Bronchopulmonary Dysplasia Predicted by Developing a Machine Learning Model of Genetic and Clinical Information

Abstract

Background: An early and accurate evaluation of the risk of bronchopulmonary dysplasia (BPD) in premature infants is pivotal in implementing preventive strategies. The risk prediction models nowadays for BPD risk that included only clinical factors but without genetic factors are either too complex without practicability or provide poor-to-moderate discrimination. We aim to identify the role of genetic factors in BPD risk prediction early and accurately.

Methods: Exome sequencing was performed in a cohort of 245 premature infants (gestational age <32 weeks), with 131 BPD infants and 114 infants without BPD as controls. A gene burden test was performed to find risk genes with loss-of-function mutations or missense mutations over-represented in BPD and severe BPD (sBPD) patients, with risk gene sets (RGS) defined as BPD-RGS and sBPD-RGS, respectively. We then developed two predictive models for the risk of BPD and sBPD by integrating patient clinical and genetic features. The performance of the models was evaluated using the area under the receiver operating characteristic curve (AUROC).

Results: Thirty and 21 genes were included in BPD-RGS and sBPD-RGS, respectively. The predictive model for BPD, which combined the BPD-RGS and basic clinical risk factors, showed better discrimination than the model that was only based on basic clinical features (AUROC, 0.915 vs. AUROC, 0.814, P = 0.013, respectively) in the independent testing dataset. The same was observed in the predictive model for sBPD (AUROC, 0.907 vs. AUROC, 0.826; P = 0.016).

Conclusion: This study suggests that genetic information contributes to susceptibility to BPD. The predictive model in this study, which combined BPD-RGS with basic clinical risk factors, can thus accurately stratify BPD risk in premature infants.

Keywords: bronchopulmonary dysplasia; exome sequencing; machine learning; prediction model; premature infants.

FIGURE 1

Schematic diagram of the study design. (A) Study flow chart. (B) Schematic of clinical characteristic collection at each timepoint. Asphyxia was defined as APGAR score of less than seven at 1 min or/and APGAR score of less than seven at 5 min. IVF, in vitro fertilization; IUGR, intrauterine growth restriction; PIH, pregnancy-induced hypertension; PPROM, preterm premature rupture of membranes; NRDS, neonatal respiratory distress syndrome; ASD/VSD, atrial septal defect/ventricular septal defect. Patent ductus arteriosus was defined by clinical signs supported by echocardiographic confirmation. Airway malformations: bronchomalacia, tracheomalacia, laryngomalacia, or subglottic stenosis. Early-onset infections: prenatal infection within 72 h of delivery or onset of neonatal pneumonia/sepsis within 7 days of birth. Late-onset infections: infections after 7 days of birth. PMA 36w: postmenstrual age 36 weeks.

FIGURE 2

Distribution of infant clinical characteristics. (A) Cluster analysis of clinical characteristics of all infants. (B) Significant clinical characteristics in three comparisons. Twenty-one clinical characteristics that significantly differed between bronchopulmonary dysplasia (BPD) and controls or between any two BPD levels (p < 0.05) were selected. Weight scale including low BW (1,500–2,499 g), very low BW (1,000–1,499 g), and extremely low BW (<1,000 g) were represented by values 1–3, respectively. Age category including extremely preterm (<28 weeks) and very preterm (28 to 32 weeks) was represented by values 1 and 2, respectively. ASD/VSD, atrial septal defect/ventricular septal defects; NRDS, neonatal respiratory distress syndrome; sBPD, severe BPD; mBPD, mild and moderate BPD.

FIGURE 3

Receiver operating characteristic (ROC) analyses of predictive models for infants with bronchopulmonary dysplasia (BPD) or severe BPD (sBPD). The comparisons of predictive models for BPD and sBPD. P-values show the areas under the ROC curves (AUROCs) between the different models. Clinical characteristics include the variables in a separate lasso model (Supplementary Figure 1). (A) ROC analyses of the prediction of BPD by the combination of BPD–risk gene set (RGS) and all clinical characteristic model, the basic clinical characteristic model, and the combined BPD–RGS and basic clinical characteristics model. (B) ROC analyses of the prediction of severe BPD by the combined sBPD–RGS and all clinical characteristic model, the basic clinical characteristic model, and the combined sBPD–RGS and basic clinical characteristics model. CCs, clinical characteristics.