Longitudinal Change of Refractive Error in Retinopathy of Prematurity Treated With Intravitreal Bevacizumab or Laser Photocoagulation


To compare progression of myopia and refractive error in former premature infants with retinopathy of prematurity (ROP) treated using intravitreal bevacizumab (IVB) or laser.


Retrospective clinical cohort study.


We identified premature infants with ROP treated using IVB from 2011 to 2020 and compared their longitudinal cycloplegic refraction data to that of infants with ROP treated using laser during the same timeframe. A subset of infants treated using IVB also underwent additional treatment using laser. We included cycloplegic refractions from 789 cumulative visits over a median 3.2 years. We used a linear mixed-effects model with a log decay function to evaluate how refraction changed with age after treatment.


In aggregate, the model estimated a significant ( P < .001) trend in refraction—from slight hyperopia to relatively more myopic states. However, progression in laser-treated eyes was significantly ( P < .001) more rapid, regardless of treatment with IVB. The number of laser spots resulted in increased myopic progression by approximately 0.16 diopters per 100 laser spots. Both ROP stage and zone had a significant effect on myopic progression, with more severe disease resulting in faster myopic progression. Random effects, including individual subject variation with nested variance for left and right eye, accounted for 86.4% of the remaining variance not explained by age and treatment.


Laser treatment for severe ROP increases the trend to severe myopia. In our sample, IVB did not affect myopic progression but did substantially reduce the amount of consequent laser required to treat ROP. The effect of laser persists after accounting for differences in ROP stage and zone.

M yopia is the most common ophthalmic sequela of preterm birth and is especially common in children with retinopathy of prematurity (ROP). There has long been interest in how the treatment of ROP impacts refractive development. Several cross-sectional studies have shown that children with type 1 ROP treated using laser photocoagulation develop a higher degree of myopia than children treated using intravitreal bevacizumab (IVB). Indeed, 1-year-old children with IVB-treated ROP show a distribution of refractive errors similar to children with spontaneously regressed ROP. In laser-treated children, the development of myopia is more rapid, particularly in the first year of life.

The BEAT-ROP clinical trial (ClinicalTrials.gov identifier: NCT0622726) showed a higher incidence of myopia and high myopia in eyes treated with peripheral laser compared with IVB at approximately 2.5 years of age, and this difference in refractive outcomes between treatment groups persisted when matched for severity of ROP. Conversely, Kuo and associates did not find any difference in refractive error between laser- and IVB-treated children at 3 years of age. A meta-analysis examining 7 studies of refractive outcomes in children treated for ROP found significantly less refractive error, a lower prevalence of high myopia, and less severe astigmatism in IVB-treated children. The authors also reported that refractive development in IVB-treated children is still relatively abnormal compared with healthy, full-term children, exhibiting a higher prevalence of both myopia and astigmatism.

The rate of myopic progression depends on ROP severity. Myopia in children with spontaneously regressed ROP stabilizes at approximately 2.5 years of age, while myopia associated with severe ROP tends to develop before 1.5 years of age and then levels off. , , While many retrospective studies have shown an increased prevalence of myopia in premature infants treated using laser, especially at younger ages, treatment differences in the rate of myopic change were only recently reported in the longitudinal study by Simmons and associates. The authors examined the rate of change in refractive error in 2 cohorts of type 1 ROP, treated using either IVB or laser. They found a significantly higher rate of myopization in the laser vs the IVB group, −5.0 diopters (D) per year vs −3.5 diopters per year in the first postnatal year, followed by slower progression. They also documented greater anisometropia in the laser group compared with the IVB group but no significant difference in visual acuity outcomes.

As the use of IVB therapy to treat ROP has burgeoned, so too have the number of cases that undergo supplemental, rather than primary, laser treatment. In some cases, laser treatment is performed for recurrent disease; in other cases, laser treatment is performed prophylactically for peripheral avascular retina. The effect of this subsequent laser treatment on refractive error has not been extensively studied. Some studies have grouped together laser subjects with IVB plus laser cohorts, which limits understanding of how this additional treatment may impact longitudinal refractive development. , Yoon and associates considered the timing of laser treatment, reporting less myopia in subjects treated using IVB plus deferred laser compared with those treated using only laser or using IVB plus concurrent laser.

