OBJECTIVES. The goals were to isolate and to estimate the genetic susceptibility to retinopathy of prematurity.
METHODS. A retrospective study (1994–2004) from 3 centers was performed with zygosity data for premature twins who were born at a gestational age of ≤32 weeks and survived beyond a postmenstrual age of 36 weeks. Retinopathy of prematurity was diagnosed and staged by pediatric ophthalmologists at each center. Data analyses were performed with mixed-effects logistic regression analysis and latent variable probit modeling.
RESULTS. A total of 63 monozygotic and 137 dizygotic twin pairs were identified and analyzed. Data on gestational age, birth weight, gender, respiratory distress syndrome, retinopathy of prematurity, bronchopulmonary dysplasia, duration of ventilation and supplemental oxygen use, and length of stay were comparable between monozygotic and dizygotic twins. In the mixed-effects logistic regression analysis for retinopathy of prematurity, gestational age and duration of supplemental oxygen use were significant covariates. After controlling for known and unknown nongenetic factors, genetic factors accounted for 70.1% of the variance in liability for retinopathy of prematurity.
CONCLUSION. In addition to prematurity and environmental factors, there is a strong genetic predisposition to retinopathy of prematurity.
Retinopathy of prematurity (ROP) is a disease process that results from disruption of the normal vascularization of the developing retina. The retinal blood supply is derived from both the choroidal and retinal circulations. The choroidal circulation is completed by ∼20 weeks of gestation, when the development of the retinal circulation is only beginning.1 Vasculogenesis in the fetus is regulated by several factors, including vascular endothelial growth factor and various regulatory cytokines.1–4 In prematurely delivered neonates, fluctuations in oxygen tension and other undetermined variables may affect these factors adversely and may result in incomplete and abnormal neovascularization of the retina.5 The outcome, despite early detection and intervention, may be retinal detachment and blindness.6–8
ROP is most prevalent and severe in neonates with extremely low birth weight (BW),9–11 with overall incidence rates estimated to be as high as 68% among infants born at <1251 g and 93% among infants born at <750 g.12 Given its prevalence and considerable morbidity, investigators have attempted to identify and to target significant contributory factors (eg, supplemental oxygen) in efforts to prevent and to treat this disease.5,9,12–16 Despite these measures, ROP remains a common problem in the NICU.9 This suggests the possible involvement of other nonenvironmental influences and the need to identify them. We hypothesized that genetic factors play a major role in predisposing neonates toward developing ROP. The goal of our present study was to analyze a cohort of preterm monozygotic and dizygotic twin pairs to determine and to estimate the genetic susceptibility.
Data on premature twins born at ≤32 weeks of gestation between January 1, 1994, and December 31, 2004, including zygosity information, were collected from 3 centers (ie, the Karolinska Institute, the University of Connecticut, and Yale University). We included only infants who survived beyond a postmenstrual age of 36 weeks, with complete data on the variables necessary for analysis. The institutional review boards of all participating centers approved the contribution of data to this study.
The zygosity of each twin pair was determined through histopathologic examination of the placenta, with confirmation from gender concordance or discordance. ROP was defined as the incomplete or abnormal vascular proliferation of the retina, as determined by experienced pediatric ophthalmologists at each institution, and was staged according to the criteria established by the International Committee for Classification of ROP.17 All stages of ROP were included in the analysis. Respiratory distress syndrome (RDS) was defined as the presence of respiratory distress with an oxygen requirement in the first 6 hours of life, accompanied by a characteristic chest radiograph. Bronchopulmonary dysplasia (BPD) was defined as the need for supplemental oxygen at postmenstrual age of 36 weeks, in association with characteristic radiographic changes.18 All levels of severity of BPD were included in the analysis. The duration of mechanical ventilation was defined as the total number of days that the infant, while hospitalized, required positive pressure ventilation. Positive pressure ventilation included high-frequency ventilation, synchronized intermittent mandatory ventilation, synchronized nasal intermittent positive pressure ventilation, and/or nasal continuous positive airway pressure. The duration of oxygen use was defined as the total number of days that the infant, while hospitalized, required the use of supplemental oxygen (>21%).
Demographic data were analyzed by using Student's t test, Wilcoxon rank-sum test, or χ2 analysis, where appropriate. Mixed-effects logistic regression (MELR) analysis was performed to identify the impact of putative risk factors on ROP. The covariates used in the model included male gender, gestational age (GA), BW, RDS, duration of oxygen use, treating institution, and BPD. The status of the outcomes from twin pairs was treated as a correlated event. The treating institution was evaluated as an overall variable and as individual institutions in comparison with a reference institution chosen at random. A MELR model was fitted to assess the relationship between the covariates listed and the outcome of interest (ROP) and to incorporate the correlation between twin pairs.
