Abstract P3-13-03: Risk factors for reoperation following breast-conserving surgery integrated into pre- and postoperative models with high accuracy

Abstract

Purpose: Breast-conserving surgery (BCS) is today applied in 60-70% of patients with invasive and in situ cancer and tumor size of 3 cm or less. However, a significant fraction of patients (10-40%) who undergo BCS require a reoperation due to incomplete excision of the tumor despite the evolvement of oncoplastic surgical techniques. We aimed to define risk factors for non-radical BCS due to positive margins in patients with in situ and invasive breast cancer operated with BCS. A specific focus was to identify preoperatively available characteristics which could be informative for the surgeon and patients when planning the primary choice of breast surgery. Methods: Patients with final diagnosis of in situ or invasive cancer who underwent BCS 2015-2016 at Skane University Hospital in Malmo were included in the study. Patients undergoing neoadjuvant therapy were excluded. Data was extracted from mammography/ultrasound files, type of surgery from clinical files and pre-and postoperative pathological diagnosis and radicality from pathological files according to a prespecified protocol. The definition of radicality was according to international consensus (ie no tumor on ink for invasive cancer and 2 mm for in situ cancer).Uni- and multivariable logistic regression analysis were executed separately for pre- and postoperatively available parameters in the SPSS software to calculate odds ratios (OR) for predictors of non-radicality. Separate models for pre- and postoperative characteristics were defined based on data derived in 2015, 2016 was used as a validation set. Accuracy of the models were presented as AUC-values. P -values of ≤0.05 were considered significant. Results: 202 patients diagnosed in 2015 were included of which 43 (21.3%) patients were considered non-radically operated, whereas 231 patients diagnosed in 2016 were eligible and had a reoperation rate of 14.3%. Multivariable analysis included determinants from the univariate analysis and clinically relevant variables with the following results for the preoperative model: mammographic size per mm: OR 1.06 (95% CI:1.01-1.11), p =0.014 , invasive lobular cancer on core needle biopsy vs not: OR 7.3 (95% CI: 1.8-29.4), P =0.005, DCIS diagnosis on core needle biopsy vs not OR 3.60 (95% CI: 1.00-13.02) p =0.051, benign core needle biopsy vs not OR 6.3 (95% CI: 0.9-46.7), p =0.070, oncoplastic surgery and presence of calcifications had p -values > 0.1. In the postoperative model pure DCIS in the specimen and total extent were significant predictive factors in the multivariate analysis. The preoperative model predicting non-radicality had an AUC of 0.81 and the postoperative model had an AUC of 0.86 in the test set (year 2015), the AUC in the validation set (year 2016) was 0.80 for the preoperative model and 0.81 for the postoperative model. Conclusion: Important preoperative prediction factors for non-radicality following BCS derived from core needle biopsies and mammograms have shown promising results and may be helpful when planning the primary choice of breast surgery. An extended study with larger cohort size is desirable. Citation Format: Ellrant J, Plassgard E, Bendahl P-O, Ryden L. Risk factors for reoperation following breast-conserving surgery integrated into pre- and postoperative models with high accuracy [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-13-03.