Introduction

Retinopathy of prematurity (ROP), the leading cause of infants’ blindness all over the world [1], is a disorder of the retinal vasculature, with pathologic vessels growing into the vitreous instead of the retina [2]. Before introducing the use of anti-vascular endothelial growth factor (anti-VEGF) agents in infants, clinical trials had confirmed the efficacy and safety of ablating the avascular peripheral retina to achieve the regression of preretinal neovascularization and prevent ensuing fibrovascular retinal detachments [2]. Treatment-warranted ROP was defined as a set of characteristics that resulted in a 50% chance of an unfavorable outcome, and cryotherapy was the first treatment modality that proved to lower that risk when compared to no treatment. [3]. Subsequently, ET-ROP study showed the efficacy of the laser treatment for a less severe form of treatment-warranting ROP (type 1 ROP), in which the risk of an unwanted outcome was approximately 15% [4].

The BEAT-ROP study [5] was the first prospective, multicenter, stratified, randomized controlled trial (RCT) that attempted a comparison of efficacy between intravitreal bevacizumab (IVB) monotherapy and conventional laser photocoagulation (LPC) for zone 1 or zone 2 posterior stage 3 + ROP. The study reached the conclusion that IVB was beneficial for infants with stage 3 + ROP for zone 1(P = 0.003), but not for zone 2 disease (P = 0.270). Although BEAT-ROP encouraged many clinicians worldwide to use IVB as first-line treatment for ROP, IVB still remains an off-label modality. A second RCT evaluating ranibizumab followed: RAINBOW study [6], an open-label, multicenter, randomized, three-arm, parallel group, superiority trial, assigned infants with treatment-requiring ROP (TR-ROP) in three groups of 0.2 mg intravitreal ranibizumab (IVR), 0.1 mg IVR and LPC, concluding that ranibizumab 0.2 mg is superior to LPC, while having less unfavorable ocular outcomes. In September 2019, the European Medicines Agency (EMA) approved ranibizumab 0.2 mg as an on-label treatment for infants with TR-ROP. The RAINBOW study provided evidence on the drug’s efficacy as well as on short-term safety issues [6]. However, more information is needed concerning the adverse events (AEs), complications, and unfavorable functional and structural outcomes of the two treatment modalities in the long term.

The aim of this systematic review was to capture the current knowledge regarding the adverse events, the complications and the unfavorable structural and functional outcomes of intravitreal anti-VEGF agents and LPC as treatment modalities for TR-ROP, so as to guide clinical ophthalmologists in their choice of the preferred treatment modality for each case of TR-ROP.

Methods

Study characteristics

This study is a systematic review of RCTs and observational studies that compared intravitreal anti-VEGF injections and LPC as treatment modalities in infants with TR-ROP, in terms of adverse events, complications, and unfavorable structural and functional outcomes. Literature search was carried out until 25/7/2020 without restrictions. The study had been registered to PROSPERO with the following registration number: CRD42020189408.

Eligibility criteria

Inclusion criteria

  • The included studies were either RCTs or observational studies.

  • The participants of each study were infants with ROP that required treatment.

  • The included studies evaluated one of the following intravitreal anti-VEGF agents as monotherapy: ranibizumab, bevacizumab, aflibercept, pegaptanib, conbercept, and compared its adverse events, complications, or unfavorable outcomes with one of the following types of LPC, also used as a monotherapy: diode laser, argon laser, Yttrium aluminum garnet (YAG) laser.

Exclusion criteria

  • The studies that did not involve humans as subjects.

  • The studies that reported early-stage outcomes that are also reported in the complete version of the study. In this case, the most complete version was included to avoid duplication of our results.

  • The studies that reported the outcome of the interventions conducted to treat an adverse event of a treatment modality for ROP, rather than comparing the outcomes of intravitreal anti-VEGF as monotherapy with LPC, also used as a monotherapy, as treatment modalities for ROP.

Study outcomes

The outcomes of interest were the comparison of adverse events, complications, and unfavorable structural and functional outcomes between intravitreal anti-VEGF treatment and LPC, and were categorized as follows:

  1. 1.

