The role of cytokines and treatment algorithms in retinopathy of prematurity

Abstract

Purpose of review

Currently, severe ROP is diagnosed by clinical evaluation and not a laboratory test. Laser is still considered standard care. However, anti-VEGF agents are being used and there are questions whether and/or if to use them, what dose or type of agent should be considered and what agent may be most beneficial in specific cases.Also unclear are the effects of laser or anti-VEGF on severe ROP, refractive outcomes or infant development. This article reviews recent studies related to these questions and other trials for severe ROP.

Recent findings

Imaging studies identify biomarkers of risk (plus disease, stage 3 ROP, ROP in zone I). Intravitreal bevacizumab or ranibizumab are reported effective in treating aggressive posterior ROP (APROP) in small series. Recurrences and effects on myopia vary among studies. Use of anti-VEGF agents affects cytokines in the infant blood and reduces systemic VEGF for up to 2 months, raising potential safety concerns. The effects of treatment vary based on infant size and are not comparable. Evidence for most studies is not high.

Summary

Studies support experimental evidence that inhibiting VEGF reduces Stage 3 ROP and peripheral avascular retina. Ongoing large-scale clinical trials may provide clarity for best treatments of severe ROP. Current guidelines hold for screening and treatment for type 1 ROP.

Introduction

Throughout the world, infants with severe retinopathy of prematurity (ROP) differ in size and age(1, 2) and from the first description of retrolental fibroplasia (RLF) in the US in the 1940's(3). In some countries, infants born older than 32 weeks gestational age and larger than 1500 g develop severe ROP, whereas in the US, screening is restricted to infants ≤ 30 weeks gestational age and ≤ 1500g birth weight, and most infants who develop severe ROP are less than 28 weeks gestational age and under 1000 g birth weight(4). Variability may be due to differences in resources for prenatal and perinatal care. High or unregulated oxygenation at birth can vasoconstrict or injure newly formed capillaries, potentially predisposing larger and older infants to severe ROP in some countries and resulting in a different pathophysiology from less developmentally mature, growth-restricted infants in the US(5). Inhibition of the bioactivity of vascular endothelial growth factor (VEGF) experimentally inhibits intravitreal angiogenesis and facilitates physiologic retinal vascularization(3). However, VEGF is important in retinal and organ development(3), and drug dose of intravitreal anti-VEGF agents cannot be assumed to have similar efficacies and safety profiles for younger, growth-restricted infants as for older, larger infants because of differences in infant development and because the same intravitreal dose of bevacizumab is diluted less in the lower blood volume of a smaller infant. Resources for screening, diagnosis and treatment vary worldwide, including the availability of trained ophthalmology experts(6). Laser that is not delivered adequately can lead to worse outcomes for severe ROP. Visualization to treat with laser may be hampered by persistent hyaloidal vasculature if laser is delivered simultaneously with anti-VEGF injections, whereas laser treatment delivered after an anti-VEGF agent and at an age in development when hyaloidal regression has already occurred may be more complete because of better visualization. If these aspects are not considered in studies, there may be erroneous conclusions.Finally, genotypes vary among populations of infants and also may play a role(7). Several studies reported variants in members of the Wnt signaling pathway in association with ROP, but a large candidate gene study of extremely low birth weight infants found variants in brain-derived neurotrophic factor (BDNF), supporting the role of neurovascular connections(7). Therefore, series published world-wide are not necessarily comparable. This article will focus on recent studies regarding cytokines, growth factors, and strategies for ROP management, including anti-VEGF treatments. The level of evidence in these series is often not high, but information may be helpful for some clinical situations and to refine hypotheses for future studies. This article discusses studies that used intravitreal bevacizumab and ranibizumab for ROP and neither is approved for this indication at the time of this writing.

Infant gestational age and birth weight in multicenter clinical trials

Infants enrolled in recent multicenter clinical trials (Early Treatment for Retinopathy of Prematurity (ETROP), Telemedicine Approaches for the Evaluation of Acute-Phase Retinopathy of Prematurity study [e-ROP]) were of younger gestational ages and smaller birth weights than those in Cryotherapy for Retinopathy of Prematurity (CRYO-ROP). Mean gestational ages declined from 27.9 to 27 weeks, and more enrolled infants had birth weights less than 750 grams (from 15.8% to 33.4%). The post-gestational ages at which ROP developed were between 34.1 and 34.8 weeks(8).