Latent variable probit modeling for twin data was then used to estimate the variance in liability for ROP.19–21 A mixed-effects probit model was fitted to estimate the genetic contribution to ROP by adjusting for all covariates used in the MELR analysis. A liability variable underlying the respective outcome was estimated. This variable was assumed to follow a normal distribution with the mean dependent on the MELR covariates. The variance was partitioned into a genetic component, a shared nongenetic component, and a random component. The sum of the first 2 components constituted the overall sharing between twins and was determined from the correlation between monozygotic twins and dizygotic twins. We estimated this heritability (ie, the ratio of the genetic variance to the total variance in liability) on the basis of a model that assumed that the correlation among twins resulted from both genetic and nongenetic components.
Anonymous clinical data, formatted in spreadsheets in Excel (Microsoft, Redmond, WA), were forwarded from each institution to the statistical core at Yale University. Statistical analyses were performed with SAS 9.1 (PROC GLIMMIX and PROC NLMIXED; SAS Institute, Cary, NC), SPSS 13.0 for Windows and Macintosh (SPSS, Chicago, IL), and GraphPad Prism 3.0 (GraphPad Software, San Diego, CA). P values of <.05 were considered statistically significant.
ROP was diagnosed in 86 (21.5%) of 400 infants from our cohort. The incidence of ROP was inversely proportional to BW, with the majority of disease occurring in the population with BW of <1000 g (Table 1); of those infants, 67 (66%) of 101 were diagnosed as having ROP, compared with 17 (11%) of 156 infants with BW of 1000 to 1500 g and 2 (2%) of 125 infants with BW of 1501 to 2000 g (Table 1). No ROP was diagnosed in the subpopulation with BW of >2000 g.
Zygosity data for 63 monozygotic and 137 dizygotic twin pairs from 3 institutions were used for analysis. These 200 twin pairs had mean GA and BW of 29 weeks and 1332 g, respectively. Despite a discrepancy in the overall number of twin pairs in each group, no statistically significant differences were observed between monozygotic and dizygotic twins, with respect to GA, BW, gender, 5-minute Apgar scores, incidences of RDS, ROP, and BPD, duration of mechanical ventilation, duration of supplemental oxygen use, and length of hospital stay (Table 2).
MELR analysis was performed using ROP as the dependent variable in an attempt to identify significant factors in our cohort that might contribute to the outcome of interest (Table 3). The analysis determined GA (odds ratio [OR]: 0.65; 95% confidence interval [CI]: 0.45–0.94; P = .024) and duration of oxygen use (OR: 1.03; 95% CI: 1.01–1.05; P = .003) to be significant predictors for ROP (Table 3). The data presented demonstrate the results of analyses with and without BPD as a covariate. Previous investigators demonstrated an association between ROP and BPD in preterm infants, suggesting the possibility of shared nongenetic and/or genetic components.22–26 Because we demonstrated a significant genetic susceptibility to BPD previously,21 we asked whether BPD and ROP might share genetic factors. To analyze this, we performed MELR analyses using BPD as the dependent variable with ROP as a covariate (data not shown) and then ROP as the dependent variable with BPD as a covariate (Table 3). The data showed that ROP and BPD were not independent predictors of each other (BPD as dependent variable: P = .090; ROP as dependent variable: P = .764).
After significant nongenetic cofactors for ROP were identified in the MELR analysis, a latent variable probit model was used to estimate the genetic susceptibility to ROP. Our model assumed that genetic and shared and unshared nongenetic factors contributed to the correlation among twins. Using this model, we determined that 70.1% (95% CI: 9%–100%; P = .026) of the variance in liability to ROP was the result of genetic factors alone.
