OBJECTIVE: To identify risk factors associated with anal intraepithelial neoplasia and develop a model for predicting the likelihood of anal intraepithelial neoplasia in heterosexual women.
METHODS: A prospective cohort of 327 patients from 2006 to 2011 with a biopsy-confirmed diagnosis of genital intraepithelial neoplasia (vulvar, vaginal, or cervical) underwent both anal cytology and anoscopy. Variables significant between those with and without anal intraepithelial neoplasia were identified using logistic regression. A forward stepwise regression analysis was carried out to identify a combination of variables that effectively predicted anal intraepithelial neoplasia.
RESULTS: In the cohort of analyzed patients, 64 (46.7%) women with anal intraepithelial neoplasia were identified, yielding a prevalence of 19.6%. Immunosuppression, vulvar dysplasia, multiple sexual partners (more than four), smoking history, and history of anal sex were positively associated with anal intraepithelial neoplasia (P<.05). These variables were analyzed using forward stepwise logistic regression. The final model used the presence of any two of three risk factors (vulvar intraepithelial neoplasia [VIN], immunosuppression, and history of anal sex) to predict 38.8% of anal intraepithelial neoplasia in our population. Area under the receiver operating characteristic curve for two of three of the factors was 0.708 (P<.05). This model has a negative predictive value of 88.2% (95% confidence interval [CI] 0.83–0.92) and positive predictive value of 43.1% (95% CI 0.31–0.56).
CONCLUSION: A simple predictive model based on the presence or absence of two of three risk factors (VIN, immunosuppression, and history of anal sex) can be used by the clinician to quantify anal intraepithelial neoplasia risk in women with genital dysplasia.
LEVEL OF EVIDENCE: II