Abstract
BACKGROUND: Cervical cancer, a major cause of cancer-related mortality in women, remains challenging to treat, particularly in recurrent or metastatic stages. Immunotherapy offers promise, but reliable biomarkers for predicting outcomes are lacking. METHOD: A cohort of 204 patients with recurrent or metastatic cervical cancer who underwent immunotherapy was included in the study. Predictive factors were identified using LASSO regression combined with multivariate Cox proportional hazards analysis. Nomograms for progression-free survival (PFS) and overall survival (OS) were constructed by incorporating significant prognostic variables. Internal validation was performed via bootstrap resampling, and clinical applicability was assessed through decision curve analysis (DCA). Risk stratification was evaluated using Kaplan-Meier survival curves, with log-rank tests for comparison. In addition, to enhance the clinical applicability of the model, we downloaded external multi-center validation datasets from public databases, including TCGA (The Cancer Genome Atlas) and GEO (Gene Expression Omnibus). Data from 306 cervical cancer patients were obtained and used as an independent validation cohort. These external datasets allowed for the verification of the developed model, ensuring its robustness and generalizability across different clinical settings. RESULTS: The analysis identified several significant prognostic factors for PFS, including histological type, maximum lesion diameter, CA125 levels, albumin concentration, lactate dehydrogenase (LDH) activity, and neutrophil-to-lymphocyte ratio (NLR). For OS, independent factors included BMI, liver metastasis, CEA levels, hemoglobin concentration, albumin levels, and LDH. The nomogram models demonstrated strong predictive accuracy, with concordance indices (C-index) of 0.706 for PFS and 0.769 for OS. Calibration curves indicated that the predicted and actual outcomes were in excellent agreement. The area under the curve (AUC) values for PFS at 1- and 2-year follow-up were 0.804 and 0.822, respectively, while for OS, the AUC values were 0.880 and 0.781. The risk stratification based on the nomogram scores revealed significant survival differences between high- and low-risk patients, with the high-risk group exhibiting poorer survival outcomes. External validation using data from the TCGA and GEO cohorts confirmed the robustness and generalizability of the nomogram models, further supporting their clinical relevance. CONCLUSION: These nomograms provide a reliable tool for predicting outcomes in cervical cancer immunotherapy, helping to personalize treatment and improve clinical management, especially for metastatic disease.