Evaluation and modification of French screening guidelines for retinopathy of prematurity

Introduction

Retinopathy of prematurity (ROP) is a leading cause of blindness and visual impairment in children (Gilbert 2007). It is caused by local retinal ischemia with the subsequent development of retinal neovascularization in premature children under oxygen therapy (Sapieha et al. 2010). Worldwide, 15.8% of premature babies develop ROP and 3.0% remain with some level of visual impairment (Blencowe et al. 2013). Up to now, 60% of cases occur in middle- to high income regions (Blencowe et al. 2013), but the incidence in developing countries is increasing (Gilbert et al. 2005).

Several treatment options were shown to improve outcome. Laser treatment in type 1 ROP reduced unfavourable outcome and improved visual acuity (Early Treatment For Retinopathy Of Prematurity Cooperative Group 2003; Early Treatment for Retinopathy of Prematurity Cooperative Group et al. 2010). Intravitreal anti-VEGF injection has also been proposed as a therapeutic option in selected ROP cases (Mintz-Hittner et al. 2011; Stahl et al. 2019). Early detection of ROP and timely intervention improved outcome in premature infants (Good 2004), pushing several countries to introduce screening programs (Holmström et al. 2015; Fierson et al. 2018).

One of the aims of a screening guideline is to define screening thresholds that come along with a high specificity and sensitivity. This would detect preterm infants at risk of developing severe ROP that needs treatment without exposing the rest to unnecessary screening. Screening guidelines should differ between countries especially between high and low-income countries since ROP depends on neonatal care and oxygen supplementation and, to a lesser extent, on genetic variation (Lasabova et al. 2018). Therefore, there is no one-fit-all screening program (Gilbert et al. 2005; Gaynon 2006; Shastry 2010; Fierson et al. 2013; Wilson et al. 2013). In France, screening followed up to 2018 the recommendations published by Holmström et al. 1993 and included all preterm infants born at gestational age (GA) ≤32 weeks.

The aim of the present study is to evaluate the current screening protocol and propose possible modifications to it.

Methods

In this retrospective multicenter study we enrolled all premature infants born at GA of ≤32 weeks over a 7-year period between November 2011 and November 2018. Six neonatal centers were included: University Hospital Center of Caen, Cotentin Public Hospital in Cherbourg-en-Cotentin, Hospital Center Sud Francilien in Corbeil-Essonnes, Le Havre Hospital Group, Louis-Mourier Hospital in Colombes and Robert Debré Hospital in Paris, in association with the Fondation Ophtalmologique Rothschild (FOR), Paris. Only children that had a complete screening record were included in the study. Institutional review boards of all centers approved the retrospective analysis of data. The study was approved by the ethical committee of the FOR and adhered to the tenets of Helsinki. All parents had to sign an informed consent for the examination under general anaesthesia and for treatment, in line with the standard procedure.

The following data was collected for every screened infant: GA, birth weight (BW), age at first screening, age at ROP detection, number of eye examinations, the need for treatment and age at treatment initiation.

Classification and treatment of ROP followed the recommendations of the Early Treatment for Retinopathy of Prematurity (ETROP) trial (Good 2004). Type 1 ROP was defined as ROP of any stage with plus disease in zone I, stage 3 in zone I, and stage 2 or 3 with plus disease in zone II. Type 2 ROP was defined as stage 1 or 2 without plus disease in zone I or stage 3 without plus disease in zone II. Premature babies were classified as no ROP, and ROP, including type 1 and type 2 ROP. Treatment was initiated in case of type 1 and children underwent treatment under general anaesthesia.

Screening initiation was advised at 4 weeks postnatal age (PNA) for all premature infants except for infants born before 28 weeks, in whom first screening was set at 31 weeks postmenstrual age (PMA) (Holmström et al. 2015). Follow-up time was 1 week or less for zone I and stage 3 zone II, 1–2 weeks for stage 2 zone II or regressing zone I, and 2 weeks for the remaining ROP cases. Screening was discontinued once the retina was fully vascularized.

All infants were dilated using tropicamide 0.5% (mydriaticum 0.5%, Laboratoires Théa, Clermont-Ferrand, France) and underwent a fundus examination using a contact wide-angle fundus camera (RetCam, Clarity Medical Systems USA) after the instillation of anaesthetic eye drops and gel or indirect ophthalmoscopy with a 20-diopter lens.

Results

In total 2246 children were included. Mean GA was 28.9 ± 2.0 weeks (range: 23.6–32) and mean BW was 1141.1 ± 332.0 g (range: 450–2360). Thirty-four percent of infants (755 infants) had a GA <28 weeks and 53.7% were boys. A total of 5288 exams were conducted with an average of 2.4 ± 2.0 exams.

