Abstract P5-18-03: A predictive model for local recurrence in patients treated for ductal carcinoma in situ of the breast (DCIS)

Abstract

Abstract Background: Ductal carcinoma in situ (DCIS) is a heterogeneous precursor, non-invasive breast lesion. There is a lack of specific DCIS molecular predictors of in breast tumour recurrence (IBTR) or progression to invasive breast cancer (IBC) after breast conserving surgery (BCS) +/- radiotherapy (RT). The aim of this was to identify novel biomarkers and combine these with clinical parameters to develop a new model to predict IBTR in patients treated by BCS for DCIS. Methods: A single institution DCIS biomarker discovery study included a case-control matched series of 180 patients (median age 61, range 35-94) treated at the Edinburgh Breast Unit between 2000 and 2010: · 88 patients with low/intermediate grade DCIS treated with BCS alone; 18 recurred within 10 years. · 92 patients with high grade DCIS treated by BCS and RT; 22 recurred within 10 years. Median follow-up was 7.4 years. RNA was extracted from DCIS lesions and whole-genome transcriptomics analysis was performed using Lexogen QuantSeq. Predictive models were generated based upon the most informative genes. Independent validation cohorts are also available and are currently being used for validation. Results: The models developed predict risk of IBTR in patients with low or intermediate grade DCIS treated with BCS alone and high grade DCIS treated with DCIS plus RT. The models were found to be independent of grade and stratify patients into binary groups of high and low risk of recurrence. A promising model was developed based on the expression of 5 genes combined with tumour diameter ≤15mm or >15mm. • In low/intermediate grade DCIS expression levels of a solute carrier family gene, kinetochore associated gene and an immunomodulatory gene are predictive of recurrence. • In high grade DCIS an additional solute carrier and a glutathione S-transferase related gene are predictive of recurrence. • In the training sets the models have 96% (high-grade) and 92% (low/intermediate grade) accuracy of prediction of subsequent recurrence and estimates of IBTR-free survival were highly significant in both groups (<0.0001). Validation of the model by RT-PCR and immunohistochemistry is underway in both the training cohort and an independent validation cohort. Conclusions: · Promising models to predict risk of IBTR in patients treated for DCIS have been developed. · Novel biomarkers that predict recurrence have been identified using new technologies that may have clinical potential. · Independent validation is currently underway. Citation Format: Martinez-Perez C, Turnbull AK, Fernando A, Ekatah GE, Arthur LM, Cartlidge CW, Johns N, Sims AH, Thomas JS, Dixon JM. A predictive model for local recurrence in patients treated for ductal carcinoma in situ of the breast (DCIS) [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 P5-18-03.