Algorithms: Identifying infants at risk of ROP

Low serum levels of insulin-like growth factor 1 (IGF-1) were associated with poor postnatal growth, wide peripheral avascular retina and greater risk of severe ROP(9). Algorithms simplified to monitor postnatal growth were reanalyzed and revised to improve specificity in Swedish and North American cohorts(10). A number of studies have proposed algorithms, but few studies provided sufficient evidence for the AAO to recommend widespread use in place of current screening recommendations. The AAO recommends additional research to optimize ROP predictive model development, validation and application before widespread use of predictive algorithms(11).

One study of 74 infants reported an association of low IGF-1 with stage of ROP in a racially diverse group(12). However, another study of 36 infant serum IGF-1 levels measured at 31, 32, and 33 weeks post-gestational age was unable to determine a threshold level of IGF-1 to exclude some infants from ROP screening. The study reported differences in IGF-1 levels based on maternal race(13).

A number of studies reviewed associations between blood levels of various factors and ROP. In a retrospective study, blood levels of VEGF and PEDF decreased with increasing postnatal day as did the PEDF/VEGF ratio in infants with ROP. In contrast, infants without ROP had constant levels, suggesting that the ratio might be useful to predict risk, but more study is needed(14).

Several studies reported increased risk of some hematologic indices and ROP. In one, infant mortality was associated with more nucleated red blood cells within days 2 to 5 of life(15). An association between high blood erythropoietin levels in extremely low gestational age infants (<28 weeks GA) at two weeks of age was associated with ROP and bronchopulmonary dysplasia not requiring ventilation(15). A single site study associated the number of blood transfusions within 30 days with greater risk of severe ROP, and within the first month of life with increased mortality in extremely low birth weight infants(16).

Several imaging studies analyzed detection of referral warranted ROP (presence of stage 3 ROP, plus disease, or zone 1 ROP; Table 1). The ROPtool was accurate to detect tortuosity by lay readers(17). The Karnataka Internet Assisted Diagnosis of ROP (KIDROP) found increased ROP in private vs. government centers and provided a “real-world” view of ROP through telemedicine screening(18). Image characteristics at 34 weeks post-menstrual age predicted severe ROP and may be useful, once validated(19).

Table 1. Definitions of Severe ROP.

	Threshold ROP	Type 1 ROP	Referral warranted ROP
Zone I	Stage 3 ROP, 5 contiguous or 8 total clock hours, Plus disease	Any ROP with Plus or Stage 3 ROP without Plus	Any ROP and/or Plus
Zone II	Stage 3, 5 contiguous or 8 total clock hours, Plus disease	Stage 2 or 3 ROP, with plus disease	Stage 3 ROP and/or Plus
Plus disease	4 quadrants	2 quadrants	2 quadrants

Current guidelines include laser for type 1 ROP and consideration of intravitreal bevacizumab for certain zone I, stage 3 ROP with plus disease after informed consent.

At this point, no blood test has been found useful to predict infants at risk of ROP, but retinal features determined by fundus imaging may prove to be valuable with additional validation.

Algorithms: Identifying infants at risk of ROP by maternal factors

Maternal preeclampsia has been reported as a risk for ROP. However, preeclampsia was reported as protective in an international cohort of 27,846 preterm infants(20) or to have no risk in a population-based cohort of very-low birth weight infants from Taiwan(21) and in a meta-analysis of studies on 925 eligible of 1142 published records(22). These observational studies differed from earlier studies reporting increased risk of preeclampsia and ROP. Preeclampsia and ROP are linked to premature birth, and this may cause bias. Clarity regarding the risk of preeclampsia and ROP is needed.

In premature infants born to diabetic and non-diabetic mothers, maternal diabetes was reported as an independent risk factor for ROP and type 1 ROP(22).