    Refractive Errors and Biometry Parameters This subsection evaluated refractive spherical power, spherical equivalent (SE), cylinder power, best-corrected visual acuity (BCVA), rates of myopia and high myopia, rates of anisometropia, rates of astigmatism and biometric results (e.g., anterior chamber depth (ACD), lens thickness (LT), axial length (AL), and central choroidal thickness (CCT)).

  2. 2.

    Adverse events, complications, and unfavorable outcomes This subsection evaluated both ocular and systemic unfavorable outcomes: rates of retinal detachment, vitreous hemorrhage, macular dragging, retinal fold, macular ectopia, endophthalmitis, ocular inflammations, cataract formation, glaucoma, corneal opacity requiring transplantation, and death.

  3. 3.

    Disease Recurrence/Disease Regression/Need for retreatment There is not a universal definition of ROP recurrence and its difference from treatment failure. Therefore, the definition of ROP recurrence, treatment failure, or treatment success that each study used, are provided when the respective results are reported.

  4. 4.

    Neurodevelopmental Outcomes Reported Bayley-3 scores of cognition, language and motor composite were assessed.

  5. 5.

    Optical coherence tomography (OCT) and optical coherence tomography angiography (OCTA) Measurements OCT assessment of the posterior part of the eye, including inner foveal thickness (IFT), outer foveal thickness (OFT), subfoveal choroid thickness (CT), foveal avascular zone (FAZ), foveal vessel density (VD), parafoveal VD, perifoveal VD and macular volume, is presented.

  6. 6.

    Other reported outcomes Comparison of serum-free VEGF levels, serum insulin-like growth factor–1 (IGF-1) levels and cardiovascular assessment between infants with TR-ROP that were treated with either intravitreal anti-VEGF or LPC, are included in this subsection.

Searches and search strategy

The following databases were searched: PubMed, Cochrane Central Register of Controlled Trials (CENTRAL), Scopus, EMBASE and Trip Database. A search in the gray literature, such as ClinicalTrials.gov and in available conference proceedings of American Academy of Ophthalmology (AAO), European Paediatric Ophthalmological Society (EPOS), American Association for Pediatric Ophthalmology and Strabismus (AAPOS) and EURETINA, has also been conducted. Furthermore, to ensure a systematic search of the existing literature, reference lists of any included study were scanned rigorously to find eligible studies that the search may had missed. The basic search terms that corresponded to each element of the research question were used for the search strategies in all databases. Every step of the systematic review process was performed by two independent researchers. In the case of disagreements, the final decision was determined by the senior author. After duplicates were removed, all studies were searched by title and abstract. Studies that did not satisfy the research question were excluded. Full-text screening was performed in the remaining, potentially eligible studies. Whenever a study had been published in different versions, the latest and most complete version was selected. The PRISMA flow diagram is presented in Fig. 1 [7].

Fig. 1
figure 1

The PRISMA flow diagram of this systematic review. (from: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009) Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6: e1000097. https://doi.org/10.1371/journal.pmed1000097)

Full size image

Risk-of-bias (quality) assessment

The risk-of-bias (quality) assessment for the RCTs and non-randomized studies of interventions (NRSI) was conducted with the use of the RoB 2.0 [8] and ROBINS-I [9] tools, respectively.

Patient consent form

No patient consent forms were needed as this is a systematic review.

Results

Refractive errors and biometry

Comparison between three groups (IVB, IVR and diode LPC)

The observational studies that compared the refractive errors and biometric measurements between IVB, IVR and diode LPC groups reported no differences between the three groups [10,11,12], as it is shown in Table 1.

Table 1 Observational studies comparing refractive errors and biometric measurements between IVB, IVR, diode LPC treatment modalities for TR-ROP
Full size table

Comparison between 2 groups

All RCTs [13, 14] and observational studies [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36] that compared intravitreal anti-VEGF injections with LPC for refractive errors and biometry are shown in Table 2. Raghuram et al. [15] reported a more myopic median refraction and a greater rate of myopia ≤ − 0.25 diopters (D) in diode LPC-treated eyes than in IVB-treated eyes at 18–24 months of age (P = 0.020, P = 0.040, respectively). These outcomes are at moderate overall risk of bias and of the highest quality among the observational studies.

