The risk for residual AIS/CIN3+ after the first conservative surgical procedure for cervical adenocarcinoma in situ - A Dutch retrospective cohort study.

Introduction: Conservative treatment options for cervical adenocarcinoma in situ (AIS) include a large loop excision of the transformation zone (LLETZ) and cold-knife conisation (CKC). Since each subsequent intervention might have adverse effects, this study aimed to 1) describe the risk of residual AIS, cervical intraepithelial neoplasia (CIN) grade 3, and cervical cancer (combined AIS/CIN3+) after the first surgical procedure and 2) develop a prediction model for residual AIS/CIN3+.

Methods: AIS/CIN3+ incidence was assessed in Dutch patients with AIS who underwent a LLETZ or CKC between 1990 and 2021. The discriminative performance (area-under-the-curve, AUC) for AIS/CIN3+ was compared between a prediction model including four prognostic variables (margin status, age, type of surgical procedure, history of CIN) and a prediction model based on surgical margins only.

Results: 4115 patients were eligible, including 2363 patients (57.4 %) treated by LLETZ and 1752 patients (42.6 %) treated by CKC. Of the 3585 patients with known follow-up, 549 patients (15.3 %) were diagnosed with residual AIS/CIN3+, of which 409 of 2152 patients after LLETZ (19 %) and 140 of 1433 patients after CKC (9.3 %) (p < 0.001). The discriminative performance of the prediction model including four prognostic variables was higher (AUC = 0.80, 95 %CI0.78-0.82) than predictions based on surgical margins only (AUC = 0.74, 95 %CI:0.71-0.76). By applying the prediction model, the number of retreatments after unclear margins could decrease by 15.2 %.

Conclusion: The prediction model developed in this study enables better quantitative assessment of the risk for residual AIS/CIN3+ after the first conservative surgical procedure for AIS compared to surgical margins only. This model could be used to guide decisions on re-excision in patients with AIS.