Algorithms: Reducing risk of ROP in premature infants by perinatal exposures

A comprehensive search of databases was undertaken to evaluate risk factors (eg., nutritional interventions, erythropoietin, supplemental oxygen, blood transfusion) associated with severe ROP. Although low oxygen saturation targets reduced risk of ROP(23), they were also reported to increase infant mortality in some studies. Poor parenteral nutrition, use of erythropoietin or blood transfusions were not associated with severe ROP. The BOOSTII, which reported low oxygen saturation targets associated with reduced ROP and increased mortality, found high oxygen saturation targets associated with a trend toward reduced extent of retinal vascular developmental growth(24). Single site studies reported duration of oxygen supplementation and high maximum inspired oxygen as significant risk factors for severe ROP(25). Because of increased mortality reported with low oxygen saturation targets, some neonatal nurseries have increased oxygen saturation targets. However, ophthalmologists are concerned, and a study reported higher and more severe ROP after changing from a lower to 91-95% oxygen saturation targets(26).

Algorithms: Treating severe ROP in premature infants by anti-VEGF and/or laser

Two large clinical trials are underway to assess treatment for severe ROP with different anti-VEGF agents. The Pediatric Eye Disease Investigative Group (PEDIG, NEI/NIH, NCT02390531) is assessing de-escalating doses of intravitreal bevacizumab for efficacy and some safety outcomes. The RAINBOW study (Novartis, NCT02375971) compares two doses of intravitreal ranibizumab to laser. Until we have information from large-scale, randomized, comparative trials, small-scale studies provide guidance, but sample sizes are often small, follow up varies, and concerns brought forth in the introduction are present. A meta-analysis of mostly observational studies and only one clinical trial using anti-VEGF agents estimated a 2.8% 6-month risk of retreatment per eye and a 1.6% risk of ocular complication without retreatment but was unable to determine risk of systemic effects based on lack of sufficient data(27).

Eye outcomes after anti-VEGF agents: retinal vascular features, recurrences and refractive status

Several studies reported changes in vascular features, namely, reduced intravitreal neovascularization and greater extension of physiologic peripheral vascularization following intravitreal bevacizumab (IVB). A study restricted analyses to aggressive posterior ROP (APROP) and reported a saw-toothed shunt and ridge(28) as common features following bevacizumab(28). Another single-center study reported findings of a scalloped appearance of the vascular/avascular junction and concluded that the need for rescue treatment with peripheral laser was associated with decreased birth weight(29). 75 infants from BEAT-ROP (Efficacy of Intravitreal Bevacizumab for Stage 3+ Retinopathy of Prematurity study) and newly diagnosed infants with APROP or posterior zone II severe ROP treated with 0.625 mg IVB were followed until 65 weeks adjusted age. 8.3% of 241 infants (7.2% of 471 eyes) had recurrences and when they occurred, were seen in 90% of infants (94% of eyes) between 45 and 55 weeks (mean 52 weeks) with a mean interval between treatment and recurrence of 16.2 weeks. Recurrences at the previous ridge or leading edge were found in former APROP but only in the advancing edge in eyes without confluent neovascularization. Retreatment led to slow regression(30). Although most recurrences were seen within 1 year of adjusted age, a report described late reactivation of ROP after bilateral IVB with a tractional retinal detachment in one eye and milder reactivation in the fellow eye at 2.5 years of age(31).

One series reported fluorescein angiographic outcomes following 0.312 mg intravitreal IVB for posterior zone II or zone I ROP, including APROP. Although regression occurred in 100% of posterior zone II eyes, only 25% of APROP regressed. 11% had recurrences at 1 week and 18% at 9-12 weeks following treatment, with leakage at the junction of vascular/avascular borders(32).

A retrospective comparison following conventional laser, IVB and laser, or IVB and deferred laser for type 1 ROP in zone I reported an unfavorable outcome in 22% of eyes treated with laser only compared to favorable results in the other two groups(33). Lower levels of myopia were reported in the IVB and deferred laser group. These findings suggest there was a benefit to intravitreal bevacizumab compared to conventional laser, but it is unknown if laser was less complete in the conventional laser or IVB and laser groups compared to the ETROP criteria.