Table 2 Randomized clinical trials and observational studies comparing refractive errors and biometric measurements between anti-VEGF and LPC treatment modalities for TR-ROP
Full size table

Adverse events, complications, and unfavorable structural outcomes

Comparison between three groups

RCTs

In the RAINBOW trial [6], unfavorable structural outcomes, defined as structural abnormalities that have potential effects on visual acuity were found in all three arms: 1 infant in the 0.2 mg IVR arm, 5 in the 0.1 mg IVR arm, and 7 infants in the LPC arm. At 24 weeks after initial treatment, death, serious adverse events (SAEs) and non-serious systemic AEs were similar between the treatment groups as 4 deaths occurred in each group. In the 0.2 mg IVR group, one infant had a moderate cataract formation, while in the 0.1 mg IVR group, one infant developed endophthalmitis in one eye. These outcomes are at low overall risk of bias.

Comparison between 2 groups

The results of all the RCTs [5, 6, 13, 37,38,39], and the observational studies [10, 19, 22, 27, 31, 33, 36, 40,41,42,43,44,45,46,47], that compared the adverse events, complications, and unfavorable outcomes between intravitreal anti-VEGF injections and LPC, are displayed in Table 3.

Table 3 Randomized clinical trials and observational studies comparing the adverse events, complications, and unfavorable outcomes between intravitreal anti-VEGF and LPC treatment modalities for TR-ROP
Full size table

Disease recurrence/disease regression/need for retreatment

In the RAINBOW study [6], treatment success, defined as alive and without treatment switch and unfavorable structural outcome or active ROP at day 169, was reported as odds ratio (OR) and 95% Confidence Interval (CI) in pairwise comparisons of the treatment arms: OR = 2.19 (CI 0.99 to 4.82; P = 0.050) of 0.2 mg.

IVR compared to LPC, and OR = 1.57 (95% CI: 0.76 to 3.26) of 0.1 mg IVR compared to LPC. This outcome is at low overall risk of bias.

The results of all the other RCTs (except RAINBOW), and the observational studies, that compared the rates of disease recurrence, disease regression, need for retreatment between IVB, IVR and LPC groups, are displayed in Table 4 [5, 6, 10, 13, 19, 22, 29, 33, 36,37,38,39,40, 42, 44,45,46, 48,49,50,51].

Table 4 Randomized clinical trials and observational studies that compare disease recurrence, disease regression and need for retreatment between intravitreal anti-VEGF and LPC treatment modalities for TR-ROP
Full size table

Neurodevelopmental outcomes

Observational studies

The observational studies [15, 17, 22, 36, 44, 47, 52,53,54,55] that compared neurodevelopmental outcomes between intravitreal anti-VEGF and LPC for TR-ROP were of moderate or serious overall risk of bias and found many similarities between the two treatment modalities, while in case a significant difference existed, LPC had the better results in terms of neurodevelopmental outcomes (Table 5).

Table 5 Observational studies comparing neurodevelopmental outcomes between intravitreal anti-VEGF and LPC treatment modalities for TR-ROP
Full size table

In the RCT of Kennedy et al. [56], 16 infants of the BEAT-ROP study were evaluated for medical and neurodevelopmental outcomes at 18–28 months corrected age (CA). The authors reported similar results in all outcomes when comparing the two treatment groups at follow-up (median cognitive score P = 0.060, language score P = 0.180, motor composite score P = 0.220, gross motor function level P = 0.850, rate of cerebral palsy P = 1.000, median CA P = 0.100, median weight percentile for age P = 0.270, median length percentile for age at follow-up P = 0.390, median head circumference percentile for age at follow-up P = 0.460). These outcomes are at medium overall risk of bias. This is the only RCT that we found concerning the comparison of neurodevelopmental outcomes between intravitreal anti-VEGF and LPC treatment modalities for TR-ROP.

