Abstract
To develop a postoperative recurrence/metastasis risk prediction model for cervical cancer based on circulating tumor cells (CTCs) and tumor fibrosis distance (TFD), and to validate its biological mechanisms through animal experiments. This model aims to guide preoperative risk stratification and personalized treatment strategies. A total of 148 patients with stage IB-IIA cervical cancer who underwent radical surgery between 2020 and 2022 were retrospectively enrolled. Preoperative CTC counts, TFD measurements, and clinicopathological data were collected. Independent prognostic factors were identified using Cox regression analysis. The model's performance was assessed using receiver operating characteristic (ROC) curves, and risk stratification was evaluated with Kaplan-Meier survival analysis. Additionally, an orthotopic tumor model was established in 60 Sprague-Dawley rats divided into six groups (Control, Model, HTFD, LTFD, HCTCs, and LH). Tumor burden, CD4+/CD8 + ratios, IL-6, TNF-α, MDA, and SOD levels were measured to explore the model's underlying metabolic and immune mechanisms. Patients with poor outcomes had significantly higher CTC counts (30.50 vs. 23.74/5 mL, P < 0.001) and lower TFD values (5.47 vs. 5.98 mm, P < 0.001). CTCs ≥ 28/5 mL (OR = 6.63) and TFD ≤ 5.7 mm (OR = 3.37) were identified as independent predictors of poor prognosis. The combined model yielded an AUC of 0.91, with a sensitivity of 90.6%, specificity of 84.5%, and a negative predictive value (NPV) of 94.7%. Patients were stratified into low-, intermediate-, and high-risk groups, with corresponding clinical management recommendations. In high-risk rat groups, elevated CTC release, reduced CD4+/CD8 + ratios, increased IL-6 and TNF-α levels, and significant oxidative damage were observed, supporting the biological plausibility of the clinical model. Unlike conventional models that mainly rely on clinicopathological features, our model incorporates both CTCs and TFD, providing superior discrimination for recurrence risk and offering a novel integrative approach for postoperative surveillance. This dual validation underscores the translational potential of the model, which may be incorporated into individualized postoperative monitoring strategies and multidisciplinary decision-making.