Single-cell high-dimensional imaging mass cytometry: one step beyond in oncology

Abstract

Solid tumors have a dynamic ecosystem in which malignant and non-malignant (endothelial, stromal, and immune) cell types constantly interact. Importantly, the abundance, localization, and functional orientation of each cell component within the tumor microenvironment vary significantly over time and in response to treatment. Such intratumoral heterogeneity influences the tumor course and its sensitivity to treatments. Recently, high-dimensional imaging mass cytometry (IMC) has been developed to explore the tumor ecosystem at the single-cell level. In the last years, several studies demonstrated that IMC is a powerful tool to decipher the tumor complexity. In this review, we summarize the potential of this technology and how it may be useful for cancer research (from preclinical to clinical studies).

Keywords: Cellular network; Hyperion Imaging System; Imaging mass cytometry; MIBI-TOF; Tumor microenvironment.

Fig. 1

Imaging mass cytometry: principle and applications. (A) Imaging mass cytometry workflow. Frozen or FFPE tissues are incubated with metal-conjugated antibodies as done for IHC. Then, slides are inserted into the analyzer (Hyperion Imaging System or MIBIscope) for data acquisition. Metals are ionized and quantified by mass spectrometry by time of flight (TOF–MS). Multiplexed images are reconstituted according the metal abundance per pixel. Hyperion Imaging System and MIBIscope images are generated with the respective software system. (B) Hyperion Imaging System (HIS) and MIBIscope acquisition systems. HIS (left panel) uses an UV laser for tissue ablation. Tissue rasterization generates a cloud of biological material that is ionized by inductively coupled plasma. Ions are then filtered by a quadrupole mass spectrometer to discard low atomic mass elements. High mass atomic ions are quantified by TOF–MS. MIBIscope (right panel) uses an O₂⁺ duoplasmotron primary ion beam to generate secondary ions. Single secondary ions are filtered and quantified by TOF–MS. (C) Single-cell file generation. For cell segmentation pixels from TIFF images are classified into nucleus (yellow), cytoplasm (blue), and background (red). All markers can be used for pixel classification. After classification, the cell boundary is determined and a segmentation mask is generated. The combination of single channel images and segmentation mask allows generating a single-cell file suitable for downstream analysis. This file associates, for each cell, its spatial coordinates and the signal intensity of each tested marker. (D, E) Cell annotation and investigation. From the single-cell files, cell subtypes can be identified. (D) The annotation can be done manually, as done during the analysis of a cell suspension by successive gating. (E) For cell exploration, unsupervised clustering can be used and its results can be visualized with dimensional reduction tools (e.g., t-SNE, UMAP) in which a color corresponds to a single-cell cluster. Each cluster is identified on the basis of the expression level of each marker visualized by a heatmap. (F, G) Spatial analysis. (F) The spatial coordinates of each cell are used to identify cells in direct contact (purple) with the cell of interest (red). Then, such neighboring cells can be analyzed on their own to determine their composition. (G) Cell interactions between all identified clusters can be comprehensively analyzed and visualized in heatmaps. X-axis, cell clusters of interest (from); Y-axis, cell clusters in contact with the cluster of interest (to); green dots, interactions; red dots, avoidance between clusters. The color intensity indicates the number of cells in contact with the cluster of interest. From the interaction analysis, a cell network for the tissue can be determined