ROP: 80 Years after Its Detection – Where Do We Stand and How Long Will We Continue to Laser?

Terry first reported retinopathy of prematurity (ROP) in 1942 [1]. In 2019, the estimated global numbers of blind and moderately or severely visually impaired people due to ROP were 2,480,000 and 423,000, respectively [2]. From 1990 to 2019, the prevalence of ROP-induced visual impairment and blindness increased, although the burden decreased slightly [2]. In the 1940s, improved neonatal care resulted in the increased survival of preterm infants, resulting in a growing number of blind moderately preterm infants. In subsequent years, excessive oxygen supplementation was identified as a primary and modifiable risk factor. With reduced oxygen supply, ROP mainly affects extremely preterm survivors, creating the “second epidemic,” which continues in high-income countries. In Sweden, the one-year survival rates for infants with gestational age (GA) 21–22 weeks have increased from 10% 2004–2007 to 39% 2017–2019 [3]. Approximately 75% of surviving infants with GA 21–22 weeks and 43% of those with GA 23 weeks were treated for ROP [4]. The incidence of ROP was constant during the study period, but the treatment frequency increased and the GA of treated infants decreased. In contrast, the ROP incidence in the USA almost doubled from 2003 to 2019, and black infants and infants with parents with lower incomes were particularly affected, indicating insufficient care for mothers and infants [5].

ROP is a growing problem in low- and middle-income countries, in settings where neonatal care has improved enough for increased survival, but is insufficiently developed to prevent ROP with poor oxygen control and a lack of screening and treatment programs [6]. In most African countries, ROP has not been a problem since infants at risk did not survive. However, the incidence of ROP is increasing in Africa, and programs to manage it are urgently needed [7]. In 2019, the highest prevalence of ROP-related blindness globally was reported in South Africa and Afghanistan [2].

The retina is part of the central nervous system, and any stage of ROP is associated with reduced brain volumes at term-equivalent age and poorer neurodevelopment at 2 years of age [8]. Treatment warranting ROP is associated with delayed brain maturation and impaired neurodevelopmental outcomes at 18 months [9].

Laser therapy for ROP requires general anesthesia more often and for a longer time than anti-VEGF injections, with an increased risk to fragile preterm infants. Laser ablation destroys the peripheral retina and reduces the release of angiogenic factors. The alternative treatments, intravitreal injections with anti-vascular endothelial growth factor (VEGF) agents, are effective in suppressing neovascularization but suppress VEGF systemically for prolonged periods with poorly known long-term systemic consequences. The large variation in the incidence and severity of ROP in different countries and settings, depending on the quality of care, shows that ROP is largely preventable. In addition, one might speculate that ROP preventive measures may also promote general health and brain development. This narrative review focuses on recent achievements in the prevention, prediction, screening, and treatment of ROP.

PubMed was searched for relevant papers published up to March 2024 using the search terms retinopathy of prematurity, prevention, prediction, screening, and treatment.

Fetal retinal vascularization occurs from the optic nerve head to the periphery, guided by a circular moving wave of O₂ regulated growth factors stimulated by a similar wave of “physiologic hypoxia” caused by increased metabolism in the maturating peripheral retina. As blood vessels form, hypoxia is relieved at the site, but hypoxia continuously forms anteriorly, moving the wave forward. It is completed at ora serrata around the term. In the first phase of ROP after very preterm birth, particularly nonphysiological oxygenation, undernutrition, infections, and other factors impair retinal blood vessel growth and may destroy already formed vessels. In most cases, physiological vascularization resumes. However, in severe cases, hypoxia in the avascular peripheral retina upregulates VEGF to a degree that leads to pathologic neovascularization (phase 2) and, in the worst cases, total retinal detachment and blindness.

Establishing an internationally recognized classification system for ROP severity allows the evaluation of incidence and interventions in all settings. The most recent international ROP classification was published in 2021 [10]. ROP is classified according to severity into stages 1–5, where stage 1 denotes a demarcation line between the vascularized and avascular retina. In stage 2, the line becomes a ridge, and in stage 3, abnormal neovascularization develops on the ridge. Partial retinal detachment not involving the macula is stage 4A and involving the macula is stage 4B. Stage 5 is total retinal detachment. Plus disease is a sign of severity and manifests as a certain degree of retinal vascular dilation and tortuosity. Preplus is less pronounced. The location is divided into three zones centered at the optic nerve head, where zone I is the most central, with the largest avascular area and the highest risk of severe disease. In the latest classification, zone definitions were refined, and aggressive ROP, a particular form of rapidly progressing disease with poorly defined stages, replaced aggressive posterior ROP because aggressive ROP does not exclusively occur posteriorly [10]. Treatment is warranted for Type I ROP defined as zone I ROP stage 3 and/or plus disease in zone I, A-ROP, or stage 2 or 3 in zone II with plus disease.

