Abstract
Purpose :
Slow retinal vascularisation is a fundamental feature of retinopathy of prematurity (ROP) but its quantification is challenging using traditional ROP screening methods. Here we measure retinal vascularisation rate derived from ultra-widefield retinal imaging-based ROP screening and explore its predictive power for threshold ROP.
Methods :
Retrospective analysis of serial (≥2 timepoints) Optomap images from babies who underwent ROP screening in Oxford from May 2019 to July 2023. Two independent investigators measured the temporal extent of retinal vascularisation as the ratio between the distance from disc centre-to-temporal vascular front and disc centre-to-fovea along a straight line that bisects the vascular arcades. Measures from two or more timepoints enabled calculation of temporal vascularisation rate (TVR), which serves as a surrogate marker for retinal vascularisation rate. Using TVR, gestational age (GA) and birthweight (BW), a random forest model was developed for predicting eyes that develop no ROP or subthreshold ROP (defined together as Group 0) versus eyes that develop threshold ROP (defined as Group 1). TVR was also compared before and after bevacizumab injection in Group 1 eyes.
Results :
Temporal vascularisation rates were measured from serial Optomap images of 137 eyes of 78 babies with mean GA of 26.0 weeks (±2.0 SD) and BW 815g (±264). TVR was significantly faster (by 1.5X) in Group 0 eyes (n=33 with no ROP and n=50 with subthreshold ROP) compared with Group 1 eyes (n=54) (p=0.03). We built a random forest model incorporating TVR and tested its predictive power for threshold ROP in a separate cohort of 14 eyes (8 babies). The model accurately predicted Group 0 (n=11) versus Group 1 (n=3) eyes in the cohort. Interestingly, comparison of TVR before and after anti-VEGF therapy in Group 1 eyes showed no significant change in the retinal vascularisation rate post-treatment (p=0.18 for right and p=0.48 for left eyes).
Conclusions :
Optomap-based ROP screening allows precise measurement of temporal vascularisation rate, which can serve as a powerful early predictor of threshold ROP. Our observation of ‘constancy’ in the rate of retinal vascularisation despite anti-VEGF therapy suggests selective targeting of extraretinal neovascularisation by the drug and provides novel mechanistic insight into the development of peripheral avascular retina (PAR) in premature infants.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.