Related Experiment Video
Updated: Sep 12, 2026
Monitoring Dynamic Growth of Retinal Vessels in Oxygen-Induced Retinopathy Mouse Model
Published on: April 2, 2021
Construction of a prediction model for retinopathy of prematurity based on placenta-derived genes and clinical
Dan Huang1, Rong Gan2, Yunpeng Zhang3
1Department of Ophthalmology Center, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, China.
Background:
Retinopathy of prematurity (ROP) is a leading cause of childhood blindness, with rising incidence as neonatal care improves. Established risk factors include low birth weight, prematurity, and supplemental oxygen, yet current clinical prediction tools have limited early accuracy. The placenta, serving as the maternal-fetal interface, may harbor transcriptomic signatures reflecting intrauterine inflammation and hypoxia that influence ROP development; however, the specific placental genes involved and their predictive value remain largely unknown. This study aimed to identify placenta-derived differentially expressed genes (DEGs) associated with ROP and construct a predictive risk model integrating gene expression and clinical variables for early ROP diagnosis and prevention.
Methods:
DEGs were screened from the GSE218039 placental RNA sequencing (RNA-seq) dataset (ROP vs. control). Functional enrichment, protein-protein interaction (PPI) network, and immune infiltration analyses were performed. Overlapping DEGs with the oxygen-induced retinopathy (OIR) mouse model (GSE200195, GSE158799) were identified. Fourteen upregulated genes were validated by quantitative real-time polymerase chain reaction (qRT-PCR) in OIR retinal tissues. Elastic-net penalized regression and multivariate logistic regression were applied to the GSE32472 dataset (postnatal days 5 and 28) to select predictive genes and clinical variables (birth weight, bronchopulmonary dysplasia) for constructing a nomogram.
Results:
A total of 1,071 DEGs (758 upregulated, 313 downregulated) were identified, enriched in cytokine-cytokine receptor interaction and chemokine signaling pathways. Seven genes (B2M, C3, ITGAX, CAPG, MCAM, MMP25, LGALS3) were significantly upregulated in OIR retinas (P<0.05). Neutrophil infiltration was increased in ROP and positively correlated with these genes. Four overlapping genes (MMP25, CAPG, MCAM, B2M) were selected from both time points and combined with birth weight and bronchopulmonary dysplasia to build a nomogram. The model achieved an area under the curve (AUC) of 0.948 [95% confidence interval (CI): 0.899-0.998] on postnatal day 5, with good calibration.
Conclusions:
The nomogram incorporating four placenta-derived genes and clinical variables shows excellent predictive performance for early ROP detection, offering a potential tool for individualized risk assessment, early intervention, and prevention of ROP in premature infants.
