Highlights
-
•
Anti–vascular endothelial growth factor for type 1 retinopathy of prematurity increased nasal and temporal peripheral vascularization.
-
•
Anti-vascular endothelial growth factor–treated eyes had marked nasal-temporal disparity in vascular extension.
-
•
Our protocol to analyze vascular extension is adaptable to other imaging modalities.
PURPOSE
Anti–vascular endothelial growth factor (anti-VEGF) therapy is an important treatment for type 1 retinopathy of prematurity (T1ROP). We previously found that regulation of VEGF signaling increased developmental vascularization into the temporal peripheral retina while reducing intravitreal neovascularization in T1ROP. To address our hypothesis that extension of retinal vascularization by early treatment with anti-VEGF could reduce T1ROP, we determined the effects of anti-VEGF therapy on retinal vascular extension in both retinal hemispheres.
DESIGN
Retrospective, masked, nonrandomized, matched, comparative case series.
SUBJECTS
Over 3 years, 34 of 382 screened infants developed bilateral T1ROP. Consent was obtained from parents of 21 infants for infants receiving bevacizumab 0.25 mg OU. Seven treated eyes (7 infants) had suitable imaging and were included in the treatment group. Seven untreated eyes (7 infants) without T1ROP were randomly selected after matching for birthweight and postmenstrual age from the 382 screened infants and assigned to the control group.
METHODS
Infants had gradable 100° field-of-view fundus photographs of both retinal hemispheres at 2 time points: baseline before treatment, if administered, and follow-up, 12 to 35 days later. No eyes with vascularization to the ora at baseline or follow-up were included. Retinal vascular extent was measured in pixels from the optic disc margin to the temporal or nasal vascular borders and divided by the optic disc diameter in pixels to create disc diameter units (DD). Measurements were confirmed by a second masked analyst. The Mann-Whitney U test and Wilcoxon signed-rank test were used for statistical analysis.
MAIN OUTCOME MEASURES
Temporal and nasal vascular extension in DD.
RESULTS
Relative to control eyes, treated eyes had greater nasal (1.27 ± 0.59 vs 0.55 ± 0.44 DD; P =.038) and temporal (1.69 ± 1.16 vs 0.62 ± 0.49 DD; P =.04) extensions. There was greater disparity in vascular extension between nasal and temporal hemispheres in the treated group than in the control group (0.98 ± 0.64 vs 0.10 ± 0.10 DD; P =.005). Nasal extension was greater than temporal extension in 3 of 7 treated vs 1 of 7 control eyes.
CONCLUSIONS
Treated eyes had significantly greater vascular extension in both hemispheres than control eyes, with greater nasal-temporal asymmetry. Although limited by sample size, our findings raise the question of uneven anti-VEGF diffusion in the infant vitreous. Further studies testing injection site, needle length, anti-VEGF agent, and molecular size differences and further follow-up periods are indicated.
INTRODUCTION
R etinopathy of prematurity (ROP) remains a leading cause of vision loss worldwide. Treatment of type 1 ROP (T1ROP) with laser therapy to the peripheral avascular retina or with agents that interfere with the bioactivity of the angiogenic factor, vascular endothelial growth factor (VEGF), has greatly reduced blindness, which was first reported in the 1940s before oxygen regulation and management. Anti-VEGF agents are becoming the first-line treatment for aggressive ROP and T1ROP in zone I or posterior zone II, in which retinal vascular development has advanced beyond the posterior pole but not to the equatorial retina. , Reduced vascular activity, noted by regression of stage 3 ROP and lessening of plus disease, can be apparent within days of treatment. The use of anti-VEGF agents is also being adopted as a first-line treatment in Germany and with increasing trends worldwide, , including in regions with limited numbers of ophthalmologists available to provide laser treatment.
