A comprehensive prognostic model for older patients with advanced non-small cell lung cancer receiving immunotherapy:...

Older patients (aged ≥ 75 years) with advanced non-small cell lung cancer (NSCLC) receiving immunotherapy represent a growing but understudied population, yet few tools incorporate multidimensional host factors that influence prognosis. We therefore developed and externally validated a real-world prognostic model using bedside-available clinical variables, and compared its performance against conventional indices.

This multicenter retrospective study included patients aged ≥ 75 years with advanced NSCLC receiving first-line immunotherapy. Cox regression analyses were performed to identify potential survival-associated host variables. A composite model, based on clinically relevant candidate variables identified through univariate screening, was developed using endpoint-specific L2-penalized Cox regression and internally validated with 1,000 bootstrap resamples. The final model was externally validated in two independent cohorts. Model performance was assessed using Harrell's C-index, inverse probability of censoring-weighted (IPCW) cumulative/dynamic area under the curve (AUC), calibration, decision curve analysis, and Kaplan-Meier survival analysis with log-rank tests. The model was formally compared with conventional indices using paired bootstrap analyses.

A total of 241 patients were included, with 115 in the development cohort and 126 in the pooled external validation cohort. We developed the CALI (Comorbidity-Advanced Lung Cancer Inflammation Index) model integrating comorbidity burden (hypertension, diabetes, chronic obstructive pulmonary disease), recent weight loss, and inflammatory-nutritional status represented by ALI. In the development cohort, 1- and 2-year AUCs were 0.587 and 0.712 for overall survival (OS), 0.584 and 0.708 for progression-free survival (PFS), with significant stratification for OS (P = 0.013) and PFS (P = 0.021). In the pooled validation cohort, 1- and 2-year AUCs were 0.561 and 0.806 for OS, 0.553 and 0.793 for PFS, also significant stratification for OS (P = 0.006) and PFS (P = 0.008). In paired comparisons, CALI showed modest but significant OS improvements over conventional indices, whereas PFS gains were less consistent.

CALI provided modest prognostic discrimination and risk stratification in older patients with advanced NSCLC receiving immunotherapy, with stronger and more consistent performance for OS than for PFS. The model offers a simple, bedside-applicable tool using routinely available clinical variables, and can be easily accessed via our online calculator (https://jccclv.github.io/CALI-calculator) for individualized risk stratification.