Abstract PO2-13-10: Predicting outcomes in locally advanced breast cancer through profiling post-neoadjuvant tissue-based minimal residual disease

Abstract Background: Neoadjuvant chemotherapy (NAC) for locally advanced breast cancer can down-stage disease prior to surgery, monitor chemo-sensitivity, and provide key prognostic information via pathologic response that guides adjuvant treatment. Greater residual disease as assessed by the residual cancer burden (RCB) index is associated with inferior outcomes but imperfectly predicts recurrence, particularly in triple negative breast cancer (TNBC) where there is heterogeneity of outcomes across RCB scores. As an alternative to RCB and PCR, blood-based minimal residual disease (MRD) from circulating tumor DNA (ctDNA) has been explored. However, poor sensitivity limits ctDNA, which is typically negative post-NAC when treatment decisions are made in patients who later recur. Methods: To address this lack of sensitivity, we developed tissue-based MRD (t-MRD) to assess molecular disease burden in post-NAC breast and lymph node resection tissue. We performed whole exome sequencing on diagnostic tumor and matched germline tissues from five pilot patients with early-stage TNBC with a range of responses to NAC. For each patient, we designed a personalized assay to track tumor-specific SNVs, combined with limited panel of genes recurrently mutated in breast cancer, to enable MRD detection. We then applied this assay to DNA from individual pathology blocks from the tumor resection via ultra-deep targeted hybrid capture. We assessed each sample for the presence of MRD using a Monte Carlo statistical framework to integrate all tumor mutations as previously described (Newman et al, Nature Biotechnology 2016). We aggregated results across all pathology blocks from each case to define each patient’s t-MRD profile and assessed the performance of t-MRD versus gold standard pathologic assessment. Results: We tracked a mean of 136 MRD reporters per case across 119 post-NAC blocks derived from five patients. Twenty-five samples had pathologic residual disease and 28 had detectable t-MRD. There was strong concordance between t-MRD mean allele frequency (AF) and tumor cellularity as assessed by pathology, defined as percent cellularity within the region containing viable tumor (Spearman r = 0.9052, p 0.0001), and excellent classification of true positive pathology samples by genomic sequencing (AUC = 0.9675, p 0.0001). Samples (n = 6) where t-MRD but not pathologic residual disease was detected had a lower mean AF than those where there was also histologic residual disease (p=0.0001). Among the five studied cases, we confirmed a molecular complete response with no t-MRD detected in the RCB-0 case (0/44 t-MRD positive blocks). In the two RCB-III cases, we observed more diffuse t-MRD than that assessed by pathology (n = 4 additional blocks with t-MRD but not pathologic disease detected), and in the RCB-II case, we detected minimal t-MRD in only a single block where pathology did not detect residual disease. Significance: By applying blood-based MRD assessment tools to tumor tissue, we developed t-MRD to profile molecular residual disease at the landmark post-NAC timepoint when disease burden is low and adjuvant treatment decisions are made. We demonstrated high concordance with gold standard pathologic assessment, with increased sensitivity and minimal false positive signal. Further studies assessing the additional prognostic value of t-MRD beyond stage-based prognostication and post-NAC RCB scores are ongoing. Citation Format: Julia Ransohoff, Mia Carleton, Alisha Birk, Gregory Bean, Melinda Telli, Ash Alizadeh, James Ford, George Duran, Michael Khodadoust, David Kurtz. Predicting outcomes in locally advanced breast cancer through profiling post-neoadjuvant tissue-based minimal residual disease abstract. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-13-10.