Tese de Doutoramento Medicina - Investigação Clínica 2022 Faculdade de Ciências Médicas, Universidade NOVA de Lisboa

Retinopathy of prematurity (ROP) is a disease of the retina associated with preterm birth, particularly very low birth weight/very preterm infants, in which retinal blood vessels fail to grow and develop normally. In spite of major advances in the understanding of postnatal risk factors for the development of ROP, it continues to be one of the leading causes of childhood blindness in high-income countries. With the improved access to neonatal technologies and survival of premature infants, the incidence of severe ROP has begun to emerge as a major cause of blindness in low-income countries, as well. ROP incidence varies considerably from country to country and even from neonatal unit to unit. While most cases are mild and resolve spontaneously, a small proportion does progress to severe ROP which, left untreated, can result in irreversible vision loss. To avoid this grim outcome, screening programs for detecting treatment-requiring ROP have been established worldwide. Infants at risk for ROP are identified using recommended birth weight (BW) and gestational age (GA) criteria. According to the current Portuguese Pediatric Society, all infants with GA ≤32 weeks (w), BW ≤ 1500 g, BW<2000 g with prolonged exposure to oxygen and selected infants that are at higher risk of ROP for having severe disease or who have undergone major surgery in the opinion of the attending neonatologist or pediatrician should undergo screening for ROP. Although the currently recommended guidelines have high sensitivity for detection of ROP, their specificity is low, as successive retinal examinations are necessary to reliably detect severe ROP in time for intervention. This is done at the cost of repeatedly subjecting infants, many of whom develop no or mild retinopathy to stressful, resource-intensive, serial diagnostic eye examinations. Due to the low diagnostic yield of current screening guidelines, the identification of additional predictive factors for ROP may help identify infants at high ROP risk and to best manage and allocate resources for those at higher risk while sparing those at lower risk. The main objective of this thesis was to establish a set of strategies that could help improve screening of preterm infants in Portugal. To this end, 3 retrospective and 2 prospective cohort studies were undertaken. In recent years, prediction algorithms have proven to be useful in some populations. To test their applicability in Portuguese cohorts, two retrospective studies were performed. The first study performed a comparison between the WINROP model and the G-ROP model for their sensitivity and specificity for Type 1 ROP and concluded that both models had similar sensitivities but none was suitable as an alternative to the current screening modality. If restricted to GA<30 weeks both models detected all Type 1 ROP infants. The second study aimed to determine the efficacy of the DIGIROP-Birth model in detecting treatment-requiring retinopathy of prematurity (TR-ROP). The obtained optimal cut-off point out to achieve 100% sensitivity for TR-ROP was 0.0016 but with moderate accuracy in the area under the curve (0.70). The number of infants requiring screening would have decreased 27.2% if this model was applied. To search for potential predictive factors for ROP development, 2 prospective studies were designed. The first cohort study assessed the association between hyperglycemia, total glycated albumin and ROP. This study revealed that hyperglycemia, but not glycated albumin, is a significant risk factor for ROP, overpowering all other recognized risk factors.The second prospective case control non-interventional study was designed to investigate whether serum Placental Growth Factor (PGF) levels differed between infants who developed severe ROP versus those without a ROP diagnosis, and to further assess its predictive value on ROP development. This study uncovered a statistically significant difference in serum PGF levels, higher at birth and at four weeks of life for infants that developed severe ROP compared to those without a ROP diagnosis. Thus, serum PGF may be important not only as a potential therapeutic target but may also serve as an early potential marker for severe ROP. Finally, the last section of this project focused on a retrospective case series of 475 infants to evaluate the incidence and risk factors for ROP. The incidence of ROP in the analyzed cohort fell within the range reported in studies from other high-income countries. In this study, hyperglycemia again overpowered all the other risk factors. This study additionally confirmed that although rare, more mature infants are also at risk for severe ROP. Older GA share the same group of risk factors but prolonged oxygen seems to play a bigger role. Lastly, this study concluded that the current Portuguese screening guidelines encompassed all infants that developed TR-ROP. Taken together, these findings confirm that prediction models cannot be used in new settings without validation and even algorithms that were validated in large cohorts can underperform when applied in another setting. At the moment, the analyzed algorithms are not able to outpace current screening guidelines however they are easy to use and might complement and facilitate present screening programs. Hyperglycemia as a risk factor overpowered all the other studied factors and merits further careful consideration, as it is a modifiable parameter that might improve ROP outcome. Also, the analyzed cohorts share some of the characteristics of those reported in larger studies from other high-income countries, as such, their data might have some applicability neonatal units included in the study. Lastly, it was uncovered for the first time, a difference in serum PGF levels between infants with severe ROP and those without a ROP diagnosis. Thus, serum PGF may be important not only as a potential therapeutic target but may also serve as an early potential marker for severe ROP that might help stratify risk profiles and optimize ROP screening. For PGF to be validated as an early marker for severe ROP, these findings need to be further evaluated in a larger cohort. In conclusion, these results provide insight on how to better incorporate predictive algorithms into everyday ROP screening, corroborate the importance of hyperglycemia as a potentially modifiable risk factor for ROP and lastly, if validated as an early marker for severe ROP, PGF might help to develop new strategies for ROP diagnosis and treatment