Predictors of High-Grade Squamous Intraepithelial Lesion Treatment Failure

ObjectiveTo estimate the association between several risk factors and high-grade squamous intraepithelial lesions (HSIL) treatment failure in order to identify predictors. MethodsThe study population included 1,548 Canadian women treated for HSIL who participated in a randomized control trial. HSIL treatment failure was the presence of histologically confirmed HSIL or worse during the two-year follow-up period. This nested-case control study included all 101 cases of treatment failure and controls that were matched 1:1 on treatment center and date of failure. Conditional logistic regression models were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) between each potential predictor and HSIL treatment failure. Independent variables that were examined included age, parity, smoking status, number of sexual partners, condom use, method of contraception, margins, number of passes, diagnosis on conisation, genotype, and number of infecting types. Interactions between smoking and margins and genotype were evaluated. ResultsHaving positive vs. negative margins (adjusted OR=4.05, 95% CI 1.57-10.48) and being positive for Human Papillomavirus (HPV)16 and/or HPV18 vs. any other type (adjusted OR=2.69, 95% CI 1.32-5.49) were predictors of HSIL treatment failure in multivariable models. ORs suggested that older age, more severe lesions, and single-type infections may be at a higher risk of treatment failure but were not statistically significant. The ORs for smoking status, number of sexual partners, condom use, contraception, parity, and number of passes were near the null value. We did not observe any evidence of interaction between smoking and genotype, nor between margins and genotype. ConclusionOnly positive margins and HPV16/18 positivity were predictors for being diagnosed with HSIL or worse within two years of treatment. However, we do not recommend automatic retreatment of those with positive margins because over 90% of those with positive margins did not fail treatment. The predictive value of HPV16 and HPV18 for HSIL treatment failure suggests that high coverage vaccination programs should contribute to a significant reduction in residual/recurrent disease.