Abstract B015: MRI features can identify DCIS patients who can be managed with endocrine therapy alone

Abstract

Abstract Background: Ductal carcinoma in situ (DCIS) represents almost a quarter of newly diagnosed breast cancers in the US. It is currently treated like a stage 1 cancer, with surgical excision, radiation, and endocrine therapy (ET). Not all cases will progress to invasive cancer and thus we may be overtreating some lesions. Using a cohort of women who participated in an Active Surveillance (AS) with neoadjuvant ET and serial MR imaging, we aimed to identify features on MRI that could stratify DCIS lesions with low- and high-risk of progressing to invasive cancer. Materials/Methods: This was an IRB-approved retrospective study of women with DCIS who chose AS for treatment and were enrolled in MR imaging studies between 2002 and 2020. All women in this study had at least 2 breast MRI scans, which consisted of normal sequences with two post-contrast timepoints. Two radiologists, who were blinded to outcomes, analyzed MRI sequences for each patient at each timepoint for features that may be indicative of risk. Baseline imaging features collected were utilized in an RPART recursive partitioning algorithm that created a classification tree based on a DCIS lesions likelihood to progress to invasive cancer. Our final cohort consisted of 63 cases of DCIS (62 women, 1 with bilateral disease) with a mean age of 58.3 years (range of 29.8-78.8 years). All 62 patients (63 cases, 100%, 63/63) agreed to take hormone therapy for their DCIS. Results: The RPART algorithm showed two baseline features that were important in predicting risk: 1) the presence of a lesion that is distinct from the surrounding tissue and 2) background parenchymal enhancement (BPE). It separated our cohort into three sub-groups based on risk of progression: A (n=32) with the lowest-risk, B (n=17) with the second lowest-risk, and C (n=14) with the highest-risk. In A and B, 6 cases of DCIS progressed to invasive cancer (12.2%, 6/49). These groups had an enrichment of mild, moderate, and marked BPE. Group A (12.5%, 4/32 progressed to IDC) did not have any lesions that were distinct above BPE. All cases in B (11.8%, 2/17 progressed to IDC) had lesions that were distinct above BPE initially but reduced in response to endocrine therapy. In C, 10 cases of DCIS progressed to invasive cancer (71.4%, 10/14). In this group, the combination of minimal BPE and a distinct lesion above BPE was enriched (78.6%, 11/14). Conclusion: Our results suggest that imaging features, such as BPE, lesion distinctness and change in BPE and lesion provide information about risk for invasive cancer. MRI features provide insight as to whether risk is diffuse or focal, and which women can safely undergo AS with hormone treatment and avoid surgery, and which women are at higher-risk and have focal lesions better treated with surgical excision. We plan to validate the use of these features to triage patients in an upcoming prospective multisite trial to stratify DCIS patients for active surveillance vs. surgical resection. Citation Format: Cristian K. Maldonado Rodas, Heather Greenwood, Rita Freimanis, Alexander D. Borowsky, Gillian Hirst, Nola Hylton, Laura Esserman, Amrita Basu. MRI features can identify DCIS patients who can be managed with endocrine therapy alone [abstract]. In: Proceedings of the AACR Special Conference on Rethinking DCIS: An Opportunity for Prevention?; 2022 Sep 8-11; Philadelphia, PA. Philadelphia (PA): AACR; Can Prev Res 2022;15(12 Suppl_1): Abstract nr B015.