Segmentation: a novel approach to immune profile analysis of the human lymph node microenvironment (P3261)

Abstract

Abstract In the lymph node microenvironment, many cells interact to provide cues for immune cell activity. Understanding the composition and spatial relationships between cells is key in deciphering mechanisms responsible for immune responses. Our novel method expands the use of immunohistochemistry (IHC) by generating quantitative immune profiles. Our integrated approach includes multiplexed IHC, high-resolution multispectral automated whole-slide scanning, and image analysis via software. Our image analysis segments cells into phenotypes and tissues into regions. In this pilot study, we compared tumor-infiltrated (Tpos) and tumor-free (Tneg) tumor-draining lymph nodes (TDLNs) from breast cancer patients. We segmented images into non-immune cell (NIC), extrafollicular (ExF), follicular (F), and tumor tissue regions. We segmented cells by five-color IHC into T cell (CD4, CD8, and Foxp3) and tumor (cytokeratin) phenotypes. Globally and compartmentally (NIC, ExF, and Tumor), there were more immune cells in Tpos TDLNs. However, in F regions, there were less immune cells in Tpos TDLNs. There were more NIC regions but less ExF and F regions in Tpos TDLNs. We also found less NIC, ExF, and F regions in relapsed vs. disease-free patients. Our segmentation analysis may provide valuable data on cell distribution and spatial relationships within specific tissue regions. Our next steps are to use this method to help define mechanisms of action for tumor metastasis and predict clinical outcome.