Association Between Retinal Microanatomy in Preterm Infants and 9-Month Visual Acuity

IMPORTANCE Preterm infants are at risk for poor visual acuity (VA) outcomes, even without retinal problems on ophthalmoscopy. Infant retinal microanatomy may provide insight as to potential causes. OBJECTIVE To evaluate the association between preterm infant retinal microanatomy and VA at 9 months' corrected age. DESIGN, SETTING, AND PARTICIPANTS This prospective observational study took place from November 2016 and December 2019 at a single academic medical center and included preterm infants enrolled in Study of Eye Imaging in Preterm Infants (BabySTEPS). Infants were eligible for enrollment in BabySTEPS if they met criteria for retinopathy of prematurity (ROP) screening, were 35 weeks' postmenstrual age or older at the time of first OCT imaging, and a parent or guardian provided written informed consent. Of 118 infants enrolled in BabySTEPS, 61were included in this analysis. Data were analyzed from March to April 2021. EXPOSURES Bedside optical coherence tomography (OCT) imaging at a mean (SD) 39.85 (0.79) weeks' postmenstrual age and monocular grating VA measurement at 9 months' corrected age. MAIN OUTCOMES AND MEASURES Presence and severity of macular edema and presence of ellipsoid zone at the fovea measured by extracting semiautomated thicknesses of inner nuclear layer, inner retina, and total retina at the foveal center; choroid across foveal 1mm; and retinal nerve fiber layer (RNFL) across the papillomacular bundle (PMB). Pearson correlation coefficients were calculated and 95% CIs were bootstrapped for the association between retinal layer thicknesses and continuous logMAR VA. Associations were analyzed between retinal microanatomy and normal (3.70 cycles/degree or greater) vs subnormal grating VA at 9 months' corrected age using logistic regression and with logMAR VA using linear regression, adjusting for birth weight, gestational age, and ROP severity at the time of OCT imaging and accounting for intereye correlation using generalized estimating equations. RESULTS The mean (SD; range) gestational age of included infants was 27.6 (2.8; 23.0-34.6) weeks, and mean (SD; range) birth weight was 958.2 (293.7; 480-1580) g. In 122 eyes of 61 infants, the correlations between retinal layer thicknesses and logMAR VA were as follows; r, 0.01(95% CI, -0.07 to -0.27) for inner nuclear layer; r, 0.19 (95% CI, 0.01to 0.35) for inner retina; r, 0.15 (95% CI, -0.02 to 0.31) for total retina; r, -0.22 (95% CI, -0.38 to -0.03) for choroid; and r, -0.27 (95% CI, -0.45 to 0.10) for RNFL across the PMB. In multivariable analysis, thinner RNFL across the PMB (regression coefficient, -0.05 per 10-um increase in RNFL thickness; 95% CI, -0.10 to -0.01; P = .046) and prior ROP treatment (regression coefficient, 0.33 for ROP treatment; 95% CI, 0.11to 0.56; P = .003) were independently associated with poorer 9-month logMAR VA. CONCLUSIONS AND RELEVANCE In preterm infants, RNFL thinning across the PMB was associated with poorer 9-month VA, independent of birth weight, gestational age, need for ROP treatment, arid macular microanatomy. Evaluation of RNFL thickness using OCT may help identify preterm infants at risk for poor vision outcomes.