Women treated for high-grade cervical intraepithelial neoplasia (grade 2 or 3) are at elevated risk for developing cervical cancer. Suggested factors identifying women at highest risk for recurrence post-therapeutically include incomplete lesion excision, lesion location, size and severity, older age, treatment modality, and presence of high-risk human papilloma virus after treatment. This question has been intensively investigated over decades, but there is still substantial debate as to which of these factors or combination of factors most accurately predict treatment failure.

In this study, we examine the long-term risk of residual/recurrent high-grade cervical intraepithelial neoplasia among women previously treated for cervical intraepithelial neoplasia 2/3 and how this varies according to margin status (considering also location), as well as comorbidity (conditions assumed to interact with high-risk human papilloma virus acquisition and/or cervical intraepithelial neoplasia progression), posttreatment presence of high-risk human papilloma virus, and other factors.

This prospective study included 991 women with histopathologically confirmed cervical intraepithelial neoplasia 2/3 who underwent conization in 2000–2007. Information on the primary histopathologic finding, treatment modality, comorbidity, age, and high-risk human papilloma virus status during follow-up, and residual/recurrent high-grade cervical intraepithelial neoplasia was obtained from the Swedish National Cervical Screening Registry and medical records. Cumulative incidence of residual/recurrent high-grade cervical intraepithelial neoplasia was plotted on Kaplan–Meier curves, with determinants assessed by Cox regression.

During a median of 10 years and maximum of 16 years of follow-up, 111 patients were diagnosed with residual/recurrent high-grade cervical intraepithelial neoplasia or worse. Women with positive/uncertain margins had a higher risk of residual/recurrent high-grade cervical intraepithelial neoplasia or worse than women with negative margins, adjusting for potential confounders (hazard ratio, 2.67; 95% confidence interval, 1.81–3.93). The risk of residual/recurrent high-grade cervical intraepithelial neoplasia or worse varied by anatomical localization of the margins (endocervical: hazard ratio, 2.72; 95% confidence interval, 1.67–4.41) and both endo- and ectocervical (hazard ratio, 4.98; 95% confidence interval, 2.85–8.71). The risk did not increase significantly when only ectocervical margins were positive or uncertain. The presence of comorbidity (autoimmune disease, human immunodeficiency viral infection, hepatitis B and/or C, malignancy, diabetes, genetic disorder, and/or organ transplant) was also a significant independent predictor of residual/recurrent high-grade cervical intraepithelial neoplasia or worse. In women with positive high-risk human papilloma virus findings during follow-up, the hazard ratio of positive/uncertain margins for recurrent/residual high-grade cervical intraepithelial neoplasia or worse increased significantly compared to that in women with positive high-risk human papilloma virus findings but negative margins.

Patients with incompletely excised cervical intraepithelial neoplasia 2/3 are at increased risk for residual/recurrent high-grade cervical intraepithelial neoplasia or worse. Margin status combined with high-risk human papilloma virus results and consideration of comorbidity may increase the accuracy for predicting treatment failure.