Background: Inaccurate colposcopic diagnosis results in inappropriate management of cervical lesions. This study aimed to develop a nomogram for predicting the probability of discordance between colposcopic and cold knife conization (CKC) pathological findings.
Methods: We retrospectively reviewed the records of patients colposcopically diagnosed with high-grade squamous intraepithelial lesions (HSIL) who underwent CKC at the Second Hospital of Shanxi Medical University from September 2018 to September 2021. To evaluate the accuracy of colposcopy for identifying HSIL, the pathological results of colposcopy and CKC were compared, with CKC results considered as the comparative gold standard. To determine the factors influencing colposcopic accuracy, the correlations between pathological discordance and clinical parameters were assessed by univariate and multivariate logistic regression. Based on the multivariate analysis results, a nomogram was developed and validated internally.
Results: Data from 607 patient records were analyzed. The colposcopic accuracy rate for identifying HSIL was 72.8%. Multivariate analysis showed that cervical intraepithelial neoplasia (CIN) grade 3 (OR=9.455, P<0.001), positive endocervical curettage (ECC) (OR=5.407, P<0.001), findings of HSIL/atypical squamous cells-cannot exclude HSIL (ASC-H)/atypical glandular cells (AGC) (OR=1.791, P=0.044), and peripheral blood lymphocyte count (OR=0.523, P=0.018) were associated with colposcopic underestimation, while CIN 3 (OR=0.422, P<0.001), positive ECC (OR=0.306, P<0.001), findings of HSIL/ASC-H/AGC (OR=0.423, P=0.004), and peripheral blood monocyte count (OR=7.989, P=0.016) were associated with colposcopic overestimation. The above factors were used to construct nomograms for predicting colposcopic under- and overestimation, which had area under the curve values of 0.815 (95% CI: 0.767-0.863) and 0.742 (95% CI: 0.690-0.793) for under- and overestimation, respectively.
Conclusions: Our results suggested a significant discordance between colposcopic and conization pathological results, which can lead to unnecessary biopsies or delayed treatment. Our nomogram models may help estimate the probability of colposcopic under- and overestimation, thus aiding clinicians to create individualized treatment plans.