Spatial Transcriptomics Advances the Use of Canine Patients in Cancer Research: Analysis of Osteosarcoma-Bearing Pet Dogs Enrolled in a Clinical Trial

Abstract

Purpose: Pet dogs spontaneously develop many of the same tumor types as humans, including osteosarcoma. Recent advances in spatial transcriptomics have improved our ability to utilize formalin-fixed paraffin-embedded tissues collected at canine autopsy. These techniques allow the canine model to be investigated alongside murine models to inform human cancer research. Herein, we present the first application of the GeoMx Canine Cancer Atlas to outcome-linked samples from canine patients enrolled in an osteosarcoma clinical trial.

Experimental design: A tissue microarray of primary osteosarcoma samples was assayed using the GeoMx Digital Spatial Profiler. Samples were stratified by disease-free interval (DFI). Patients within the upper and lower tertiles were assigned to the high- (n = 8) and low-DFI (n = 8) groups, respectively. Analyses included the identification of differentially expressed genes, pathway enrichment, and cell deconvolution.

Results: Genes enriched in high-DFI tumors included PTEN and CDKN1B. Low-DFI tumors were enriched for NCAM1. Pathways enriched in the low-DFI group included MYC, MTORC1, and oxidative phosphorylation. High-DFI tumors were enriched for IFNα response and allograft rejection pathways, among others. Macrophages and CD8+ T cells were elevated in high-DFI tumors.

Conclusions: Gene and pathway enrichment found to be associated with DFI in this study overlap with that described in human patients with osteosarcoma, underscoring the value of the canine model in metastatic osteosarcoma research.

Figure 1.. Gene and pathway enrichment in canine osteosarcoma patients stratified based on their disease-free interval (DFI).

(A) Representative images of H&E and vimentin labeling of a core from the study TMA and Kaplan-Meier curve for the Low and High DFI groups. (B) Illustration showing the genes that are differentially expressed between the High and Low DFI groups. (C) List of the pathways significantly enriched in the Low and High DFI groups based on GSEA pathway analysis.

Figure 2.. Cell deconvolution analysis and comparison to matched bulk mRNA sequencing dataset.

(A) Graphical depiction of cell compartments from tumors in the Low and High DFI groups. (B) Macrophage proportions in the Low and High DFI groups. (C) CD8+ T cell proportions in the Low and High DFI groups. (D) Comparison of cellular proportions between the GeoMx platform and bulk mRNA sequencing where each dot represents a specific cell type (e.g., CD8 T cells) estimated by RNAseq (x-axis) and GeoMx (y-axis) for each patient (n = 9). (E) Comparison of pathway enrichment between the GeoMx platform and bulk mRNA sequencing for cases that had both (n = 9). The top 5 pathways (unshaded) were significantly different between DFI groups across platforms. Pathways shaded in grey or green were significantly different between DFI groups only in the RNAseq or GeoMx datasets, respectively.