An Omic and Multidimensional Spatial Atlas from Serial Biopsies of an Evolving Metastatic Breast Cancer

Abstract

SummaryMetastatic cancers often respond to treatment initially but almost universally escape therapeutic control through molecular mechanisms intrinsic to tumor cells, as well as extrinsic influences from immune cells, stroma, and structural microenvironments. We explore the extent to which we can learn these mechanisms and associated therapeutic vulnerabilities by linking comprehensive molecular and multiscale imaging analyses of tumor biopsies to detailed clinical information for a patient with metastatic breast cancer. Our analyses of three serial biopsies include DNA, RNA, and protein profiling, plus quantification of tumor microenvironment composition using multiplex immunostaining and electron microscopy. Features from these analyses are linked to treatments and treatment-response measurements, including CT and PET images, circulating tumor DNA and tumor protein levels in blood, and treatment-limiting readouts of blood counts and liver function. Importantly, the measurement modalities unique to this study complemented routine diagnostics to suggest actionable mechanisms of response and resistance for each treatment phase.

Robust Segmentation of Cellular Ultrastructure on Sparsely Labeled 3D Electron Microscopy Images using Deep Learning

“…We evaluated the proposed ResUNet on 3D FIB-SEM images of three longitudinal tissue biopsy samples (Bx1, Bx2 and Bx4) acquired over three phases of cancer treatment from a single patient with metastatic ER+ breast ductal carcinoma and one biopsy sample (PTT) acquired from a patient with pancreatic ductal adenocarcinoma (PDAC). Extensive additional information about the three biopsies from the patient with metastatic breast cancer are available (Johnson et al, 2020). For each dataset, a ResUNet model was trained using a small subset of manually labeled images and the trained model was used to generate segmentation masks on the rest of the unlabeled images.…”

“…Multiple imaging technologies, including multiplex immunohistochemistry (Kalra and Baker, 2017), cyclic immunofluorescence imaging (Lin et al, 2018) and electron microscopy (EM) (Riesterer et al, 2020), have been used to map cellular states and inter-and intracellular interactions. EM is an important component in this imaging suite, providing nanometer resolution views of intra and intercellular interactions (Johnson et al, 2020) that are not apparent in images generated using light microscopy. This complete picture of spatial relationships can reveal potential therapeutic targets that can be related back to the macroscale heterogeneity and microenvironment of the tissue.…”

“…SEM has also been used to examine cancer-derived cultured cells at nanometer resolution to gain insights, for instance, into cancer cell signaling networks (Pham andIchikawa, 1992, López et al, 2017). The ultrastructure of inter-and intracellular interactions in 3D FIB-SEM images of human cancer biopsies have been explored by our group with emphasis on aspects of nanobiology that might affect therapeutic response (Johnson et al, 2020). Those studies reveal intricate interactions between tumor and stromal cells via micrometers long filopodia-like protrusions (FLP) suggesting the formation of the stomal barrier possibly driven by cytokine-receptor interactions that could be manipulated therapeutically to increase drug access and immune surveillance (Bazzichetto et at., 2019).…”

“…EM image segmentation in general is difficult due to similar intensity and texture variations across all parts of the image (Karabağ et al, 2020). The high density of ultrastructural features and the convoluted and intertwined nature of cancer and stromal cells makes it even more challenging (Johnson et al, 2020). The ultrastructure of tumor cells and the tumor microenvironment is very different from the widely studied neural and other normal cells (Baba and Câtoi 2007, Zink et al, 2004, Coman and Anderson., 1955.…”

The Human Tumor Atlas Network’s beginning steps toward the future of collaborative multi-omic discovery

“…In this issue of Cell Reports Medicine , Brett Johnson and colleagues, as part of the Human Tumor Atlas Network (HTAN), combine comprehensive single-cell genomic sequencing and histologic imaging to provide a tour-de-force biological narrative of a woman’s metastatic breast cancer as it evolves in space and over time. 1 Funded by the Cancer Moonshot initiative, the goal of the HTAN is to provide an efficient means of sharing high-throughput sequencing and imaging data across multiple institutions, with standardized workflows for data collection, processing, and presentation. 2 Distinguishing features of the HTAN database include: (1) serial analyses of tumors and blood ranging from pre-malignancy to metastatic disease, (2) comprehensive clinical annotation of treatment history and clinical response, and (3) a focus on developing spatial methods for integrating single-cell genomic data with histopathologic findings.…”