In the present study, we considered patients treated using 1) laser alone, 2) IVB alone, or 3) IVB with subsequent laser. We examined how refractive error progressed over a period of >8.5 years.


As part of an ongoing study of visual development in children with a history of preterm birth, we reviewed charts of all infants (gestational age at birth 22-32 weeks) who were treated for type 1 ROP between April 2011 and November 2020. We included all individuals who had ≥2 cycloplegic refractions and complete clinical data on treatment (ie, number of laser spots applied). We excluded those with disorders associated with abnormal refractive development or inability to accurately refract, such as retinal detachment involving the macula, glaucoma, aphakia, microphthalmia, and persistent hyperplastic primary vitreous. This retrospective clinical cohort study was approved by the Boston Children’s Hospital Institutional Review Board, is in accordance with Health Insurance Portability and Accountability Act regulations, and adheres to the tenets of the Declaration of Helsinki.

Infants were screened for ROP according to joint guidelines and treated if type 1 ROP criteria were met. Selection of treatment modality was at the discretion of the attending ophthalmologist, as was the dose of IVB and/or laser treatment parameters.

Cycloplegic streak retinoscopy was conducted by experienced pediatric ophthalmologists during the course of clinical care. Cycloplegia was obtained primarily using cyclopentolate (or Cyclomydril for infants) or combination drops (cyclopentolate, tropicamide, and phenylephrine). Visual acuity was assessed using preferential looking Teller cards, Lea symbols, or HOTV or Snellen letters. For analyses, acuities were converted to the logarithm of the minimum angle of resolution.

We implemented a linear mixed-effects (LME) model to evaluate how spherical equivalent and visual acuity changed with postmenstrual age (following log transformation). The model included both fixed and random effects and was fit using restricted maximum likelihood estimates of the parameters. The model included an intercept at term birth and estimates for the effect of both IVB treatment and number of laser spots on myopic progression. Random effects were individual patient with a nested random effect for eye (nested within individual). In a second, more complicated model with additional estimates for ROP severity, we included interaction effects for ROP zone, stage, and clock-hours of affected retina. Finally, we evaluated how astigmatism and anisometropia differed as a function of IVB treatment, number of laser spots, and ROP zone, stage, and clock-hour using LME models. The P values reported are estimates of the probability that the parameter described is significantly contributing to the model, ie, the estimated coefficient is not 0. Statistical analysis was conducted in RStudio (R Core Team, http://www.R-project.org/ ) using the lme4 and lmerTest packages for LME modeling. ,


Of the 88 subjects included in this study, 22 (40 eyes) received only IVB, 48 (90 eyes) received only laser, and 18 (36 eyes) received both IVB and laser. Table 1 shows the baseline characteristics of the study population. Most subjects (n = 78) underwent bilateral treatment, but 10 (n = 4 treated using IVB and n = 6 treated using laser) received monocular treatment. We included cycloplegic refractions from 789 cumulative visits. A total of 2 to 13 refractions (median 3 refractions) spanning 0.5 to >8.5 years (median 3.2 years) were obtained in each patient. The first of these was obtained at a median age of 27.5 weeks (range 13-106 weeks). The median interval between refractions was 30.75 weeks (range 3-343 weeks).


Patient Characteristics Between Treatment Groups

IVB (n = 22) IVB Then Laser (n = 18) Laser (n = 48)
Patient Characteristic
Gestational age (wks) 24.0 (23.0-29.0) 24.5 (23.0-28.0) 25.0 (22.0-30.0)
Weight (g) 545.0 (410.0-960.0) 567.5 (460.0-950.0) 690.0 (330.0-1760.0)
Age at first laser treatment (wks) NA 48.3 (20.6-99.9) 14.3 (5.3-21.4)
Age at first CRx (wks) 26.9 (17.6-84.3) 31.2 (17.7-106.3) 26.5 (12.6-104.4)
Time followed (yrs) 1.4 (0.2-5.3) 3.0 (0.7-6.8) 3.2 (0.2-8.9)
Interval between CRx (wks) 20.7 (3.3-68.0) 22.6 (6.6-60.8) 40.9 (5.9-343.0)
Asian 1 (6.7) 3 (20) 5 (12)
Black/African American 2 (13) 3 (20) 8 (20)
Declined 0 (0) 0 (0) 1 (2.5)
Other 2 (13) 2 (13) 9 (22)
White 10 (67) 7 (47) 17 (42)

CRx = cycloplegic refraction; IQR = interquartile range; IVB = intravitreal bevacizumab; NA = not applicable.

