Subclassification of Small Cell Lung Cancer Based on Gene Expression Signatures and Machine Learning.

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• Small-cell lung cancer (SCLC) is classified into four molecular subtypes (NAPY) based on transcription factor (TF) activity: NEUROD1, ASCL1, POU2F3, and YAP1.
• Current diagnostic methods for SCLC subtypes lack consensus, hindering effective treatment strategies.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for accurate SCLC-NAPY subtyping using TF-specific gene expression signatures.
• To identify subtype-specific genomic alterations and molecular patterns for improved SCLC classification and prognosis.

Main Methods:

• Analysis of transcriptomic and genomic data from 460 SCLC patients.
• Extraction of TF-specific gene expression signatures to train ML models for NAPY subtype prediction.
• Assessment of genomic alterations and cancer pathway signatures (RosettaSX) for subtype-specific signals.

Main Results:

• The ML model achieved approximately 90% accuracy in predicting SCLC-NAPY subtypes in clinical samples and cell lines.
• Survival analyses revealed significant prognostic differences among SCLC subtypes.
• A robust diagnostic algorithm for NAPY classification was proposed, integrating TF expression with downstream signature activity.

Conclusions:

• The developed ML model provides a highly accurate and functionally robust method for SCLC-NAPY subtyping.
• This classification facilitates the identification of novel molecular and clinical associations across SCLC subtypes, paving the way for personalized treatment approaches.