The dynamics of tumor progression is driven by multiple factors, which can be exogenous to the tumor (microenvironment) or intrinsic (genetic, epigenetic or due to intercellular interactions). While tumor heterogeneity has been extensively studied on the level of cell genetic profiles or cellular composition, tumor morphological diversity has not been given as much attention. The limited analysis of tumor morphophenotypes may be attributed to the lack of accurate models, both experimental and computational, capable of capturing changes in tumor morphology with fine levels of spatial detail. Using a three-dimensional, agent-based, lattice-free computational model, we generated a library of multicellular tumor organoids, the experimental analogues of in vivo tumors. By varying three biologically relevant parameters-cell radius, cell division age and cell sensitivity to contact inhibition, we showed that tumor organoids with similar growth dynamics can express distinct morphologies and possess diverse cellular compositions. Taking advantage of the high-resolution of computational modeling, we applied the quantitative measures of compactness and accessible surface area, concepts that originated from the structural biology of proteins. Based on these analyses, we demonstrated that tumor organoids with similar sizes may differ in features associated with drug effectiveness, such as potential exposure to the drug or the extent of drug penetration. Both these characteristics might lead to major differences in tumor organoid's response to therapy. This indicates that therapeutic protocols should not be based solely on tumor size, but take into account additional tumor features, such as their morphology or cellular packing density.

Progression from a ductal carcinoma in situ (DCIS) to an invasive ductal carcinoma (IDC) involves changes in the surrounding extracellular matrix (ECM) fibril patterns before and during the development of ductal microinvasions, i.e., small cohorts of tumor cells that breach the duct and migrate through the stroma. We used a combination of mathematical modeling (with the hybrid agent-based model silicoDCIS) and advanced image analysis techniques applied to histological and fluorescent images in order to identify the rules of cell-ECM interactions that guide the emergence of various alignment patterns of the ECM fibrils. This includes the three tumor associated collagen signatures (TACS) previously observed in laboratory experiments. This integrated approach provides an in silico tool for testing biomechanical hypotheses of tumor cell-tumor matrix interactions. These findings can be compared to the patient histology samples and may help to define criteria for identification of DCIS to IDC transition and future development of new diagnostic methods. Citation Format: Katarzyna A. Rejniak, Sharan Poonja, Shreya Mathur, Mehdi Damaghi, Marilyn Bui. Dynamics of fibril collagen remodeling in early DCIS invasions: integrating in silico modeling and tumor histology [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 B010.

Many solid tumors are characterized by dense extracellular matrix (ECM) composed of various ECM fibril proteins that provide structural support and biological context for the residing cells. The growing tumor cell colonies are capable of remodeling the ECM structure in tumor immediate vicinity to form specific microenvironmental niches. The changes in fibril patterns of the collagen (one of the ECM proteins) surrounding the tumor can be visualized experimentally using both histology and fluorescent imaging. In particular, three diverse tumor associated collagen signatures (TACS) were identified and related to tumor behavior, such as benign growth or invasion. Here we will use an off-lattice hybrid agent-based model (MultiCell-LF) to identify the rules of cell-ECM interactions that guide the development of various patterns of alignment of the ECM fibrils.

In vivo tumor microenvironments are heterogeneous in their architecture, cellular contents, metabolic landscape, and extracellular matrix (ECM) composition. Moreover, these microenvironments can undergo temporal and spatial changes. This may alter tumor responses to anticancer therapies. To examine the complex and dynamic changes within tumors and in tumor microenvironments, we developed a computational single cell-based model of the three-dimensional multicellular organotypic cultures: Organoid3D. This model was calibrated to experimental data and reproduced growth dynamics and morphologies of four human breast cell lines: a nontumorigenic epithelial MCF-10A, a mildly tumorigenic MCF-10AT1, and metastatic MCF-10CA-1a and MCF-10CA-1d lines grown in various microenvironmental conditions. By explicitly including the fibril structure of the ECM in the model, we examined how ECM properties promote or suppress the growth of in silico tumor organoids. The quantitative integration of experimental data for organoids exposed to different concentrations of doxorubicin allowed us to derive hypotheses on relative importance of microenvironmental factors and chemotherapeutic treatments on organoid growth. Thus, the in silico organoid model interrogated with experimental data provides a tool for fast and broad hypotheses testing and presents an opportunity to explore experimental conditions beyond what is physically feasible in laboratory experiments. Citation Format: Katarzyna A. Rejniak, Sharan Poonja, Jessica Kingsley, Shreya Mathur, Ibrahim Chamseddine, Aleksandra Karolak, Dmitry A. Markov, Lisa J. McCawley. In silico organoids: A model for deconvolution of microenvironmental and drug effects on tumor growth [abstract]. In: Proceedings of the AACR Special Conference on the Evolving Landscape of Cancer Modeling; 2020 Mar 2-5; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2020;80(11 Suppl):Abstract nr A47.

Progression from a ductal carcinoma in situ (DCIS) to an invasive tumor is a major step initiating a devastating and often lethal metastatic cascade. One sentinel event that initiate this process is the development of ductal microinvasions, i.e., small cohorts of tumor cells that breach the basement membrane surrounding the duct and migrate through the extracellular matrix (ECM) leading to irreversible changes in tumor and stromal architecture. We used a combination of advanced image analysis techniques applied to patients’ histology data to extract features which identify specific properties of individual tumor cells inside the duct and on the invasive front. By integrating these histology-based data with a hybrid agent-based mathematical model, we investigated the biomechanical interactions between the cells and the ECM fiber architecture, and microenvironmental physical and metabolic features that define tumor niche prone to microinvasions. The identified physical properties of the matrix elucidate conditions that can facilitate or prevent the progression of such microinvasions. The presented method provides also a tool for quantifying morphological and immunohistochemical properties of individual cells within the mammary ducts and ductal microinvasion, as well as for testing biomechanical hypotheses of tumor cell-tumor matrix interactions. These findings can be compared to the patient histology samples and help define criteria for identification of changes in tumor architecture and future development of new diagnostic methods. Citation Format: Katarzyna A. Rejniak, Sharan Poonja, Shreya Mathur, Jessica Kingsley, Marilyn Bui. ECM mechanical and metabolic architecture during early ductal invasions: integrating in silico modeling, histology-based machine learning and mechanobiology [abstract]. In: Proceedings of the AACR Virtual Special Conference on the Evolving Tumor Microenvironment in Cancer Progression: Mechanisms and Emerging Therapeutic Opportunities; in association with the Tumor Microenvironment (TME) Working Group; 2021 Jan 11-12. Philadelphia (PA): AACR; Cancer Res 2021;81(5 Suppl):Abstract nr LT017.