Understanding Ductal Carcinoma in Situ Invasion using a Multiscale Agent-Based Model^∗

Ductal carcinoma in situ (DCIS) is commonly treated clinically through surgical resection. Although surgical options exist for resection, it is unclear which is optimal to reduce the likelihood of future invasive disease. This is further complicated by challenges in determining correct surgical margins from disease diagnostics, with mammographic imaging misidentifying surgical margins by as much as 2 cm vs. histological examination. We have implemented a threedimensional, hybrid multiscale model of DCIS to study disease initiation and progression. In order to shed new light on current biological questions and clinical challenges surrounding the disease, we present here an improved version of this model, with more biologically relevant molecular signaling pathways, cell phenotype hierarchies, and duct architecture variation. In particular, a cell necrosis, lysis and calcification pathway has been incorporated into the model to help better understand the relationship between diagnostic imaging and the true extent of disease invasion. We observe that deficiencies in availability of molecular signaling molecules that upregulate cell proliferation may be overcome by dynamic shifts in phenotypic distributions within the disease mass. Hypoxia, necrosis, and calcification together functioned as a hypoxia relief mechanism, and were observed to maintain a consistent distance between the DCIS leading edge and the site of necrosis onset, providing insights for improving surgical margins.