ROP remains a significant and prevalent cause of morbidity among preterm infants worldwide.27 In the more highly developed countries of the world, with screening programs and interventions aimed at prevention and treatment, ROP still accounts for 3% to 11% of blindness in children.27 Although the onset and progression of ROP may be influenced strongly by the GA of the newborn, factors that are unrelated to prematurity are likely involved. Previous investigators attempted to show an association between genetic factors and susceptibility to ROP.28–34 Potential candidate genes that have been evaluated involve known pathophysiologic mediators involved in this disease processes. These have included primarily mediators of angiogenesis in the developing retina, with particular focus on vascular endothelial growth factor,28,29 the Norrie disease gene (NDP),30–33 and angiotensin-converting enzyme.34 These studies should be interpreted with some degree of caution. Variations of candidate genes were present in small subsets of subjects and/or were limited geographically to specific ethnic populations, making the generalizability of the results uncertain. Although a significant amount of information regarding these genetic polymorphisms has been published,28–34 no previous definitive study has confirmed that a genetic susceptibility to ROP exists. Also, information regarding the extent of that genetic contribution has not been identified previously.
Our cohort consisted of 200 monozygotic and dizygotic twin pairs sampled from 2 centers in the United States and 1 in Sweden. By comparing concordance within monozygotic twin pairs, which share 100% of their genetic information, and concordance within dizygotic twin pairs, which share at least 50%, and controlling for known and unknown nongenetic covariates, we were able to estimate the genetic contribution to ROP.
The statistical model to estimate the genetic susceptibility to ROP separated variance into 3 major components, namely, additive genetic effects, unshared or residual nongenetic effects, and unidentified common nongenetic effects.19,35 To determine the effect of additive genetic effects, we identified and estimated the contributions from unshared or residual nongenetic effects and unidentified common nongenetic effects. Residual nongenetic effects were represented only by variables included in the MELR; these included, among other factors, the effects of prematurity,22,23,25 use of supplemental oxygen,22,25 RDS,22,23 and BPD,22–26 all previously determined risk factors for ROP. A second component included in our model, unidentified common nongenetic effects, factored in the potential effects of unidentified factors not incorporated into the MELR analysis (ie, variables not available from our data set); these included the potential influences of race,16,36 intraventricular hemorrhage,22,23,25 periventricular leukomalacia,22 necrotizing enterocolitis,37 and septicemia,22,23,26 in addition to unknown unidentified factors. By modeling the effects of these nongenetic components, we were able to determine that 70.1% of the variance in liability to ROP was attributable to genetic factors alone.
Although the statistical analyses of data for our cohort determined a significant genetic susceptibility to ROP, we recognize that there are some limitations to our data. Although criteria for the diagnosis and staging of ROP are standardized, there is still the potential for error and disagreement, even among the most experienced examiners.11 In our cohort, however, the same individual, or group of individuals, from each site performed all retinal examinations over the entire study period. In addition, our cohort was restricted to twin pairs with available zygosity information and was therefore limited in number. Specific subgroup analyses to address potential differences in genetic susceptibility according to BW and/or ROP stage could not be performed. Despite this limitation, no differences in the percentages of infants with severe ROP and/or BW of <1000 g were observed between monozygotic and dizygotic twin pairs. Lastly, the covariates collected in our data set did not include all of the potential contributing factors for ROP. We did attempt to control for these unknown variables in our statistical model, however.
Using our statistical model, we have identified a significant genetic susceptibility to ROP. The extent to which genetic factors contribute to this major cause of infant morbidity accentuates the need for a shift in the paradigm used to identify and to treat disease processes. Similar models using data from twin pairs revealed substantial genetic heritability for several disorders, including BPD (53% heritability),21 schizophrenia (82%),38 schizoaffective disorders (85%),38 type I bipolar disorder (93%),39 Alzheimer disease (79%),40 and corneal thickness, which is an important factor in glaucoma (95%).41 Significant genetic effects have also been shown to influence reading and mathematics performance (82% heritability).42 The identification of a genetic influence has led to the isolation of specific genes associated with some of these disorders, including Alzheimer disease, schizophrenia, and dyslexia.43–47 It is possible that similar discoveries may be achievable for ROP, and dual therapy, aimed at limiting potential environmental risk factors and identifying and targeting specific genetic factors, may become the model for future interventions.
Dr Bizzaro was supported in part by National Institute of Child Health and Human Development training grant T32 HD07094. Dr Jonsson was supported by Sallskapet Barnavard and Stiftelsen Frimurare Barnhuset in Stockholm. Dr Ment was supported by National Institute of Neurological Disorders and Stroke grants NS27116, NS35476, and NS42027. Dr Gruen was supported by National Institute of Neurological Disorders and Stroke grant R01 NS43530. Dr Zhang was supported by National Institute on Drug Abuse grants K02 DA017713 and R01 DA016750. Dr Bhandari was supported by National Heart, Lung, and Blood Institute grant K08 HL074195.
