Identification and validation of cuproptosis related genes and signature markers in bronchopulmonary dysplasia disease using bioinformatics analysis and machine learning

Mingxuan Jia, Jieyi Li, Jingying Zhang, Ningjing Wei, Yating Yin, Hui Chen, Shixing Yan, Yong WangBMC Medical Informatics Decis. Mak 2023 · DOI

Introduction

Bronchopulmonary Dysplasia (BPD) remains a significant challenge in neonatal care, affecting approximately 35% of preterm infants. This chronic lung condition often results in long-term respiratory complications and neurodevelopmental delays. Despite its prevalence, the molecular drivers of BPD are not fully understood, and clinical management is primarily supportive.

Recent research has highlighted cuproptosis—a form of regulated cell death triggered by intracellular copper accumulation—as a potential factor in various diseases. While well-documented in oncology, its role in neonatal lung pathology is a new frontier. This study aims to bridge this knowledge gap by identifying cuproptosis-related genes (CRGs) that could serve as reliable diagnostic biomarkers for BPD.

Methods and Datasets

The researchers implemented a multi-stage bioinformatics and machine learning pipeline to analyze high-throughput gene expression data:

  1. Data Acquisition: The study utilized infant cord blood samples from the Gene Expression Omnibus (GEO). The discovery phase used dataset GSE108754, while validation was performed using GSE190215 and GSE188944.
  2. Differential Expression and Network Analysis: Using the limma R package, the team identified differentially expressed CRGs. Weighted Gene Co-expression Network Analysis (WGCNA) was then applied to find gene modules strongly associated with BPD.
  3. Machine Learning Optimization: Four algorithms—Generalized Linear Models (GLM), Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGB)—were compared to identify the most accurate predictive model.
  4. Immune Profiling: CIBERSORT was used to evaluate immune cell infiltration, exploring the link between cuproptosis and the neonatal immune environment.

Key Insights: The Cuproptosis-BPD Connection

The analysis pinpointed two critical copper-death-associated genes: NFE2L2 and GLS. In BPD patients, NFE2L2 was significantly downregulated, whereas GLS was upregulated. This inverse relationship suggests a complex regulatory role for cuproptosis in the developing lung.

Differential expression analysis of cuproptosis-related genes (CRGs) shows NFE2L2 downregulation and GLS upregulation in BPD samples.

Furthermore, genomic mapping revealed that both NFE2L2 and GLS are located on human chromosome 2. The positive correlation between these genes and their shared chromosomal location suggests a genetic "hotspot" that may be central to BPD susceptibility.

Circos plot illustrating the chromosomal distribution of cuproptosis-related genes, highlighting the proximity of key markers on chromosome 2.

The study also found that these CRGs influence the immune microenvironment, particularly affecting Monocytes, which were found to be positively regulated by the identified cuproptosis genes.

Key Results and Model Performance

The study successfully identified a five-gene signature for BPD prediction: NFATC3, ERMN, PLA2G4A, MTMR9LP, and LOC440700.

Surprisingly, the Generalized Linear Model (GLM) outperformed more complex algorithms like XGBoost. The GLM-based model achieved remarkable predictive accuracy in external validation:

  • 94.2% AUC on the GSE190215 dataset.
  • 98.7% AUC on the GSE188944 dataset.

Additionally, clustering analysis revealed five metabolic pathways, including pantothenate and CoA biosynthesis, that differ significantly between BPD subgroups, offering potential targets for future therapeutic intervention.

Limitations and Future Directions

While these findings are robust, the authors emphasize that this is a retrospective, data-driven study.

  • Experimental Validation: The biological role of the identified genes and the specific mechanisms of cuproptosis in lung tissue must be confirmed through in vitro and in vivo experiments to prove causality.
  • Clinical Implementation: Prospective trials are required to validate the diagnostic utility of the five-gene signature and the associated risk-prediction nomogram in a clinical setting.
  • Functional Characterization: The role of non-coding elements like LOC440700 remains largely unknown and warrants further investigation.

Summary updated

BibTeX
@article{jia2023identification,
  author = {Mingxuan Jia and Jieyi Li and Jingying Zhang and Ningjing Wei and Yating Yin and Hui Chen and Shixing Yan and Yong Wang},
  title = {{Identification and validation of cuproptosis related genes and signature markers in bronchopulmonary dysplasia disease using bioinformatics analysis and machine learning}},
  journal = {BMC Medical Informatics Decis. Mak.},
  year = {2023},
  doi = {10.1186/s12911-023-02163-x},
}
Figure