Our laboratory has been pursuing the question of preventing the proliferative stage of ROP, or phase II of the original hypothesis posed in the 1950s, by promoting retinal vascular development. We used a relevant animal model of oxygen-induced retinopathy in rats that allowed us to test the outcome of peripheral retinal vascular development, measured by the area of peripheral avascular retina. We tested the effect of oxidative stress, nutrition, oxygen, and signaling effectors, including VEGF. , On the basis of our studies, including those with pharmaceutical neutralizing antibodies and small molecules, we developed a novel gene therapy approach using cell-specific promoters to target endothelial cells in the rat retina to knock down, but not eliminate, various messenger RNAs. We surprisingly found that of the 6 compounds we tested for knockdown of messenger RNAs, only knockdown of VEGF receptor 2 in retinal endothelial cells inhibited intravitreal neovascularization and extended peripheral retinal vascularization. ,,,,, To expand to human infants, we looked retrospectively at a cohort of infants screened for ROP in Utah and measured vascular extension temporally in those who developed T1ROP and consented to bilateral anti-VEGF treatment with bevacizumab (0.25 mg). We found that compared with a control prematurity-matched group of infants with less severe ROP than T1ROP, temporal vascular extension to the ora serrata was greater in infants treated with anti-VEGF. This evidence encouraged us to pursue the line of inquiry into whether anti-VEGF might be considered in the future to extend vascularization at least into zone II and reduce the risk of severe ROP.
Given the limited analyses of retinal vascular development and its extension nasally, we investigated the effects of anti-VEGF therapy on the nasal retina compared with the temporal retina. We also addressed whether the measurement of vascular extension was affected by magnification errors when obtaining retinal images. Other errors can potentially occur when multiple measurements are obtained across different fields of view (FOVs), a method that is necessary to measure the full temporal vascular extension but not nasal extension. We were also interested in whether the outcomes could be replicated using a method that would normalize the measurements to the individual diameter of the optic disc, which would allow measurement in studies that used different imaging modalities than the camera used in our study.
We, therefore, evaluated nasal and temporal vascular extent and extension using disc diameter units (DD) in a prematurity-matched treatment and control cohort and evaluated whether within-eye nasal-temporal extension patterns differed between groups.
METHODS
Institutional review board approval was obtained, and the study adhered to the Declaration of Helsinki. All premature infants screened for ROP (<30 weeks’ gestational age, <1500 g birthweight) at the University of Utah UHealth Hospital and Primary Children’s Hospital between January 1, 2019, and February 1, 2022, were eligible for inclusion. Data and imaging were obtained through the University of Utah electronic medical records.
Screening and follow-up examinations were performed according to the American Association for Pediatric Ophthalmology and Strabismus and the American Academy of Ophthalmology recommendations. Infants underwent dilated eye examinations with contact fundus photography and indirect ophthalmoscopy by the same ophthalmic photographer. Infants were diagnosed as having no ROP, incomplete vasculature, type 2 or less severe ROP, or T1ROP according to the Early Treatment for Retinopathy of Prematurity definitions. Confirmation of T1ROP was made by a pediatric vitreoretinal specialist.
A total of 382 infants were screened. Of these, 138 infants without ROP were excluded. Among the remaining 244 infants with ROP, 210 had type 2 ROP or less and were eligible as controls and 34 had T1ROP. Of the 34 with T1ROP, 13 received laser therapy and were excluded from our analyses. The remaining 21 infants received bilateral 0.25 mg intravitreal bevacizumab after informed parental consent was obtained and were initially enrolled in the treatment group. Four were subsequently excluded: 1 due to withdrawn consent and 3 who received treatment at outside facilities, leaving 17 treated infants. A further 10 infants were excluded at the analysis stage due to inadequate imaging, characterized by incomplete or poor-quality images. The final treatment group comprised 7 infants. Seven control infants were then randomly selected from the pool of 210 eligible controls after matching for birthweight and postmenstrual age (PMA) ( Figure 1 ).
Patient selection flowchart. ROP = retinopathy of prematurity.
IMAGING
Fundus images were captured using the same contact fundus camera (NeoLight Phoenix ICON), which has a 100°FOV. Baseline imaging was obtained 1 day to 2 weeks before treatment for the treated infants. Imaging on the day of treatment was avoided to reduce the risk of infection. Follow-up imaging was performed 12 to 35 days after baseline; timing varied to ensure infants were medically stable and to accommodate multidisciplinary care coordination. To account for differences in baseline and follow-up times between groups, treatment and control infants were matched as closely as possible based on PMA.
