Inflammatory biomarker-based prognostic model for immunotherapy outcomes in patients with recurrent or metastatic cervical cancer.

210

Background: Conventional predictive biomarkers, including programmed death ligand-1 (PD-L1) expression and tumor mutational burden testing, fail to predict immunotherapy efficacy in patients with recurrent or metastatic cervical cancer (r/m CC), highlighting the need for accessible and dynamically evaluable indicators. Methods: This study aimed to develop an inflammatory biomarker-based prediction model to identify patients with advanced CC who may benefit from anti-PD-1/PD-L1 therapy. Results: A total of 263 patients with r/m CC receiving immunotherapy were randomly divided into training (70%, n = 185) and validation (30%, n = 78) cohorts. The maximum log-rank test was used to determine optimal cutoff values for pretreatment inflammatory markers (neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, monocyte-to-lymphocyte ratio, and systemic immune-inflammation index); posttreatment markers were stratified by comparison with baseline levels. Least absolute shrinkage and selection operator (LASSO) regression was applied for feature selection, and the selected variables were incorporated into a Cox proportional hazards model to establish the prognostic model. Model performance was evaluated using time-dependent receiver operating characteristic (ROC) curves, concordance indices, calibration curves, and decision curve analysis. Conclusions: Optimal cutoff values for pretreatment neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, monocyte-to-lymphocyte ratio, and systemic immune-inflammation index were 2.82, 243.41, 0.39, and 580.10, respectively. The LASSO-Cox model showed strong predictive performance. Calibration curves confirmed consistency between predicted and observed overall survival (OS) or progression-free survival (PFS). Time-dependent ROC curves yielded training set area under the ROC curve values (1-, 2-, and 3-year) of 0.700, 0.740, and 0.759 (OS) and 0.732, 0.835, and 0.757 (PFS), respectively, as well as validation set area under the ROC curve values of 0.715, 0.771, and 0.803 (OS) and 0.732, 0.835, and 0.757 (PFS), respectively. Decision curve analysis demonstrated a stable net clinical benefit. A nomogram was developed, with Kaplan–Meier analysis showing poorer OS or PFS in high-risk patients stratified by the nomogram.