The Genetics and Clinical Variables of Retinopathy of Prematurity: A Matched Case Control Study

Purpose: Retinopathy of prematurity (ROP) is a major cause of childhood blindness worldwide; it is a proliferative retinal vascular disease in preterm infants. Some infants progress to severe disease despite absence of these clinical risk factors such as low birth weight, gestational age and oxygen. Genetic factors seem to have etiologic roles in retinopathy of prematurity (ROP). This matched case-control study examined genetic factors for ROP. Methods: Conrmed 71 severe ROP cases and gestational age, birth weight and days of oxygen therapy matched controls (1:1) were enrolled from September 2015 to August 2020. Exome sequencing was performed and accomplished by next-generation sequencing. Gene mutations on heritable retinal vascular diseases related to Norrin signaling pathway were analyzed. Results: Seven heterozygous variants in exon 2 and the 3′ UTR were identied in the FZD4 gene in 28 patients with ROP. Variants c.*3770A > T and c.*2971T > C were found to be signicantly associated with ROP (P = 0.033 and P = 0.017 respectively). Screening of LPR5 revealed ve heterozygous variants. The rate of variant c.575 + 6T > C was much higher in infants with ROP (P = 0.009). One variant in exon 3 (c.67-9delT) was found in the TSPAN12 gene in one infant with ROP, whereas another variant of TSPAN12, c.*1139A > T, was identied in both groups. Two variants of ZNF408, c.581_592del AGTGGTGACAGA and c.1007C > T, were found in infants with ROP, with the rate of change of the former being much higher in infants with ROP (P < 0.001 and P = 0.033 respectively). Conclusions: This study revealed genetic factors in Norrin signaling may contributed to the development and severity of ROP, which warrants further detailed investigation worldwide. in preterm characterized by abnormal retinal vessel growth. FZD4, LRP5, and ZNF408 are involved in activation of the Norrin/β-catenin signaling pathway and are considered the causative genes for ROP progression, to for FEVR.[4, rst mutation analysis of the FEVR pathogenic gene in Chinese ROP patients. Various types of mutations in the NDP gene have been reported.[21] gene heterozygous transition from T to C at nucleotide and in each identied pathogenic and development ROP. ZNF408-associated ROP is not well known. Our report suggests that the p.S336L mutation disrupts the expression of important genes in angiogenesis and sheds further light on the molecular mechanisms underlying ZNF408-associated ROP. There were some limitations to the present study. We only did gene polymorphisms analysis for type 1 ROP or threshold ROP patients, which may lead to visual impairment. Most of Type 2 or mild ROP recovered spontaneously, and major parents of these infant declined to participant this study. It could be necessary to conduct further studies with larger sample sizes to screen Norrin signaling gene mutations in Chinese ROP patients and to clarify whether these mutations are risk factors for the occurrence and development of ROP. Further, several infant characteristics varied between severe ROP and non-ROP groups in this study. C (rs10898563) of the FZD4 gene, c.575 6T > C (rs4988322) of the LRP5 gene, c.*715T > C (rs3747350) of the NDP gene, c.67-9delT(rs774111149) of the TSPAN12 gene, and novel founder mutations c.1007C > T (p.S336L) and c.581_592del AGTGGTGACAGA (p. Val194_Val197del AVVTEinsA, rs148055528) of the ZNF408 gene. These mutations were classied into SNPs, for which details in the dbSNP. FZD4, variant and non-frameshift variant of ZNF408 signicantly


Introduction
Retinopathy of prematurity (ROP) is a retinal vascular proliferative disease in very preterm infants and is the main cause of visual impairment and blindness in children. [1] The pathogenesis of severe ROP is still unclear, partly due to the lack of complete retinal angiogenesis in preterm infants, and is associated with a variety of factors, including very low birth weight, short gestational age, and longer oxygen treatment. [2] In addition, the incidence of ROP is different in different races. [3] This suggested that various changes in the nuclear coding genes might be involved in the occurrence and development of ROP. [2] Familial exudative vitreoretinopathy (FEVR) is an inherited vitreoretinopathy, with clinical manifestations similar to ROP, [4] involving several genes with different variations, including insertions, point mutations, or deletions, which include Frizzled-4 (FZD4), lipoprotein receptor-related protein 5 (LRP5), Norrie disease protein (NDP), and tetraspanin-12 (TSPAN12). [2,5] FZD4, LRP5, and TSPAN12 encode the ligand-receptor complex of the Norrin/β-catenin signaling pathway, which is involved in the activation of retinal vascularization.
