综合免疫炎症表型在接受手术治疗的宫颈癌中的预后价值：生存建模和免疫组化验证

Postoperative recurrence risk in cervical cancer remains heterogeneous, and conventional clinicopathological factors may not fully capture the contribution of the immune microenvironment and systemic inflammation. We investigated whether an integrated immune-inflammatory phenotype combining stromal tumor-infiltrating lymphocytes (TILs) and the systemic immune-inflammation index (SII) could improve recurrence-free survival (RFS) stratification after surgery.

This retrospective cohort study included 612 patients with cervical cancer who underwent primary surgery between January 2020 and December 2025. Stromal TILs were assessed on hematoxylin-eosin sections, and pretreatment SII was calculated from blood counts. An integrated immune-inflammatory phenotype was defined as favorable, poor, or intermediate. Kaplan-Meier analysis, restricted cubic spline modeling, and multivariable Cox regression were performed. Prognostic performance was compared across a clinical Cox model, an immune-extended Cox model, LASSO-Cox, CoxBoost, and random survival forest (RSF) using C-index, time-dependent area under the curve (AUC), integrated Brier score (IBS), calibration, and 36-month landmark decision curve analysis. Tissue-level validation was performed using immunohistochemical assessment of CD8, CD163, and PD-L1.

During follow-up, 119 patients experienced an RFS event. Kaplan-Meier analysis showed significant RFS differences according to TIL category, SII category, and integrated phenotype. Restricted cubic spline analysis demonstrated a significant overall association between SII and recurrence risk, without marked nonlinearity. In multivariable analysis, FIGO IIIC disease, positive margin status, and the poor integrated phenotype remained independently associated with worse RFS. In model comparison, the immune-extended Cox model had the lowest IBS (0.278), LASSO-Cox achieved the highest C-index (0.782) and the best discrimination at 24 and 36 months, and RSF showed the highest 60-month AUC. In 36-month decision curve analysis, LASSO-Cox and RSF showed the greatest net benefit at intermediate thresholds. Immunohistochemical validation showed that the favorable phenotype was characterized by higher CD8+ cell density, lower CD163+ cell density, a higher CD8/CD163 ratio, and higher PD-L1 combined positive score.

An integrated immune-inflammatory phenotype combining stromal TILs and SII was independently associated with postoperative recurrence risk in cervical cancer and corresponded to distinct tissue immune states. This phenotype may provide a practical framework for recurrence risk stratification, while LASSO-Cox and RSF offer complementary prognostic perspectives.