Retinopathy of prematurity (ROP) was found in 683 infants (30.4%), of whom 145 (6.5%) had type 2 ROP and 58 (2.6%) had type 1 ROP and were referred for treatment. Table 1 reports on GA and BW according to the presence or absence of ROP and according to ROP status. Laser treatment was the main treatment option. Forty-nine infants received laser only, 6 anti-VEGF followed by laser and 3 anti-VEGF only. Table 2 shows the distribution of ROP, type 1 and 2 ROP among GA groups.

Among infants that developed type 1 or 2 ROP none had a GA 31 weeks or more. Two infants with GA of 30 weeks developed type 2 ROP and one of them progressed to type 1 ROP requiring prophylactic laser treatment. Only one infant with a weight above 1250 g developed type 2 ROP that regressed spontaneously. Any inclusion of higher GA or BW will not improve the sensitivity of the screening. Table 3 shows sensitivity and specificity of different GA and/or BW cutoffs to diagnose a type 1 ROP, and the number of infants needed to screen in order to detect a type 1 ROP.

Discussion

The aim of the present multicentric retrospective study was the evaluation of the ROP screening protocol in France. This is, to our knowledge, the largest study to evaluate the ROP screening protocol in France.

Thirty-four percent of 2246 infants born with a GA ≤32 weeks developed any stage of ROP. Six point 5% developed type 2 ROP and 2.6% required treatment due to type 1 ROP. None of the infants with a GA of 31 weeks or more developed type 1 or 2 ROP. Two infants with GA of 30 weeks developed type 2 ROP and one of them needed laser treatment because of progression to type 1 ROP.

Seventy-three percent of infants had the first fundus examination before the sixth postnatal week and 91% before the eighth postnatal week. The evaluation of the Swedish ROP screening protocol, which was based on the Swedish national register for retinopathy of prematurity found similar results. Inclusion criteria was a GA < 32 weeks (Holmström et al. 2012).

Both GA and BW were associated with the development of type 2 ROP and the need for treatment (Early Treatment For Retinopathy Of Prematurity Cooperative Group 2003; EXPRESS Group 2010; Holmström et al. 2012). Therefore, GA and BW should be considered when evaluating a screening protocol. However, GA depends on an adequate dating of the pregnancy and a 1 or 2 week error can wrongly exclude an infant from ROP screening (EXPRESS Group et al. 2009). In our study, 7 infants with a GA of 29 and 30 weeks developed type 2 ROP and 3 infants developed type 1 ROP. Based on our data we evaluated in Table 3 several cutoffs for GA alone, BW alone and a combination of both. We found that by decreasing the upper limit of the screening program by 1 week to GA 31 weeks, sensitivity of the screening protocol stays at 100% while the number of screened children decreases by 13.4% (1946 vs. 2246 infants) and the number of fundus examination decreases by 7.7% (4880 vs. 5288 FE). This finding is in accordance with the Swedish, British, American, and Canadian screening guidelines (Wilkinson et al. 2008; Holmström et al. 2012; Jefferies, & Canadian Paediatric Society, Fetus and Newborn Committee 2016; Fierson et al. 2018). Furthermore, we found that by combining a GA ≤ 30 weeks and a birthweight ≤1250 g (de la Cruz Bértolo et al. 1999) we could further increase the specificity of the screening protocol and decrease the number of infants to screen. Despite this, we recommend as a first step to decrease the upper limit of the screening by 1 week to include infant ≤31 weeks of GA and further evaluate the latter combination (GA ≤ 30 weeks or birthweight ≤1250 g).

Most ROP screening guidelines recommend to start screening at 31 weeks GA for children born with a GA of 27 weeks or less, or to start at 4–6 weeks PNA for more mature infants (Wilkinson et al. 2008; Holmström et al. 2012; Fierson et al. 2013). The same recommendations were applied to the French national guideline. The results of the present study are in accordance with these recommendations. The youngest type 2 ROP was detected at a PMA of 30 weeks but no treatment was initiated before the PMA of 33 weeks or a PNA of 6 weeks. Therefore, we recommend initiating screening at PNA of 6 weeks for infants with GA of 26 weeks or more and at 31 weeks for younger infants. Those recommendations are in concordance with the findings of Holmström et al. (2019).

The study is limited by its retrospective nature and by the fact that it is limited to a single country, therefore, the results cannot be generalized to other countries. Furthermore, the study is hospital-based and not population-based since no national registry exists which might affect study population. The inclusions of many centers aimed at increasing the representativity. Furthermore, the use of tropicamide alone might have limited in some cases the diagnosis of mild cases of ROP especially in zone III because of incomplete pupillary dilation. We do not believe that any type 1 or 2 ROP have been missed.

In conclusion, this study demonstrated that it is safe to decrease the screening of premature infants to < 31 weeks of GA and to start screening at 6 weeks of PNA for infants with a GA of 26 weeks or more and at 31 weeks PMA for younger infants. Referral of severely ill more mature infants should be left to the discretion of the neonatologist. Further studies should be done to evaluate the screening cutoff of GA ≤ 30 weeks or a BW ≤1250 g, which seems to have a better specificity.