Development and validation of a prognostic model to independently predict cervical cancer recurrence using the immunophenotyping profiles in peripheral blood.

208

Background: The prognostic value of peripheral blood immunophenotyping for cervical cancer (CC) recurrence remains unclear. This study investigated associations between peripheral blood immune profiles and CC recurrence, developing a predictive model for progression-free survival (PFS). Methods: We analyzed cellular immunity, humoral immunity, and complete blood count (CBC) profiles in 667 CC patients post-primary therapy. A peripheral blood-derived test (PBDT) was developed using multivariate and LASSO Cox regression models to predict PFS. Model performance was evaluated using receiver operating characteristic curves and concordance index (C-index). Results: Among 667 patients, the recurrence group (n = 102) demonstrated significantly elevated B lymphocyte proportions ( P = 1.77e−12), complement C3 ( P = 0.04), complement C4 ( P = 3.14e−05), and immunoglobulin G levels ( P = 0.005), while Cytotoxic T cell proportions were reduced ( P = 0.015). The PBDT LASSO Cox model showed superior performance with higher C-index (0.81; 95% CI: 0.75-0.86) versus multivariate Cox regression (0.78; 95% CI: 0.72-0.84). In validation, the LASSO model achieved excellent discrimination for 1-year (AUC = 0.845; 95% CI: 0.727-0.963), 3-year (AUC = 0.782; 95% CI: 0.684-0.88), and 5-year PFS (AUC = 0.799; 95% CI: 0.699-0.898). Importantly, PBDT emerged as an independent recurrence predictor (HR = 6.62; 95% CI: 4.33-10.14, P = 3.19e−18), outperforming traditional clinical factors including age, radiation therapy, and FIGO stage. Conclusions: Peripheral blood immunophenotyping profiles strongly correlate with CC recurrence. The PBDT model represents a promising non-invasive prognostic tool for predicting CC recurrence, potentially enabling personalized risk stratification and treatment optimization.