Several risk factors are associated with ROP. Immaturity with low GA, low birth weight, and oxygen supplementation are major risk factors. Poor early neonatal weight gain and low circulating levels of insulin-like growth factor 1 (IGF-1) are strongly associated risk factors often used in ROP prediction [11]. Other risk factors are bronchopulmonary dysplasia [12], sepsis [13], severe necrotizing enterocolitis requiring surgery [14], patent ductus arteriosus [15], duration of parenteral nutrition [16], blood transfusions [17], and thrombocytopenia [18]. Some risk factors can be used to predict ROP and some can be modified to prevent ROP. These issues will be addressed in more detail later in this paper.

Oxygen supplementation is life-saving for extremely preterm infants, but is also a risk factor for ROP. After preterm birth, “physiological hypoxia” of the peripheral retina is replaced by relative hyperoxia with oxygen supplementation, which varies. Intermittent apnea also contributes to fluctuating oxygenation, which is particularly detrimental to the development of retinal blood vessels [19]. Surprisingly, infants with severe ROP had lower mean SpO₂ than those with no or mild ROP [19]. The optimal oxygen target saturations are still unknown.

The Neonatal Oxygenation Prospective Meta-analysis Collaboration compared a constant oxygen saturation (SpO₂) target of 85 to 89% and 91 to 95% in five similar studies. Lower target saturation (85–89% vs. 91–95%) reduced the risk of severe ROP but increased mortality and NEC. Currently, the European Consensus Guidelines on the Management of Respiratory Distress Syndrome recommend saturation targets between 90 and 94% and alarm limits between 89 and 95% [20].

Evidence suggests that targeting lower SpO₂ during the first weeks after birth and changing to a higher target later, when retinal hypoxia occurs, may reduce the incidence of severe ROP [21]. Graded SpO₂ targets from birth to 36 weeks postmenstrual age (PMA) in infants with GA 24–27 weeks significantly reduced not only severe ROP and ROP requiring treatment but also bronchopulmonary dysplasia and severe intraventricular hemorrhage (grade III–IV) [22]. The most immature infants with GA of 22–23 weeks seem to benefit from lower SpO₂ targets than recommended with regard to severe ROP requiring treatment [23]. To improve oxygen control, staff education and the implementation of manuals for oxygen titration or automatic oxygen control may increase the time spent within the saturation target [24, 25].

Most extremely preterm infants become anemic and receive one or more transfusions of adult erythrocytes. Adult hemoglobin (HbA) has a lower oxygen affinity than fetal hemoglobin (HbF) and appears to deliver more oxygen to the brain [26].

A low HbF fraction is an independent risk factor [27]. During the third trimester, the fraction of HbF is approximately 70–80% and decreases to a few percent at 6 months after term equivalent age [28]. Most extremely preterm infants receive one or more transfusions of adult red blood cells, replacing HbF with HbA. However, HbF synthesis continues at a PMA-dependent rate similar to that in utero [29]. One might speculate that repeated transfusions with adult erythrocytes not only deliver more oxygen to some tissues but also contribute to fluctuations in oxygenation. A median sample-related withdrawal of 58% of total blood volume during the first two postnatal weeks was reported in infants with GA <30 weeks, and a median transfused adult donor blood volume of more than 100% of infants’ total blood volume during the corresponding period [30]. Restricted transfusion policies reduce risk [17]. Possible preventive interventions include the avoidance of blood loss through delayed cord clamping and minimized blood sampling volume. A promising alternative is the transfusion of allogeneic cord blood erythrocytes, which increases the HbF level at 32 weeks PMA compared to adult erythrocyte transfusion [31]. An ongoing multicenter, randomized, controlled trial investigates the effects of allogeneic cord blood erythrocytes on HbF and severe ROP in infants with GA ≥24 and <28 weeks [32].

Studies on the effects of nutritional intervention in preterm infants have mainly addressed neurodevelopmental outcomes. However, according to a meta-analysis, initiating parenteral lipids (≥1.5 g/kg/day within the first 24 h of birth) has a favorable effect on ROP compared to lower initial doses and/or later initiation [33]. A Cochrane review concluded that very low evidence suggests that higher amino acid intake in parenteral nutrition reduces ROP, but not severe ROP [34].