[6] The gene alterations linked to FEVR are considered associated with the development of ROP because of the phenotypic similarity between FEVR and ROP. [2] Norrin-β-catenin signaling plays a very important role in regulating retinal angiogenesis. FZD4, LRP5, and TSPAN12 proteins are expressed in endothelial cells of the eye and form the FZD4/LRP5 complex. The Norrin ligand binds to the FZD4/LRP5 complex and TSPAN12 and activates the downstream β-catenin signaling pathway, consequently initiating gene expression related to retinal vascularization. [7,8] The involvement of genes in Norrin signaling is simpli ed and illustrated in Figure (Figure 1). [8,9] The similar clinical manifestations of ROP and FEVR further increase the possibility of Norrin-β-catenin signal transduction genes participating in the pathogenesis of ROP. However, few studies have detected mutations in Norrin-β-catenin signaling genes in ROP worldwide based on different populations, and thus, further research is needed. [4] Therefore, this study aimed to screen Norrin-β-catenin signaling genes in Chinese ROP patients to understand their contribution to disease pathogenesis. We further evaluated the correlation between the observed genotype and phenotype of the variants to understand their relative role in predicting disease severity and visual outcome.

Study design and setting
A matched case control study was performed from September 2015 to August 2020 on a Chinese population; subjects were born with birth weights (BW) <1500 g and gestational age (GA) < 32 weeks, at the Children's Hospital a liated to Zhengzhou University. A case was de ned as a patient of BW < 1500 g and GA <32 weeks with a conformed severe ROP, diagnosed by an ophthalmologist using "The International Classi cation of Retinopathy of Prematurity revisited" 2005. [10] A control was de ned as a preterm infant without any stage of ROP. Controls were individually matched with each case (1:1) for GA(±3 days), BW (±50grams) and days of oxygen therapy (±3 days). Infants with congenital malformations and genetic metabolic diseases were excluded from the study. The parents or legal guardians of all infants provided informed consent before participation in the study. The study was approved by the Children's Hospital a liated to Zhengzhou University Ethics Committee, and conducted in compliance with the Declaration of Helsinki.
All infants involved in this study were screened for ROP by an ophthalmologist according to the ROP screening schedule. [11] The examining ophthalmologist performed the follow-up examinations based on retinal ndings. The stages of ROP were classi ed according to the "International classi cation of retinopathy of prematurity revisited". [10,12] ROP was subdivided into stages 1-5 according to the International Classi cation of ROP, [10] de ned as follows: 1) none: immature or mature vascularization; 2) mild ROP: stage 1 or stage 2 ROP in zone II or III without plus disease; 3) type 2 ROP: stage 1 or 2 ROP without plus disease in zone I or stage 3 ROP without plus disease in zone II; 4) type 1 ROP: any stage ROP with plus disease, stage 3 ROP without plus disease in zone I, or stage 2 or 3 ROP with plus disease in zone II; 5) Threshold disease ROP was de ned as ve contiguous or eight cumulative clock hours of stage 3 ROP in zone 1 or 2 with "plus" disease. [10,13] 6) aggressive posterior ROP (AP-ROP) was de ned as an "uncommon, rapidly progressing, severe form of ROP" that "usually progresses to stage 5" ROP if untreated. [10] 7) Severe ROP was de ned as prethreshold disease type 1, threshold disease or AP-ROP meeting for the criteria of treatment-requiring ROP [14,15]. The frequency at which infants were re-assessed/re-screened depended on the results of the previous ROP assessment/screening. The termination of ROP screening varied depending on retinal vascular development and was carried out before 45 weeks post-menstrual age if no type 1 or worse disease was present. [12] DNA extraction Peripheral blood samples (0.3-1mL) were obtained the day after severe ROP diagnosis by the examining ophthalmologist. For non-ROP infants with same gender, gestational ages and birth weights similar to those of the severe ROP cases, the peripheral blood samples were obtained on the day of ROP screening termination. Using the chemagic ™ DNA blood kit (PerkinElmer, Massachusetts, Massachusetts, USA) on the chemagic ™ 360 system (PerkinElmer), the genomic DNA was isolated from the peripheral blood of enrolled subjects following the manufacturer's guidelines. NanoVue ™ Plus (GE Life Sciences, New Jersey, USA) was used to measure the quantity and quality of DNA. Electrophoresis on 1% agarose gels was performed to detect the integrity of the DNA.