OCT and OCTA measurements

The observational studies [16, 20], that compared the macular OCT and OCTA measurements between intravitreal anti-VEGF and LPC in infants with TR-ROP were of critical overall risk of bias and in case a significant difference existed, IVB treatment showed lower mean foveal, parafoveal, perifoveal, and inner foveal thickness (Table 6).

Table 6 Observational studies comparing OCT and OCTA measurements between intravitreal anti-VEGF and LPC treatment modalities, in infants with TR-ROP
Full size table

Finally, studies that compared other reported outcomes [54, 57,58,59], such as serum-free VEGF levels, serum Insulin-like growth factor 1 (IGF-1) levels, tricuspid E-wave values, odds of returning to respiratory baseline by 48 h, number of diagnoses at time of discharge, hospitalization days, days for oxygen requirement, and duration of hospitalization, are summarized in Table 7.

Table 7 Studies comparing different reported outcomes between intravitreal anti-VEGF and LPC/PRP as treatment modalities for TR-ROP that are useful for clinical ophthalmologists
Full size table

Discussion

This systematic review of the literature regarding treatment of ROP revealed that higher quality studies concluded that LPC leads to greater rates of myopia than intravitreal anti-VEGF treatment, while the rate of adverse events and of unfavorable neurodevelopmental outcomes is similar. However, there was controversy among the included studies concerning the rate of ROP recurrence. Studies agree on findings regarding the refractive outcome, the rates of adverse events, and the neurodevelopmental outcomes, while they differ in disease recurrence rates. Notably, authors defined their outcomes differently and results are thus not directly comparable. Furthermore, most studies are observational and of moderate risk of bias, so safe conclusions cannot be drawn.

The RCTs [14] and the observational studies [15, 19] of the highest quality concluded that eyes treated with LPC developed more myopic refraction than the eyes treated with intravitreal anti-VEGF injections, a finding shared with the majority of observational studies as well [16, 20,21,22,23, 27,28,29, 33,34,35,36]. Rates of adverse events, complications and unfavorable outcomes were similar between intravitreal anti-VEGF agents and LPC in high quality RCTs like RAINBOW [6] and BEAT-ROP [5]. Results in lower-quality RCTs [13, 37,38,39] and observational studies [10, 19, 22, 27, 31, 33, 36, 40,41,42,43,44,45,46,47] did not generally differ.

Disease recurrence, disease regression and need for retreatment were similar between intravitreal anti-VEGF injections and LPC in the RAINBOW study [6], the highest-quality RCT available in our systematic review concerning that outcome. Some of the lower-quality RCTs concluded that the rate of disease recurrence was greater in the intravitreal anti-VEGF injection group [37,38,39], while others like BEAT-ROP [5] concluded the exact opposite. Finally, other RCTs [13] agreed with the findings of RAINBOW [6]. Some of the observational studies of higher quality concluded that rates of ROP recurrence were greater in the intravitreal anti-VEGF injection group [19], while others reported similar ROP recurrence rates between the two treatment groups [29, 45, 48], and others concluded that rates of ROP regression were greater in eyes treated with intravitreal anti-VEGF injections [50].

The only RCT that conducted a comparison of neurodevelopmental outcomes between intravitreal anti-VEGF and LPC, which was at medium overall risk of bias, found no differences in the neurodevelopmental outcomes between the two treatment modalities [56]. Most of the observational studies of higher quality also report this result [15, 36, 52, 54]. However, some high-quality observational studies supported that the intravitreal anti-VEGF group had worse neurodevelopmental outcomes in the Language-Social domain Developmental Quotient (DQ) at 18 months CA [53], motor composite score at 18 months CA [55], or more neurodevelopmental disabilities [55]. This may be due to the fact that there is only one RCT comparing neurodevelopmental outcomes between intravitreal anti-VEGF and LPC, and due to the observational design of the other studies.

