm (CRVE), and 0.04 (AVR), while they were 10 μm, 11 μm, and 0.05 for the FC. LoA for the inter-device comparison were -20 μm to 12 μm (CRAE), -12 μm to 21 μm (CRVE), and -0.08 to 0.02 (AVR). Intra-device ICCs ranged between 0.93 and 0.96 (single measures) and between 0.97 and 0.98 (average measures), for both instruments; inter-device ICCs were 0.85, 0.87, and 0.79 (single measures) and 0.92, 0.93, and 0.88 (average measures) for CRAE, CRVE, and AVR, respectively. The lower CRAE, AVR and higher CRVE values reported by the SLO were highly significant (P<0.0001).
Conclusions :
These findings indicate that estimation of vascular caliber depends on the imaging modality. Compared to the FC, the SLO underestimates arteries and overestimates veins by 5 μm. These discrepancies could be possibly attributed to varying penetrating levels or to the hemoglobin absorption of different wavelengths. Hence, results among different studies should be interpreted with caution.
This is a 2020 ARVO Annual Meeting abstract.