The prognostic analysis and a machine-learning based disease-specific survival state model in pulmonary large-cell...

Pulmonary large-cell neuroendocrine carcinoma (PLCNEC) is a rare and highly malignant lung cancer. Due to the paucity of data from clinical studies, its clinical characteristics and treatment remain controversial. The present study explored factors influencing the prognosis and survival outcomes of patients with PLCNEC and developed a dependable prognostic model using machine learning.

The clinical data of PLCNEC patients were extracted from the Surveillance, Epidemiology, and End Results (SEER) database between 2010 and 2020. A total of 2,897 PLCNEC patients were enrolled and univariate and multivariate Cox regression analyses were performed to explore independent prognostic factors for disease-specific survival (DSS). Ten machine learning algorithms were utilized to predict the 2-year survival. The clinicopathological data collected from The First Affiliated Hospital of Sun Yat-sen University between 2010 and 2022 were used to test the trained machine.

Sex [hazard ratio (HR) 1.168, 95% confidence interval (CI): 1.063-1.284], age (HR 1.262, 95% CI: 1.144-1.391), surgery (HR 0.481, 95% CI: 0.413-0.559), chemotherapy (HR 0.450, 95% CI: 0.404-0.501), bone metastasis (HR 1.284, 95% CI: 1.124-1.466), brain metastasis (HR 1.167, 95% CI: 1.023-1.331), liver metastasis (HR 1.223, 95% CI: 1.069-1.399), American Joint Committee on Cancer-Node (AJCC-N), and tumor stage were independent prognostic factors. The gradient boosting decision tree (GBDT) performed better than other models, with an F1-score of 0.791 and an area under the curve of 0.831.

Male, age ≥65 years, distant metastasis to the bone, liver, and brain are associated with a worse prognosis in PLCNEC patients, while surgery and chemotherapy are associated with improved prognosis. GBDT showed promising performance in predicting 2-year survival, which can serve as a valuable reference for clinical diagnosis and treatment of PLCNEC.