Abstract
The objective was to develop a nomogram for predicting positive margins after cold knife conization (CKC) in patients with high-grade squamous intraepithelial lesion (HSIL). This retrospective study included patients who underwent CKC at Baoding No. 1 Central Hospital between December 2013 and March 2024. Patients were divided into training (between December 2013 and December 2022) and validation (between January 2023 and March 2024) sets. The least absolute shrinkage and selection operator regression was applied to filter and select relevant variables. Multivariable logistic regression was used for nomogram construction. The model performance was evaluated using various methods, including receiver operating characteristics, decision curve analysis, and calibration analysis. The training and validation sets included 985 and 227 patients, respectively. Age (OR = 1.046, 95% CI: 1.028-1.064, P < .001), cervical intraepithelial neoplasia quadrants by punch biopsy (OR = 1.561, 95% CI: 1.348-1.808, P < .001), HSIL type (OR = 1.711, 95% CI: 1.102-2.657, P = .017), and gland involvement (OR = 1.552, 95% CI: 1.073-2.247, P = .020) were associated with positive margins and used for nomogram construction. The predictive model yielded area under the curves of 0.744 and 0.754 in the training and validation sets, respectively. Decision curve analysis indicated a net benefit when using the nomogram, and the calibration curves demonstrated a good fit. This study constructed a nomogram model for predicting positive margins after CKC in patients with HSIL. This nomogram may enable early and accurate patient evaluation, potentially improving clinical outcomes.