Mothers’ own milk (MOM) is the best choice for all infants and is superior to donor human milk (DHM) in promoting growth and development. Ethical concerns preclude randomized trials comparing MOM and DHM. However, a randomized trial of 243 infants with GA <30 weeks compared DHM and preterm formula as substitutes for MOM when the supply was insufficient. A group of infants who received only MOM was also included. Of those fed exclusively MOM, 5.6% developed proliferative ROP stage 3 versus 19% fed DHM in addition to MOM versus 14% in the preterm formula-supplemented group, suggesting a protective effect of MOM against severe ROP [35].

The omega-3 long-chain polyunsaturated fatty acid (LCPUFA) docosahexaenoic acid (DHA), is essential for brain and retinal development and accumulates in the fetus from approximately 30 weeks of GA. The supply is dependent on maternal intake. The role of the omega-6 LCPUFA arachidonic acid (ArA) in neurodevelopment and ROP has received little attention. ArA is constantly supplied to the fetus during gestation. It is enriched in the immune system and vasculature and plays essential roles in vascularizing neuronal tissues [36].

Supplementation of fatty acids in preterm infants has mainly comprised omega-3 LCPUFA in various forms. In a mouse model of ROP, omega-3 LCPUFA supplementation reduced pathologic neovascularization [37]. The use of fish-oil-based parenteral lipid solutions rich in omega-3 LCPUFA has yielded inconsistent results. In 2017, a systematic review and meta-analysis concluded that providing fish-oil lipid emulsions may reduce the incidence of severe ROP or the need for laser therapy in preterm infants [38]. However, in 2019, a Cochrane review found no support for the preventive effect of omega-3 LCPUFA-rich fish-oil compared to non-fish oil emulsions in reducing severe ROP [39]. High levels of omega-3 LCPUFA suppress ArA and other omega-6 LCPUFA [40, 41]. Lower serum fractions of ArA are associated with ROP [42]. In addition, higher mean daily serum levels of DHA during the first 28 postnatal days are associated with a lower frequency of severe ROP, but only in infants with sufficiently high ArA levels, indicating the importance of proper ArA/DHA balance [43].

In the Mega Donna Mega trial, infants born with GA <28 weeks were randomized to receive enteral ArA (100 mg/kg/day) plus DHA (50 mg/kg/day) or no supplementation from within 3 days after birth until 40 weeks PMA. The supplement reduced severe ROP by 50% [44]. In a later Norwegian study, infants born at GA <29 weeks, subjected to a similar intervention, had reduced incidence of severe ROP (12.7% in the control group vs. 5.4% in the supplemented group). This difference was not statistically significant, possibly because of a lower ROP risk in a less immature population [45].

Some clinical studies have reported an association between neonatal hyperglycemia and ROP [46]. However, there is no consensus that hyperglycemia is a risk factor for ROP [46]. In addition, insulin treatment may not improve the outcomes of preterm infants with hyperglycemia [47].

Experimental evidence suggests that the nonselective beta-adrenoreceptor blocker propranolol may downregulate VEGF, IGF-1 mRNA, and HIF-1α in hypoxic retina [48]. In clinical studies, propranolol reduced ROP progression, but safety concerns regarding oral administration were expressed in a review from 2023 [49]. In a retrospective nonblinded study with historical controls, treatment with 0.2% propranolol eye drops every sixth hour for an average of 52 days in infants with stage 1 or 2 ROP prevented progression to stage 3 ROP, without adverse effects [50]. However, in most of these infants, the ROP likely would have regressed spontaneously.

Screening for ROP involves repeated, often painful, and stressful retinal examinations using intense light and dilated pupils. In Sweden, only 5.6% of screened infants required treatment in 2021 [51]. A reliable prediction of the risk of sight-threatening ROP may reduce the number of unnecessary examinations. Weight-gain-based algorithms such as WINROP (Weight, IGF-1 [insulin-like growth factor 1], Neonatal, ROP), G-ROP (Postnatal Growth and ROP), CHOP ROP (Children’s Hospital of Philadelphia ROP), ROPScore, CO-ROP (Colorado ROP), and PINT (Premature Infants in Need of Transfusion) ROP were found to have adequate sensitivity but inadequate specificity in a systematic review and meta-analysis in 2021 [52]. Notably, the sensitivity has to be 100%, because no cases of blindness are tolerable.

