An analysis of the factors involved in the diagnostic accuracy of colposcopically directed biopsy

Abstract

Objective. To compare the results of colposcopically directed biopsy with the final diagnosis established by the analysis of the surgical specimen, and to determine the clinical and colposcopic factors on which the reliability of biopsy is based. Material and methods. Five hundred and sixty-seven women were seen by the same colposcopist who also performed the directed biopsies and/or the endocervical curettage. The final histological diagnosis identified 29 normal aspects (5.2%), 58 low-grade cervical intraepithelial neoplasias (CIN) (10.2%), 448 high-grade CINs (79.0%), 16 microinvasive cancers (2.8%) and 16 occult invasive cancers (2.8%). The influence of several factors - such as age, parity, menopause, pregnancy, history of cervical treatment, site of the squamocolumnar junction, localization, size and severity of the lesions - on the pertinence of the biopsy was studied in a uni- and multifactorial analysis. Results. Colposcopy was satisfactory in 399 patients (70.4%) in whom the colposcopic aspect was consistent with the final histological diagnosis in 81.2% of cases. The global agreement between biopsy diagnosis and final diagnosis was observed in 89.6% of cases. It was 84.2% for low-grade CINs, 95.8% for high-grade CINs, 31.2% for microinvasive cancers and 81.2% for invasive cancers. No clinical or colposcopic factor could be identified as independent factor associated with the diagnostic agreement with the directed biopsy. Conversely, concordance of biopsy was related to the final diagnosis since the only independent risk factors were a high-grade CIN (adjusted risk ratio (ARR) = 1.52, 95% CI = 1.11 - 2.08; p = 0.006) and a microinvasive or invasive cancer (ARR = 0.56, 95% CI = 0.39 - 0.81; p = 0.002). Conclusion. To ensure that a microinvasive cancer has not been overlooked, the excision of high-grade CINs seems to be justified, whatever the clinical status and the colposcopic aspect.