Abstract PD5-01: Magnetic resonance imaging insights from an active surveillance cohort of women with DCIS

Abstract

Abstract Purpose: Standard treatment for ductal carcinoma in situ (DCIS) involves surgical excision usually with radiation therapy or mastectomy and often endocrine therapy, treatments that are the same as for invasive cancer (IDC). It is clear, however that not all patients with DCIS will develop IDC and so the question is who benefits from early surgical intervention? Active Surveillance (AS) with endocrine risk reduction presents an opportunity to improve outcomes by identifying those lesions that can be managed with endocrine therapy alone without immediate surgical intervention. An active surveillance cohort of women who chose not to have surgery at diagnosis were followed with serial MRIs to identify the imaging features associated with successful AS and those with IDC. Methods: Patients with DCIS were enrolled in MRI surveillance studies between 2002 and 2019 and analyzed retrospectively with IRB approval. Per medical record review, these patients sought to avoid surgical intervention. Inclusion criteria included at least two breast MRIs performed for purposes of surveillance. The final cohort of patients included 64 cases with at least two breast MRIs, with 27 patients having more than 4 MRIs. All breast MRIs consisted of routine sequences, including both pre and post IV contrast images with at least 2 post-contrast time points. Lesion conspicuity, change in lesion, background parenchymal enhancement (BPE), change in BPE, and likelihood of invasive cancer at each MRI timepoint, were subjectively measured independently by two breast radiologists. A Likert scale was used to grade each imaging feature. Radiologists were blinded to the clinical outcome of whether the patient had IDC at surgery or not. Input variables included all imaging features collected and classification trees were trained and bootstrapped on 90% of the data using recursive partitioning to distinguish imaging features predictive of clinical outcome. Proportionality tests were conducted to test whether IDC was associated with age, menopausal status, and breast composition among other clinical variables.Results: Women in the cohort had a mean age of 53.6 years (range 29.8 to 78.9). 98.3% were HR+. Of the 64 cases in the cohort, 57 received endocrine therapy (89.1%). A total of 31 cases (48.4%) eventually had surgical excision and 33 (51.6%) remained on AS. At surgery, 17 patients had IDC (26.6%). Classification trees revealed that the most distinguishing features in the model correlating with IDC were if the lesion was distinct from background at MRI timepoint 1, an increase in BPE between MRI timepoints 1 and 2, and an increase in the lesion size or conspicuity between MRI timepoints 1 and 2. At diagnosis, 56.3% demonstrated a more diffuse pattern of enhancement where the DCIS lesion was not distinguishable above background. Of those with lesions that did not stand out above background at diagnosis and whose BPE did not increase, only 1 of 31 patients developed IDC (3%) with mean follow up of 4.62 years. Patients with IDC were proportionately older (>60) and post-menopausal (73% with IDC were postmenopausal), although only 57% were postmenopausal at diagnosis. Other variables such as breast composition were not enriched in either the IDC versus the non-IDC population. Conclusion: Our study suggests that imaging markers such as BPE and conspicuity of lesion enhancement may provide information to better understand and stratify the risk of DCIS and avoid overtreatment. Importantly, MRI may provide insight as to when a diagnosis of DCIS is more likely to be a global risk factor amenable to endocrine risk reduction, versus a lesion best treated with surgical excision. Pathology correlation is underway. We are currently developing methods to improve reproducibility and harmonization between radiologists, and performance on a validation set will be presented. Citation Format: Heather Greenwood, Rita Freimanis, Case Brabham, Rita Mukhtar, Gillian Hirst, Paul Kim, April Liang, Laura Esserman, Nola Hylton, Amrita Basu. Magnetic resonance imaging insights from an active surveillance cohort of women with DCIS [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PD5-01.