The studies that reported OCT and OCTA measurements were of very low quality due to critical overall risk of bias and generally concluded that foveal thickness was lower, while mean foveal avascular zone (FAZ) was higher in the IVB group [16, 20].

Finally, an RCT with high overall risk of bias that compared the serum levels of free VEGF and IGF-1 between IVB and LPC treatment groups found lower serum levels of these two biochemical markers in the IVB group [57]. The observational studies that reported different outcomes related to the wide spectrum of adverse events were of very low quality and therefore analyzing these studies is out of the scope of this systematic review [54, 58, 59].

Intravitreal anti-VEGF treatment leads to lower rates of myopia, having a similar rate of adverse events and unfavorable neurodevelopmental outcomes as LPC. Therefore, intravitreal anti-VEGF treatment seems to have the preferable outcomes overall. However, no safe conclusions can be drawn concerning the rates of disease recurrence. More primary studies need to be conducted to give a definite answer to which treatment modality has greater rates of disease recurrence and to verify the aforementioned findings.

Strengths and weaknesses

This study captured the comparison of all the adverse events, complications, and unfavorable structural and functional outcomes between intravitreal anti-VEGF injections and LPC that have been reported in the literature concerning the treatment of TR-ROP. Due to its systematic nature, this study aimed to summarize all current knowledge in view of facilitating clinical decisions. The adverse event comparison was stratified in sections that correlated with clinical significance, distinguishing different clinical entities of interest. Finally, the critical appraisal of the included studies was thoroughly conducted by assessing the risk of bias of each outcome of each individual study. Two independent researchers conducted the risk-of-bias assessment to limit bias as much as possible.

On the other hand, it should be mentioned that there is high heterogeneity between the included studies due to the very wide spectrum of the outcomes of our interest. This is the reason for not performing a quantitative synthesis of the results (meta-analysis).

Furthermore, some reported outcomes have been defined in different ways from study to study. For example, in the result section of disease recurrence or regression and need for retreatment, RAINBOW study defined treatment success as: alive and without treatment switch and unfavorable structural outcome or active ROP at day 169 [6], while BEAT-ROP study defined treatment failure as: the recurrence of neovascularization in one or both eyes arising from the retinal vessels and requiring retreatment by 54 weeks’ postmenstrual age [5]. That is, an obstacle encountered in many other outcomes of interest and therefore the included studies were synthesized in a descriptive way to succeed in providing the reader with conclusions that reflect everyday clinical practice. Lastly, an endogenous limitation that is related to the pathophysiology of ROP is that it is not clear if ROP recurrence is an adverse event or a failure of the respective treatment modality; therefore, ROP recurrence and adverse events were analyzed in different result sections.

Clinical implications

Almost all the included studies, and most importantly, the higher quality studies like RAINBOW [6], agreed that LPC treatment leads to greater refractive errors and greater rates of myopia than intravitreal anti-VEGF treatment modalities. Similarly, almost all included studies reported no differences in the rates of adverse events, complications, unfavorable structural outcomes, and unfavorable neurodevelopmental outcomes between LPC and anti-VEGF. Findings of different high-quality studies, in terms of ROP recurrence/regression and need for retreatment, are overall controversial. This may be partly due to different definitions of ROP recurrence in the different studies that investigated this outcome or due to the vague nature of the outcome itself, because some authors interpret ROP recurrence as an adverse event, while others as failure of the applied treatment.

Eyes treated with LPC for TR-ROP tend to have more myopic refraction than the eyes treated with intravitreal anti-VEGF injections, while the rates of adverse events, complications, unfavorable structural outcomes, and unfavorable neurodevelopmental outcomes between the two treatment modalities seem to be similar. The RAINBOW study is designed to follow up participants until the age of five years so hopefully more data will become available soon [6]. There is a need for more primary studies, and a consensus needs to be agreed upon concerning the definition of the outcomes of interest. This would help in lowering the heterogeneity of future systematic reviews and in providing clinical ophthalmologists with more precise and more high-quality evidence about the comparison of intravitreal anti-VEGF injections and LPC in the treatment of TR-ROP.