IMAGE ANALYSIS
All images were deidentified and randomized by the ophthalmic photographer before analysis by the masked analyst. Images of the posterior pole, temporal retina, and nasal retina were analyzed and used to measure vascular extent. Baseline and follow-up images were first assessed to ensure that no baseline or follow-up images were fully vascularized to the ora serrata. Images were assessed for quality and adequate views at baseline and follow-up. One eye of each infant was included in the study because of limitations with image quality that resulted in only 1 analyzable eye.
A step-by-step methodology for obtaining measurements is shown in Figure 2 . All measurements were obtained using the line and ruler tool with pixel units on Adobe Photoshop version 26.11. Using the posterior pole image, the optic disc diameter was measured at the largest point horizontally. Temporally, bifurcations in the superior and inferior arcades were identified and marked between baseline and follow-up images. A line was drawn between the 2 bifurcations and halved. A separate line was drawn from the same temporal edge of the optic disc to the edge of the previously halved line. This line now formed measurement 1 (L1), and the angle created between the 2 lines was noted. Next, using the temporal image, the same bifurcations were identified in the superior and inferior arcades. A line was drawn and then halved. A line was drawn from the edge of the halved line to the temporal vascular/avascular boundary. The angle between the extending line and the halved line was the same as the angle formed in the posterior pole image. The extending line was then placed appropriately to meet the halved line, and the measurement of this extending line was now measurement 2 (L2). L1 and L2 were combined to create the measurement for the temporal extent in pixels. Using the nasal image, a line was drawn between the superior and inferior nasal vessels emerging from the optic disc. An extending line at a 90° angle was then drawn from the nasal vascular/avascular boundary to the bifurcating line. The extending line was shortened so that it extended from the disc. This measurement (L3) comprised the nasal vascular extent. In follow-up images, the same bifurcating points and points from the disc edge were used.
Measurement methodology using a representative fundus image. The methodology for measuring retinal vascular extents is depicted through a series of steps, marked by black lines in each image: A. The disc diameter is measured in pixels. B. A line is drawn between bifurcating points in the superior and inferior arcades. These points are kept consistent between baseline and follow-up and posterior pole and temporal images. C. The bifurcating line is halved. D. A line is drawn from the optic disc edge to the halved bifurcating line (L1). The angle formed by these 2 lines is recorded. E. In the temporal image, steps B and C are repeated to form a halved bifurcating line. An extending line is then drawn from the halved bifurcating line to the temporal vascular/avascular boundary, using the same angle as step D. F. The extending line is cropped to extend from the edge of the halved bifurcating line to the temporal vascular/avascular boundary, and the distance formed (L2) is saved. G. In the nasal image, a line is drawn across superior and inferior disc vessels to bifurcate the disc. H. A perpendicular extending line is drawn from the bifurcating line to the nasal vascular/avascular boundary. I. The perpendicular extending line is then advanced to the edge of the optic disc. The distance formed, L3, is saved and used as the nasal extent in pixels. Final calculations include the following: L1 + L2 = temporal extent. L3 = nasal extent.
We defined vascular extent as pixel measurements from the optic disc margin to the junction of the vascular/avascular retina and defined the difference in extent between baseline and follow-up as vascular extension. The temporal and nasal extent in pixels was divided by the disc diameter in pixels to obtain the length of the temporal extent in disc diameter units (DD). Using DD as units ensured that the effects of zoom were accounted for between separate visits. In addition, because angles and landmarks were kept constant between images, the effects of rotation and warping were minimized.
A masked vitreoretinal specialist reread the deidentified images, and together, the masked analysts came to a consensus on the correct measurements. Baseline and follow-up images were measured along the same horizontal meridian nasally and temporally.