However, the infants’ weights are not always readily available to ophthalmologists. Therefore, DIGIROP-birth was created based on birth characteristics GA, BW, sex, and postnatal age in 7,609 infants born at GA 24–30 weeks and registered in the Swedish National Registry for ROP (SWEDROP) [53]. DIGIROP-birth estimates momentary and cumulative risks for treatment warranting ROP with a predictive ability comparable to that of weight-based methods. Interestingly, the risk increases up to 12 weeks of postnatal age, irrespective of GA [53]. Adding data on ROP status and age at first diagnosis of ROP to the algorithm resulted in DIGIROP-screen with equal or higher sensitivity and specificity than weight gain-based models [54]. Both DIGIROP-birth and DIGIROP-screen are available for validation and research at www.digirop.com, without cost. When parenteral nutrition for 14 days or more was found to be a strong risk factor for severe ROP in a population of more than 11.000 screened infants, further refinement of the prediction tool resulted in the Revised DIGIROP 2.0, with 100% sensitivity and high specificity, superior to WINROP and G-ROP [16]. At Sahlgrenska University Hospital, Sweden, a digital, standardized web-based protocol has been developed that is linked to SWEDROP, where screening results are automatically recorded, and suggestions for further refinement of the screening examination protocols are provided.

With the increasing number of infants at risk for severe ROP, the burden of ROP screening and treatment has increased, and the availability of specially trained ophthalmologists is poor, especially in low-resource settings. Telemedicine, with bedside wide-field retinal photography and the transfer of images to a specialist for ROP grading, has proven to be a safe and cost-effective complement to ophthalmoscopy [55]. Smartphone imaging may be a screening tool in the future [56]. Machine learning algorithms for automated ROP staging can potentially improve the screening accuracy and reduce the burden of ROP screening [57].

Laser photocoagulation of the avascular retina to halt the production of angiogenic factors such as VEGF has long been the most common treatment for sight-threatening ROP. Over the last decade, the use of easily administered intravitreal injections of anti-VEGF agents has increased worldwide. Randomized controlled trials (RCTs) have compared laser treatment with intravitreal bevacizumab (BEAT-ROP trial) [58], ranibizumab (RAINBOW trial) [59], and aflibercept (FIREFLEYE trial) [60]. In addition, a new agent, conbercept, was compared with ranibizumab in an RCT, and no difference in recurrence rate or PMA at retreatment was found. In a recent meta-analysis of RCTs, the retreatment rate after anti-VEGF treatment was found to be similar to that after laser therapy [61]. The time to recurrence was significantly shorter with laser treatment. Lower (compared to commonly used) doses of anti-VEGF result in reduced time to recurrence, indicating that a shorter period of repeated follow-up examinations is needed [61]. Myopia development is more common after laser [62].

Anti-VEGF agents are valuable alternatives to lasers for the treatment of ROP. They are easily administered and are efficient. However, they leak into the circulation and suppress systemic VEGF levels, which have unknown effects on developing tissues including the brain. Bevacizumab remains in the circulation and suppresses VEGF levels for 2 months. The most common dose is 0.625 mg per intravitreal injection; however, lower doses may be as efficient. Bevacizumab doses as low as 0.002 mg were associated with reduced circulating VEGF levels during the first 4 weeks after treatment. Further studies on doses and long-term consequences are needed. The prolonged period of recurrence risk implies the need for detailed retinal examinations at ages when examinations have become a challenge, and sedation or general anesthesia is often needed. Laser therapy requires general anesthesia more frequently than anti-VEGF treatment, requires considerable skill, and requires more time. Laser therapy will continue to be a valuable therapy for severe ROP as a primary treatment, for the treatment of recurrences after anti-VEGF therapy as rescue therapy, and to reduce the need for intensive follow-up.

ROP increasingly affects infants born during the second trimester of pregnancy and/or in less developed settings lacking oxygen monitoring, as well as programs for ROP screening and treatment. Screening examinations are painful, and laser treatment is destructive. ROP would not occur if preterm births could be prevented. Measures to improve maternal health and supplement women with DHA deficiency may help prolong gestation [63]. For preterm infants, optimizing oxygen delivery, reducing blood sample volumes, providing the mother’s milk, and supplementing with ArA and DHA can help prevent ROP. Screening should be individualized to avoid unnecessary eye examinations in infants at low risk, focusing on those at high risk.

The authors have no conflicts of interest to declare.

This study was supported by the Swedish Research Council (2015-00810, 2016-01131, and 2022-01562), the Swedish state under the agreement between the Swedish government and the county councils: the ALF agreement (ALFGBG-71971 and ALFGBG-812951), The Wallenberg Clinical Scholars (KAW 2018.0310), and De Blindas Vänner.

A.-L.H., A.H., and L.E.H.S. contributed to the design of the study. A.L.H. drafted the initial manuscript. A.-L.H., A.H., and L.E.H.S. critically revised the manuscript, agreed to be fully accountable for ensuring the integrity and accuracy of the work, and have read and approved the final manuscript.