Prognostic scores after surgical treatment for cervical intraepithelial neoplasia: a proposed model and possible implications for post-operative follow-up.

ABSTRACT

OBJECTIVE: To develop a prognostic model for women who underwent surgical treatment for cervical intraepithelial neoplasia. DESIGN: Cohort study. Patient inclusion and follow-up occurred retrospectively and prospectively. SETTING: Barretos Cancer Hospital, Barretos, São Paulo, Brazil. POPULATION Women (n = 242) diagnosed with cervical intraepithelial neoplasia who were submitted to conization. METHODS: Immediately prior to surgical treatment, a cervical cytology sample was collected from each individual included in the study by endocervical brushing and stored in a preservative solution with methanol. A human papilloma virus-DNA test was conducted using an aliquot of the endocervical brushings. The surgical specimens were subjected to immunohistochemical analysis of p16 (immunohistochemical analysis 4a) protein expression. MAIN OUTCOME MEASURES: Two-year disease-free survival rates calculated for each study variable. Identified variables in the multivariate Cox model were used for elaboration of prognostic scores. RESULTS: Variables associated with outcome included age (p = 0.033), tobacco use (p < 0.001), final histopathological diagnosis (p = 0.007), surgical margins (p < 0.001), high-risk human papilloma virus status (p = 0.008), human papilloma virus-16 status (p < 0.001) and immunoexpression of p16 in the cytoplasm (p = 0.049). By the Cox model, independent risk factors for disease recurrence/persistence were tobacco use (hazard risk = 3.0; 95% confidence interval 1.6-5.6), positive surgical margins (hazard risk = 3.2; 95% confidence interval 1.6-6.1), human papilloma virus-16 (hazard risk = 3.3; 95% confidence interval 1.6-6.9) and age over 45 years (hazard risk = 2.7; 95% confidence interval 1.1-6.6). CONCLUSIONS: Establishment of a prognostic score can represent a valuable tool for determining the risk of cervical intraepithelial neoplasia recurrence after conization. The use of clinical (age and tobacco use), pathological (surgical margins) and molecular (human papilloma virus-16 genotyping) factors can facilitate more appropriate patient follow up according to risk stratification.