Deep-Learning-Based Classification of Lung Adenocarcinoma and Squamous Cell Carcinoma Using DNA Methylation Profiles:...

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• Accurate classification of non-small-cell lung cancer (NSCLC) into lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC) is crucial for treatment and prognosis.
• Current subtyping methods require improvement for optimal patient management.

Purpose of the Study:

• To develop and validate a deep learning model for classifying NSCLC subtypes using genome-wide DNA methylation profiles.
• To identify key DNA methylation biomarkers for NSCLC classification.

Main Methods:

• Utilized variance-based feature selection to identify 5000 discriminative CpG probes from Illumina HumanMethylation450 BeadChip data.
• Developed a five-layer deep neural network with batch normalization and dropout regularization for classification.
• Trained and validated the model on The Cancer Genome Atlas (TCGA) data, with external validation on Gene Expression Omnibus (GEO) datasets (GSE39279, GSE56044).

Main Results:

• The model achieved 96.92% accuracy and an AUC-ROC of 0.9981 on the TCGA test set.
• Cross-dataset validation demonstrated robust generalization, with a GEO-trained model achieving 88.92% accuracy and 0.9724 AUC-ROC on TCGA data.
• SHAP analysis identified key CpG biomarkers influencing classification decisions.

Conclusions:

• Deep learning models leveraging DNA methylation data offer a reliable approach for NSCLC subtype classification.
• These findings suggest potential clinical applicability for improving NSCLC diagnosis and treatment strategies.