Long-term survival prediction in lung cancer patients with type 2 diabetes using machine learning: a 20-year cohort study

Background: Lung cancer is a leading cause of cancer-related mortality, and the presence of comorbid conditions, particularly type 2 diabetes mellitus (T2DM), can significantly impact treatment and survival outcomes. This study aims to enhance our understanding of survival among lung cancer patients with T2DM by identifying key risk factors and developing predictive models. Methods: We conducted a retrospective analysis of diabetic patients diagnosed with lung cancer using data from the Hong Kong Hospital Authority Data Collaboration Laboratory, covering the years 2000 to 2020. Five survival analysis algorithms were employed: Cox proportional hazards regression, LASSO Cox regression, boosting, survival tree, and random survival forest (RSF). Model performance was assessed with time-dependent area under the curve (AUC) and concordance index (C-index). The best-performing model utilized SHAP (Shapley Additive Explanations) analysis to identify critical risk factors. Findings: Our analysis included 5,491 lung cancer patients with T2DM, with an average diagnosis age of 72.59 years and a mean survival time of 30.16 months. Significant risk factors associated with poorer prognosis included older age at diagnosis (adjusted hazard ratio [aHR]=1.06, 95% CI [1.05, 1.06], p<0.001), longer duration of T2DM (aHR=1.05, 95% CI [1.04, 1.06], p<0.001), and smoking behavior (aHR=1.41, 95% CI [1.29, 1.54], p<0.001). Conversely, the use of anti-lipid (aHR=0.84, 95% CI [0.77, 0.93], p<0.001) and anti-diabetic medications (aHR=0.85, 95% CI [0.79, 0.91], p<0.001) was linked to improved prognosis. The RSF model outperformed others with the highest AUC (0.883) and C-index (0.78), identifying age at diagnosis (410.65), duration of T2DM (188.52), smoking status (152.98), sex (75.37), and LDL cholesterol levels (70.62) as the top five influential factors. Interpretation: This study highlights the complex interplay between T2DM and lung cancer prognosis, emphasizing the need for tailored treatment strategies. Our findings suggest that addressing modifiable risk factors, such as smoking and medication adherence, may improve survival outcomes for lung cancer patients with T2DM.