Construction and validation of a prediction model for postoperative recurrence risk of cervical high-risk lesions using nomogram analysis.

Abstract

Patients with cervical high-risk lesions face a heightened risk of recurrence following cervical conization, making early prediction of recurrence essential for effective treatment and follow-up. This retrospective cohort study aimed to develop a recurrence risk prediction model using clinical factors to enhance prediction accuracy and guide clinical decisions. The study included 120 female patients undergoing their first cervical conization with positive surgical margins. Clinical data such as age, human papilloma virus (HPV) typing, surgical margin status, colposcopy results, and postoperative treatments were analyzed. Univariate and multivariate logistic regression identified age ≥ 45 years, HPV 16/18 infection, and positive surgical margins at the internal or external os as independent predictors of recurrence. A nomogram model was constructed and validated externally, achieving 90.3% accuracy in predicting recurrence in an additional 31 patients. Decision curve analysis confirmed the model's higher net benefit compared to single-factor predictions. We conclude that the recurrence risk prediction model, based on age, HPV typing, and surgical margin status, offers high accuracy and clinical utility, supporting individualized patient management and precise treatment planning.