Deep learning based histomorphological phenotyping and prognostic stratification for combined SCLC and LCNEC - PubMed
4 hours ago
- #Neuroendocrine Carcinoma
- #Deep Learning
- #Prognostic Stratification
- A deep learning model named GTBIS was developed for histomorphological phenotyping and prognostic stratification of combined small and large cell neuroendocrine lung carcinoma (cSCLC-LCNEC).
- In multicenter cohorts (n=670), GTBIS accurately differentiated between SCLC and LCNEC and stratified cSCLC-LCNEC patients treated with chemoradiotherapy into subgroups with different prognoses.
- Patients classified as having an SCLC-like phenotype (favorable-prognosis) showed significantly better five-year overall survival (100% vs. 39.5%) and disease-free survival (87.5% vs. 36.0%) compared to those with an LCNEC-like phenotype (poor-prognosis).
- GTBIS classification was validated as an independent strong prognostic factor through multivariable analysis, and interpretability linked the favorable-prognosis phenotype to proliferative pathways and the poor-prognosis phenotype to pathways like epithelial-mesenchymal transition and hypoxia.