Next-generation sequencing and data processing Next-generation sequencing was performed for patients to determine the molecular cause of ROP using the NextSeq500 PE150 (Illumina, San Diego, CA, USA). The ClearSeq Inherited Disease Panel was adapted for the clinical testing of every enrolled pro-band. For variant calling, the GATK Best Practices were employed to detect single nucleotide variations/small indels. Samtools, GATK, and ANNOVAR were used separately for the detection of copy number variations, and the results were merged. Genetic analysis was complemented by qPCR and duplex-PCR analysis. The observed variants were further veri ed by sequencing and evaluated for their association with disease susceptibility. The harmful effects of the variants observed among these candidate genes were predicted using the bioinformatics software ClustalW Omega, SIFT,[16] PolyPhen, [17] and Mutation Taster. [18] Statistical analysis Statistical Package for Social Science (SPSS) version 21.0 (SPSS Inc., Chicago, Illinois, USA) was used to analyze statistical data. The relationship between detected variants and clinical factors in ROP and non-ROP patients was tested using the Chi-squared or Fisher's exact tests for categorical variables and the ttest for continuous variables. Multiple logistic regression analysis was used and adjusted for potential confounders. Crude relative risks (RR) with 95% con dence intervals (95%CIs) were estimated. The signi cance of values was set at P < 0.05 with a 95% con dence interval. The observed variant allele frequencies were compared using the Exome Aggregation Consortium database.

Baseline characteristics
In the study period, 1620 preterm infants (gestational age at birth ≤ 32 weeks and birth weight ≤ 1500 g) were born and assessed for eligibility. Among them, 81 infants with severe ROP, Eleven infants were excluded because of congenital anomalies and informed consent was not obtained. 71 severe ROP cases were recruited, and as well as 86 infants with similar gestational ages, birth weights and days of oxygen therapy without ROP as the control group. Fifteen infants were excluded because of congenital anomalies and refused to participate in the control group. (Fig. 2). Most of the maternal characteristics were similar between the two groups, although some infant characteristics varied. The mechanical ventilation days (P = 0.035) were longer in the severe ROP group compared to those in the non-ROP group. The incidence of necrotizing enterocolitis, sepsis, or bronchopulmonary dysplasia (BPD) was higher compared to the non-ROP group ((P = 0.028 and P = 0.023, respectively) ( Table 1).  (rs182049165) and c.*2971T > C (rs10898563), were much higher in the severe ROP group than in the non-ROP group (P = 0.033 and P = 0.017, respectively). Two variants of FZD4, c.*2660C > T (rs11234890) and c.*203A > G (rs3740661), were detected in both groups (non-ROP group: 3% and 1%, respectively; severe ROP group: 6% and 3%, respectively; Table 2).
Sequencing data for the TSPAN12 gene There was a heterozygous TSPAN12 gene variant in one ROP patient. The TSPAN12 c.67-9delT (rs774111149) substitution was a splicing change in exon 3. Another variant of TSPAN12, c.*1139A > T (rs192303288), was identi ed in both groups (10% for both severe ROP and non-ROP patients; Table 2).
There were heterozygous and homozygous NDP gene variants in two different patients from each group ( Table 2). The NDP c.*715T > C (het. or hom.) substitution was an unknown change in the 3′UTR (

Discussion
Severe ROP can lead to blindness in preterm infants, which is characterized by abnormal retinal vessel growth. FZD4, LRP5, and ZNF408 are involved in activation of the Norrin/β-catenin signaling pathway and are considered the causative genes for ROP progression, similar to that for FEVR. [4,8,19,20] This study rst reported a mutation analysis of the FEVR pathogenic gene in Chinese ROP patients. Various types of mutations in the NDP gene have been reported. [21] However, in this study, entire NDP gene sequencing (translation region) revealed a heterozygous variant comprising a transition from T to C at nucleotide position 715, and only two patients in each group were identi ed as having this alteration. This variant might not be pathogenic and might not be involved in the development of ROP.