- Accepted July 9, 2006.
- Address correspondence to Vineet Bhandari, MD, DM, Department of Pediatrics, Yale University School of Medicine, 333 Cedar St, PO Box 208064, New Haven, CT 06520-8064. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
Dr Feng's current address is Section on Statistical Genetics, Department of Biostatistics, Ryals Public Health Building 420B, University of Alabama at Birmingham, Birmingham, AL 35294.
- ↵Reynolds JD. Retinopathy of prematurity. In: Nelson LB, Olitsky SE, eds. Harley's Pediatric Ophthalmology. 4th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2005:77– 78
- Stone J, Itin A, Alon T, et al. Development of retinal vasculature is mediated by hypoxia-induced vascular endothelial growth factor (VEGF) expression by neuroglia. J Neurosci.1995;15 :4738– 4747
- Neufeld G, Cohen T, Gengrinovitch S, Poltorak Z. Vascular endothelial growth factor (VEGF) and its receptors. FASEB J.1999;13 :9– 22
- ↵STOP-ROP Multicenter Study Group. Supplemental therapeutic oxygen for prethreshold retinopathy of prematurity (STOP-ROP), a randomized, controlled trial, part I: primary outcomes. Pediatrics.2000;105 :295– 310
- ↵Good WV, Hardy RJ, Dobson V, et al. The incidence and course of retinopathy of prematurity: findings from the early treatment for retinopathy of prematurity study. Pediatrics.2005;116 :15– 23
- Kinsey VE, Arnold HJ, Kalina RE, et al. Pao2 levels and retrolental fibroplasia: a report of the cooperative study. Pediatrics.1977;60 :655– 668
- ↵International Committee for Classification of ROP. An international classification of retinopathy of prematurity. Pediatrics.1984;74 :127– 133
- ↵Shennan AT, Dunn MS, Ohlsson A, Lennox K, Hoskins EM. Abnormal pulmonary outcomes in premature infants: prediction from oxygen requirement in the neonatal period. Pediatrics.1988;82 :527– 532
- ↵Neale MC, Cardon LR. Methodology for Genetic Studies of Twins and Families. Dordrecht, Netherlands: Kluwer Academic Publishers; 1992
- ↵Bhandari V, Bizzarro MJ, Shetty A, et al. Familial and genetic susceptibility to major neonatal morbidities in preterm twins. Pediatrics.2006;117 :1901– 1906
- ↵Hussain N, Clive J, Bhandari V. Current incidence of retinopathy of prematurity, 1989–1997. Pediatrics.1999;104 (3). Available at: www.pediatrics.org/cgi/content/full/104/3/e26
- ↵Purohit DM, Ellison RC, Zierler S, Miettinen OS, Nadas AS. Risk factors for retrolental fibroplasia: experience with 3,025 premature infants: National Collaborative Study on Patent Ductus Arteriosus in Premature Infants. Pediatrics.1985;76 :339– 344
- ↵Gilbert C, Fielder A, Gordillo L, et al. Characteristics of infants with severe retinopathy of prematurity in countries with low, moderate, and high levels of development: implications for screening programs. Pediatrics.2005;115 (5). Available at: www.pediatrics.org/cgi/content/full/115/5/e518
- ↵Cooke RW, Drury JA, Mountford R, Clark D. Genetic polymorphisms and retinopathy of prematurity. Invest Ophthalmol Vis Sci.2004;45 :1712– 1715
- ↵Toh TY, Liew SHM, MacKinnon JR, et al. Central corneal thickness is highly heritable: the twin eye studies. Invest Ophthalmol Vis Sci.2005;46 :3718– 3722
- ↵Holmans P, Hamshere M, Hollingsworth P, et al. Genome screen for loci influencing age at onset and rate of decline in late onset Alzheimer's disease. Am J Med Genet B Neuropsychiatr Genet.2005;135 :24– 32
- Tosato S, Dazzan P, Collier D. Association between the neuregulin 1 gene and schizophrenia: a systematic review. Schizophr Bull.2005;31 :613– 617
- Williams NM, O'Donovan MC, Owen MJ. Is the dysbindin gene (DTNBP1) a susceptibility gene for schizophrenia? Schizophr Bull.2005;31 :800– 805
- ↵Meng H, Smith SD, Hager K, et al. DCDC2 is associated with reading disability and modulates neuronal development in the brain. Proc Natl Acad Sci USA.2005;102 :17053– 17058
- Copyright © 2006 by the American Academy of Pediatrics