54 infants with type I ROP, 33 with posterior disease and 21 in posterior zone II, were treated with IVB or laser. Faster regression occurred after IVB for zone I ROP; however, there was no difference for disease in peripheral zone II. 12% of eyes with posterior disease had recurrences in both eyes following IVB and required laser at an average of 12.7 weeks(34). There was a suggestion that spherical equivalents were lower in infants with posterior (+0.37) compared to peripheral zone II (+3.0) ROP. Some series report higher rates of myopia in eyes treated with laser than in those treated with IVB or ranibizumab(35). Other studies report that having type I ROP, compared to non-type 1 ROP or no ROP, was associated with myopia and that outcomes were similar following laser or IVB at 3 years of age(36), whereas still others report posterior disease, compared to peripheral disease, was associated with myopia(34). Lack of statistical significance may be from small sample sizes, and longer follow up is needed to assess refractive outcomes after laser vs. anti-VEGF agents..

A series of intravitreal ranibizumab (0.5 mg) showed reduced plasma VEGF one day after treatment but no difference at one, 2 or 4 weeks after treatment(37). Another study reported reduced serum VEGF to a greater extent and up to 2 months after IVB treatment compared to ranibizumab; both treatments were efficacious for treating severe ROP(38).

A prospective randomized trial of infants with zone II, stage 2 or 3 ROP with plus disease comparing intravitreal ranibizumab to laser reported recurrences in 26 ranibizumab-treated eyes (13 infants) vs. 2 laser-treated eyes (1 infant) at 6 months follow up(39). A retrospective study reported favorable outcomes in 87.5% of infants with intravitreal ranibizumab only and 71% with ranibizumab and laser delivered at the time of injection or deferred(40). A retrospective study reported more recurrences after ranibizumab than IVB(41), but another consecutive study reported no differences in recurrence rate at one year and full vascularization in 60% of ranibizumab-treated eyes compared to 46.7% IVB-treated eyes(42).

Studies reported mixed results. Until large-scale, randomized, powered clinical studies or trials provide greater insight, recommendations from the American Academy of Pediatrics are to treat type 1 ROP with laser and consider intravitreal bevacizumab for zone I, stage 3, plus disease (0.625 mg in 0.025 mL) after careful informed consent.

Anti-VEGF agent systemic outcomes: levels of blood VEGF, cytokines, and systemic effects

A review of 12 studies was performed to assess the association of systemic VEGF and ROP. Investigators measured cord blood, serum, plasma and tissue samples and had mixed results with ROP associated with: low VEGF at birth; no difference in VEGF levels; increased VEGF in postnatal blood; and an increase in VEGF at the time of ROP followed by a decrease at the time of treatment. Serum VEGF was reduced for 2 months following a single intravitreal injection(38) raising concerns about the effect of prolonged reduction of VEGF might have in the developing premature infant(43). Studies do not agree as to the association of VEGF measured in the blood and the severity of ROP, so use of anti-VEGF agents is not supported by finding a certain systemic level of VEGF. Also, there is concern of reduced systemic VEGF from anti-VEGF treatment in developing preterm infants.

One study analyzed whole blood of 13 preterm infants with type 1 ROP prior to treatment with IVB and then again at 42 days after treatment. A control group of age-matched untreated preterm infants without ROP was tested at both time points. Plasma soluble VEGFA appeared similar in infants with type 1 ROP or no ROP at base line, average age of 34.9 weeks post-menstrual age (34 weeks, control group). At day 42, both groups had reduced VEGFA; however, anti-VEGF treated infants had over a 5 fold decrease in VEGFA compared to the non-treated group without type 1 ROP(43). Other cytokines, angiogenic factors and receptors were decreased (VEGF-D, angiopoietin-2, soluble VEGF receptors 1 and 2, soluble TNFα Receptor I and II, soluble gp130, IL-6R) or increased (soluble VEGFC, placental growth factor, endothelin 1, FGF1). VEGFA, VEGFC, VEGFD, endothelin 1, BMP-9, VEGFR1, IL6R, sGP130 and TNFRI appeared substantially different between the two groups. (Previously, IVB was found to reduce serum IGF-1 in treated type I ROP vs. untreated preterm infants without ROP.) The significance of these findings is unknown,, but suggests the need for careful monitoring of developing preterm infants treated with intravitreal anti-VEGF agents.