As is known, proline is an atypical amino acid that plays a unique role in the rigidity of protein conformations. Several studies found that proline to serine missense substitutions cause compromised protein rigidity, indicated that it might be a causative factor for FEVR or ROP. [22,23] FZD4 coupling variants might increase the incidence of retinal damage, which can lead to ROP. It was also indicated that biomarkers for an increased risk of ROP development might comprise these variants. [24] It has been previously reported that LRP5 in FEVR patients is associated with various types of molecular abnormalities. The genetic variants of LRP5 with 10 different single-base substitutions were analyzed in 53 Japanese patients with advanced ROP in 2013, and variants in the coding sequence of this region were mostly discovered. [4] We also identi ed a variant in exon 2, exon 9, exon 10, and exon 19 of LRP5 (Q89R, D666N, E644E, and D1363D, which were determined to be nonsynonymous, synonymous, or splicing change mutations based on the unchanged protein sequence. The NLHBI GO Exome Sequencing Project (ESP, Seattle, WA, USA) was used to classify synonymous changes as common polymorphisms, with minor allele frequencies known to be less than 10%. [4] In the Norrin/β-catenin signaling pathway, LRP5 is a receptor that functions with FZD4. [25] The presence of the c.3357G > A (p.V1119V) variant was reported previously in severe ROP in a Chinese and Japanese population. [21,26] These variants might be associated with the development of ROP.
Variants of TSPAN12 in the 3 UTR and exon 3 were identi ed in this study, including a splicing change, c.67-9delT (rs774111149), and another c.*1139A>T (rs192303288) variant in the 3 UTR. TSPAN12 is one component of a four-tetraspanin protein and is involved in various activities associated with cell proliferation and signaling pathways; [27] its speci c function is to activate the FZD4/LRP5 receptor complex by interacting with Norrin or LRP5.
Both changes were discovered in one tenth of this study population (13 of 134 subjects). The c.*1139A > T (rs192303288) change was found in all subjects with the 3 UTR variant. As proposed, the 3 UTR is involved in regulatory pathways of transcription. In all subjects with the 3 UTR variant, the c.*1139A > T (rs192303288) change was found. [29] As reported in previous studies, the role of this region is to control gene expression involving SNPs at the 3 UTR. [30] Furthermore, some studies showed that any 3 UTR variant might be associated with susceptibility to some diseases. [31] Novel mutations in ZNF408 were found in this study. ZNF408 encodes a transcription factor of the zinc nger family. Three variants of ZNF408 in downstream bases, exon 4, and exon 5 were discovered as follows: the unknown change rs146198477, non-frameshift change c.581_592del AGTGGTG (p. Val194_Val197del AVVTEinsA, rs148055528), and nonsynonymous change c.1007C > T (p. S336L, rs150368802) in the exon. According to the type and number of zinc ngers, zinc nger genes can be separated into many subclasses. [32] Each nger constitutes an independent domain, which is stabilized by zinc ions attached to cysteine and histidine, along with a hydrophobic core in the inner structure. Some reports suggested that variants of ZNF408 lead to abnormal retinal vascularization in humans and might be related to disease processes in FEVR. [33] However, the molecular mechanism of ZNF408associated ROP is not well known. Our report suggests that the p.S336L mutation disrupts the expression of important genes in angiogenesis and sheds further light on the molecular mechanisms underlying ZNF408-associated ROP.
There were some limitations to the present study. We only did gene polymorphisms analysis for type 1 ROP or threshold ROP patients, which may lead to visual impairment. Most of Type 2 or mild ROP recovered spontaneously, and major parents of these infant declined to participant this study. It could be necessary to conduct further studies with larger sample sizes to screen Norrin signaling gene mutations in Chinese ROP patients and to clarify whether these mutations are risk factors for the occurrence and development of ROP. Further, several infant characteristics varied between severe ROP and non-ROP groups in this study.
In conclusion, ve Norrin signaling genes involved in ROP were detected in the present study based on a Chinese population. They were identi ed as c.*3770A > T (rs182049165) and c.*2971T > C (rs10898563) of the FZD4 gene, c.575 + 6T > C (rs4988322) of the LRP5 gene, c.*715T > C (rs3747350) of the NDP gene, c.67-9delT(rs774111149) of the TSPAN12 gene, and novel founder mutations c.1007C > T (p.S336L) and c.581_592del AGTGGTGACAGA (p. Val194_Val197del AVVTEinsA, rs148055528) of the ZNF408 gene. These mutations were classi ed into SNPs, for which details are available in the dbSNP. The present study suggested that patients with the non-synonymous variant in FZD4, the splicing variant of LRP5, and non-frameshift variant of ZNF408 might have a signi cantly increased risk for the development ROP. Subjects. Consort Diagram Flow chart of included infants. ROP= retinopathy of prematurity