STATISTICAL ANALYSIS
Statistical analyses were performed using Python (version 3.12.10) with NumPy (version 2.2.6) and SciPy (version 1.16.2) libraries. Owing to small sample sizes, nonparametric tests were used for the primary between-group and paired comparisons. The Mann-Whitney U test was used for unpaired, between-group data, whereas the Wilcoxon signed-rank test was used to analyze differences in paired data (direct comparisons between baseline and follow-up for either the treatment or control groups). To characterize nasal-temporal asymmetry, we calculated a temporal-to-nasal extension ratio for each eye and used the Levene test to compare the variance of these ratios between groups. CIs (95%) for between-group differences (CI of difference) were calculated using bootstrap resampling with 10,000 iterations. Statistical significance was defined as P <.05.
RESULTS
PATIENT CHARACTERISTICS
A total of 14 eyes from 14 infants were included in the study: 7 eyes in the treatment group (2 OD, 5 OS) and 7 in the control group (1 OD, 6 OS). Baseline characteristics are provided in Table 1 . There were no significant differences in mean birthweight, PMA, or gestational age between groups at baseline ( P = 1.0, P =.405, and P =.425, respectively). In the control group, no eyes had zone I ROP and 7 eyes had zone II ROP. The treatment group consisted of 1 eye with zone I ROP and 6 eyes with zone II ROP.
TABLE 1
Infant Characteristics.
| Characteristics | Control | Treatment | 95% CI of Difference | P Value |
|---|---|---|---|---|
| Number included | 7 | 7 | — | — |
| Mean gestational age (wk) | 24 ± 1 | 23 ± 1 | [−1, 0] | .429 |
| Mean birthweight (g) | 624 ± 103 | 615 ± 90 | [−101, 84] | 1.000 |
| Mean PMA at baseline image (wk) | 35.5 ± 2.8 | 34.5 ± 2.4 | [−3.5, 1.5] | .405 |
| Infants with zone I ROP (n) | 0 | 1 | — | — |
| Infants with zone II ROP (n) | 7 | 6 | — | — |
PMA = postmenstrual age; ROP = retinopathy of prematurity.
VASCULAR EXTENSION
At baseline, the control group demonstrated a significantly greater vascular extent both nasally (9.3 ± 2.0 DD vs 5.8 ± 2.0 DD, P =.011) and temporally (12.3 ± 2.6 DD vs 8.7 ± 2.5 DD, P =.017) compared with the treatment group ( Table 2 , Figure 3 ). This difference was expected given that the control group had type 2 or less severe ROP compared with the treatment group, which comprised solely T1ROP. At follow-up, nasal vascular extent remained significantly different between the treatment and control groups (control: 9.9 ± 2.3 DD, treatment: 7.1 ± 1.7 DD, P =.026); however, differences in temporal vascular extent were no longer significant (control: 12.9 ± 3.0 DD, treatment: 10.4 ± 2.7 DD, P =.097), suggesting catch-up growth.
TABLE 2
Vascular Extent in Temporal and Nasal Hemispheres at Baseline and Follow-Up and Vascular Extension.
| Measurements | Control | Treatment | 95% CI of Difference | P Value |
|---|---|---|---|---|
| Nasal baseline extent (DD) | 9.3 ± 2.0 | 5.8 ± 2.0 | [−5.4, −1.6] | .011 |
| Nasal follow-up extent (DD) | 9.9 ± 2.3 | 7.1 ± 1.7 | [−4.7, −0.8] | .026 |
| Nasal extension from baseline to follow-up (DD) | 0.55 ± 0.44 | 1.27 ± 0.59 | [0.22, 1.21] | .038 |
| Nasal extension from baseline to follow-up divided by days (DD/d) | 0.032 ± 0.028 | 0.072 ± 0.045 | [0.004, 0.076] | .073 |
| Temporal baseline extent (DD) | 12.3 ±2.6 | 8.7 ± 2.5 | [−6.0, −1.0] | .017 |
| Temporal follow-up extent (DD) | 12.9 ± 3.0 | 10.4 ± 2.7 | [−5.3, 0.3] | .097 |
| Temporal extension from baseline to follow-up (DD) | 0.62 ± 0.49 | 1.69 ± 1.16 | [0.26, 1.98] | .040 |
| Temporal extension from baseline to follow-up divided by days (DD/d) | 0.037 ± 0.029 | 0.086 ± 0.058 | [0.009, 0.097] | .041 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
