Abstract
Purpose :
The purpose of this study was to evaluate the association between retinal microanatomy at 36 weeks postmenstrual age and Teller card grating visual acuity (VA) at 9 months in preterm infants.
Methods :
For each infant and for each eye, we analyzed optical coherence tomography (OCT) images from one bedside imaging session captured at 35-37 weeks postmenstrual age. We also analyzed retinopathy of prematurity (ROP) stage (the same day as OCT imaging), infant demographics, and Teller card grating VA at 9 months corrected age. We captured all measures prospectively from preterm infants enrolled in the STudy of Eye imaging in Preterm infantS (BabySTEPS; NCT02887157). Expert graders, masked to other study data except for age, extracted OCT microanatomy features (e.g., presence/severity of macular edema) and retinal thicknesses, including: inner layer complex (retinal nerve fiber layer [RNFL], ganglion cell, and inner plexiform layer) thickness at the foveal center; inner nuclear layer thickness at the foveal center; total retina thickness at the foveal center; choroid thickness averaged across the foveal 1mm; and RNFL thickness at the papillomacular bundle (PMB). We analyzed for associations of OCT retinal microanatomy features with categorized VA (normal vs. subnormal, using a threshold of 3.70 cycles/degree) using logistic regression models and with continuous logMAR VA using linear regression models. We accounted for inter-eye correlation by using generalized estimating equations.
Results :
Of the 130 eyes of 65 infants with both OCT and VA data, 71 (55%) eyes of 40 infants had subnormal VA (<3.70 cycles/degree), and 12 (9%) eyes of 6 infants had prior ROP treatment. On univariable linear regression, low gestational age, low birthweight, prior ROP treatment, and RNFL thinning at the PMB were associated with higher logMAR. On multivariable linear regression, ROP treatment (p=0.006) and RNFL thinning at the PMB (p=0.007) were independently associated with higher logMAR. On logistic regression, only prior ROP treatment (p=0.021) was associated with subnormal VA, though RNFL thinning at the PMB (p=0.097) approached significance.
Conclusions :
RNFL thinning at the PMB is associated with poorer 9-month VA in preterm infants, independent of ROP severity and demographic features.
This is a 2021 ARVO Annual Meeting abstract.