Abstract 6352: Defining spatial molecular residual disease by personalized tumor tissue profiling

Abstract Introduction: Localized cancers are increasingly treated with neoadjuvant therapies prior to resection. For operable breast cancers, this approach can aid in breast-conservation, therapeutic response assessment, and prognostication. For patients with residual disease, addition of adjuvant therapy can reduce recurrence risk and improve outcomes. Pathological assessments of neoadjuvant responses, however, imperfectly predict survival, resulting in under- and over-treatment of many patients. Current liquid biopsies are also falsely negative in most patients with residual disease, limiting their actionable utility. Methods: We developed a novel personalized strategy for detection of spatial molecular residual disease (sMRD) and applied it to monitoring neoadjuvant response in 767 spatially-cataloged resection tissues from 29 diverse breast cancers. We utilized Faxitron digital specimen mammography images to generate 3D spatial renderings of each tumor bed that we oriented within the breast and overlaid with sMRD and pathologic detection data from individual samples to evaluate disease distributions. Results: We benchmarked sMRD against histopathological residual disease assessment of individual resection samples from each case, demonstrating strong concordance (sensitivity=88.9%, specificity=86.2%). Spatial case reconstructions captured local and distant recurrence patterns: we identified molecularly-involved, histologically negative margins at eventual local recurrence sites; metastatic progressors exhibited a broader maximal linear spread of sMRD relative to non-progressors (“Dmax”, 5.26cm vs 3.34cm, p=.041) and greater sMRD spatial dispersion interspersed with genomically undetected regions, while responders exhibited more compact disease foci (standard deviation of distances to highest detected sample per case, 1.43cm vs 0.96cm, p=.02). We additionally identified somatic alterations in multiple classes of genes selected through chemotherapy that were associated with clinical recurrence, including genes involved in DNA repair (BRCA1, FANCA), genomic stability (POLE), and ciliogenesis (CSPP1, DNAAF4, CFAP54), suggesting a novel role for loss of ciliogenesis as an emergent chemoresistance mechanism. Finally, we developed a multivariate predictive sMRD model of recurrence in which patients with high versus low sMRD scores experienced inferior recurrence-free survival (log-rank p=.001). Conclusions: We demonstrate that sMRD profiling permits ultrasensitive, quantitative assessment of genomic patterns of tumor evolution associated with clinical recurrence and outperforms routine histopathologic assessment. Our results suggest sMRD could guide risk-adapted personalized adjuvant therapies, heralding a shift of molecular therapeutics from the metastatic to the adjuvant setting. Citation Format: Julia D. Ransohoff, Mia A. Carleton, Sofia Miron Barroso, Nick Phillips, Alisha Maltos, Gregory Bean, Jordan Goldstein, Crystal Zhou, Ruwan Gunaratne, Hitomi Hosoya, George Duran, Michael Khodadoust, James Ford, Robert Tibshirani, Melinda L. Telli, David Kurtz, Ash A. Alizadeh. Defining spatial molecular residual disease by personalized tumor tissue profiling [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6352.