ABSTRACT Objective To investigate factors contributing to discrepancies between colposcopic‐directed biopsy (CDB) and conization in patients with cytology‐positive cervical lesions to reduce missed diagnoses of cervical cancer and unnecessary conizations. Methods A retrospective analysis was conducted, including 1407 patients with positive ThinPrep cytologic test (TCT) results who underwent both CDB and cervical conization (loop electrosurgical excision procedure and cold knife conization), selected from a cohort of 200,814 screened individuals (January 2020–April 2025). The histopathological diagnosis of specimens obtained from cervical conization served as the gold standard, and Firth‐corrected logistic regression models were utilized to evaluate pathological discrepancies and their influencing factors. Results McNemar's test revealed a significant discrepancy between the diagnosis of HSIL+ by CDB and post‐conization ( p < 0.01). The overall concordance rate between colposcopic biopsy and post‐conization diagnoses was 42.7% (601/1407), with a Cohen's kappa coefficient of 0.336, indicating slight agreement. Univariate logistic regression analysis demonstrated that diagnostic downgrading in cone specimens was significantly associated with several factors: age, cytology results, glandular involvement, the time interval between biopsy and conization, and transformation zone (TZ) type. Specifically, downgrading was more likely in patients < 35 years of age, with TCT results showing ASC‐US, LSIL, or ASC‐H, no evidence of glandular involvement, a biopsy‐to‐conization interval < 90 days, and TZ 1. Multivariate logistic regression analysis further identified the colposcopic impression of chronic cervicitis, TCT results of ASC‐US or LSIL, CDB results of chronic cervicitis, LSIL, and no glandular involvement as significant factors associated with downgraded conization diagnoses. Conclusion Significant pathological discrepancies exist between CDB and conization, and clinicians should consider these factors to create more precise and individualized treatment strategies.
Zheng et al. (Sun,) studied this question.