Development and validation of prognostic nomograms for patients with cervical cancer and liver metastasis: a SEER-based study.

Background: Cervical cancer with liver metastasis (CCLM) is associated with a dismal prognosis, with a 5-year survival rate significantly lower than that for other metastatic patterns. However, no dedicated prognostic tool currently exists to predict individualized survival outcomes for these patients. This project seeks to develop and evaluate prognostic nomograms for forecasting overall survival (OS) and cancer-specific survival (CSS) in CCLM. Methods: Patients diagnosed with CCLM between 2010 and 2020 were identified from the Surveillance, Epidemiology, and End Results (SEER) database. After applying inclusion and exclusion criteria, 704 eligible patients were enrolled and randomly divided into a training cohort (n=492) and a validation cohort (n=212) in a 7:3 ratio. The optimal age cutoff was determined using X-tile software. Least absolute shrinkage and selection operator (LASSO)-Cox regression was employed to screen prognostic variables and mitigate overfitting. Independent prognostic factors were identified through multivariate Cox regression analysis and incorporated into nomograms. Model performance was assessed using the concordance index (C‑index), time‑dependent receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA). A risk stratification system was developed based on nomogram-derived total scores. Results: The 0.5-, 1-, and 2-year survival rates for CCLM patients were 48.3%, 23.0%, and 8.0%, respectively. Multivariate Cox regression identified chemotherapy [hazard ratio (HR): 0.297, P<0.001], radiotherapy (HR: 0.747, P=0.004), and lung metastasis (HR: 1.493, P<0.001) as independent prognostic factors for OS. For CSS, chemotherapy (HR: 0.296, P<0.001), radiotherapy (HR: 0.760, P=0.008), lung metastasis (HR: 1.543, P<0.001), and histological type (HR: 1.334, P=0.01) were additionally identified. The C-index for the OS nomogram were 0.720 (training) and 0.727 (validation), while those for the CSS nomogram were 0.729 (training) and 0.724 (validation). Area under the curve (AUC) values exceeded 0.70 at all time points in both cohorts. Calibration curves demonstrated good agreement between predicted and observed survival probabilities. DCA confirmed favorable net clinical benefit. The risk stratification system effectively categorized patients into low-, intermediate-, and high-risk groups, with median OS of 11 months versus 1 month for low- and high-risk groups, respectively (P<0.001). Conclusions: This study developed the first specialized nomograms for predicting 0.5-, 1-, and 2-year OS and CSS in patients with CCLM. These nomograms demonstrated robust discriminative ability and clinical utility, providing clinicians with reliable tools for individualized prognosis assessment, risk stratification, and personalized treatment planning. External validation in geographically diverse cohorts is warranted to confirm generalizability.