Values shown as median (range) or n (%).

Characteristics of ROP and ROP treatment are described in Table 2 . ROP zone, stage, and clock-hour are recorded as maximum severity of the acute phase of disease, regardless of previous treatment. Eyes that received laser after IVB therapy received fewer laser spots than eyes receiving laser as the sole therapy. At the time of subsequent laser treatment, 17 of 18 individuals (34/36 eyes) were noted to have peripheral avascular retina, and 7 of 18 (14/36 eyes) were treated for recurrent ROP. Clinical data from these 18 individuals can be found in the Supplementary Material.


Retinopathy of Prematurity Characteristics per Eye

Characteristic a IVB IVB Then Laser Laser
ROP zone
I 12 (27) 18 (56) 6 (6.3)
II 32 (73) 14 (44) 89 (94)
ROP stage
1 2 (4.5) 0 (0) 0 (0)
2 15 (34) 5 (15) 10 (10)
3 27 (61) 29 (85) 80 (83)
4 0 (0) 0 (0) 6 (6.2)
No. of clock hours 9.50 (6.00-12.00) 10.00 (8.00-12.00) 8.00 (6.00-12.00)
Laser spots NA 676 (463-1032) 1098 (839-1568)
IVB dose (mg) 0.50 (0.44-0.62) 0.50 (0.25-0.50) NA

IQR = interquartile range; IVB = intravitreal bevacizumab; NA = not applicable; ROP = retinopathy of prematurity.

a Characteristic at maximum severity of the acute phase of disease. Values shown as median (IQR) or n (%). ROP zone data were missing for n = 4 eyes; IVB dosage was missing for n = 8 eyes.

There were notable group differences regarding the stage and zone of maximum severity of the acute phase of disease. Most eyes that had zone I ROP received IVB, and a majority of eyes with zone II disease had laser treatment, either as monotherapy or after IVB treatment. For LME modeling, ROP stage was binarized to less severe (stages 1 and 2) or more severe (stages 3 and 4).


Figure 1 shows the results of the model fits for the median number of laser spots for each of the three treatment groups as well as individual courses of refractive development. Normative refractive error data are indicated by the gray ribbon. ,


Linear mixed-effects model predictions of refractive error (spherical equivalent) based on postmenstrual age (years) across median number of laser spots for 3 different treatment groups (0, 676, and 1098 spots, for IVB, IVB with laser, and laser groups, respectively). Data are plotted as individual points and slopes show model prediction. Shading indicates the 95% confidence interval for the model fit. D = diopter; IVB = intravitreal bevacizumab.

The model estimated a significant ( P < .001) trend in refraction, from slight hyperopia to relatively more myopic states, as is the course of emmetropization in healthy eyes. The model showed no significant impact of IVB treatment on spherical equivalent (SE; P = .895). However, treatment of eyes using laser significantly increased myopic progression ( P < .001) at a rate of approximately 0.16 D per log year for every 100 spots. Based on the model, longitudinal SE changes are described by the equation:

<SPAN role=presentation tabIndex=0 id=MathJax-Element-1-Frame class=MathJax style="POSITION: relative" data-mathml='SE(Age)=1.08−1.31log10(1+Age/40)−0.0629log10(1+Age/40)·IVB−0.00155log10(1+Age/40)·Laser’>𝑆𝐸(𝐴𝑔𝑒)=1.081.31log10(1+𝐴𝑔𝑒/40)0.0629log10(1+𝐴𝑔𝑒/40)·𝐼𝑉𝐵0.00155log10(1+𝐴𝑔𝑒/40)·𝐿𝑎𝑠𝑒𝑟SE(Age)=1.08−1.31log10(1+Age/40)−0.0629log10(1+Age/40)·IVB−0.00155log10(1+Age/40)·Laser

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Sep 11, 2022 | Posted by in OPHTHALMOLOGY | Comments Off on Longitudinal Change of Refractive Error in Retinopathy of Prematurity Treated With Intravitreal Bevacizumab or Laser Photocoagulation

Full access? Get Clinical Tree

Get Clinical Tree app for offline access