Besides recurrences, persistent avascular retina, and myopia, other systemic complications were reported following IVB(44). Hypotension 3 days after IVB was reported in a clinically stable infant. In a retrospective study from the Canadian Neonatal Network, neurodevelopmental outcomes determined by Bayley Scales of Infant and Toddler Development Third Edition at 18 months corrected age of infants less than 29 weeks gestation reported significant reduction in motor score and 3 times greater odds of severe neurodevelopmental disabilities [severe cerebral palsy, hearing aids, or bilateral blindness] in IVB treated infants than those treated with laser. The study was retrospective and included 27 IVB treated and 98 laser treated eyes but pointed to concerns regarding IVB and need to monitor long-term safety(45). Other studies were unable to identify differences in neurodevelopment 2 years after treatment(46).

Algorithms: Other treatment trials

A Phase II clinical trial studied the infusion of IGF-1 and IGF-1 binding protein 3 to levels of IGF-1 to normal intrauterine serum concentrations in order to reduce the risk of severe ROP and improve infant growth and cognitive development in preterm infants (Shire, NCT01096784). Reports show safety and ability to infuse to the level anticipated. The primary outcome of reduction in severe ROP was not attained, but secondary outcomes (intraventricular hemorrhage and bronchopulmonary dysplasia) were(9) (47). Subjects will be invited into a continuation study (Shire, NCT02386839).

Previous studies reported increased risk of severe ROP in infants treated with erythropoietin for anemia of prematurity, but this effect may have depended on the timing of administration. Erythropoietin has neuroprotective effects, and a clinical trial (NCT01378273) is ongoing to test the effect of erythropoietin on cognitive development and other outcomes, including ROP(48).

Inositol is an essential compound and was studied in clinical trial [NCT01954082] to assess the effects on premature infant morbidity, including ROP. Despite initial safety studies(49), the study was stopped.

Conclusions

ROP varies throughout the world. Differences in genotype, resources for prenatal and perinatal care, and the number of trained(6) experts to accurately diagnose and treat ROP may account in part for this. Evidence for treatment with anti-VEGF agents is accruing, yet the level of evidence for most studies is low. Studies report IVB at doses currently used reduce blood cytokines and VEGF for at least 2 months and more severely and longer than intravitreal ranibizumab. Recurrences are estimated at <10% for APROP and usually manifest within a year but some reports suggest longer follow up in needed. There is some evidence of impaired neurodevelopment after IVB and that myopia is lessened after IVB compared to eyes treated with l laser, but larger randomized studies and longer follow-up are needed. Concerns exist about the use of lower oxygen saturation targets to reduce ROP risk because of reported increase in mortality; however, there are also concerns that severe ROP will increase with higher oxygen saturation targets.. When considering outcomes in the literature, it is important to determine if any ROP or severe ROP is evaluated, what infant gestational ages and birth weights are studied, the infant oxygen exposure, perinatal resources and genetic influences. Clinical trials are ongoing, but current recommendations from the AAP and AAO for treatment of type 1 ROP still apply.

Key Points.

• Infants with severe ROP vary in size and age throughout the world, and studies regarding efficacy, dose and safety may not be comparable.

• Imaging biomarkers predict infants at risk, whereas evidence for cytokines or growth factors is not strong

• Inhibition of VEGF reduces severe ROP and improves physiologic retinal vascularization in agreement with experimental studies, but the best treatment algorithm in terms of efficacy, dose, type of agent, and the role of anti-VEGF therapy vs. laser remains unknown..

• Use of anti-VEGF in APROP may reduce degree of myopia compared to laser but long-term randomized studies are needed

• Intravitreal bevacizumab at 0.625 mg results in recurrences in ∼10% of infants with APROP on average at 52 weeks adjusted age