The heterogeneous cells within complex human tumors are like Darwin’s infamous finches on the Galápagos Islands—the most well-adapted survive even the most aggressive treatments. But the effects of natural selection within a tumor can happen seemingly overnight and turn a treatment from promising to ineffective.

Figure 1. Natural selection complicates treatment of highly heterogeneous tumors. Intrinsically resistant cells within a tumor often survive treatment and quickly form a highly treatment-resistant tumor. Effective treatments against heterogeneous tumors must eliminate resistant cells by targeting them before conventional therapy. Treatment strategies include pushing cells to respond via phenotype switching or specifically eliminating resistant cells (Tammela T and Sage J. Annu Rev Cancer Biol 4: 99–119 (2020).).

Highly heterogeneous tumors pose a significant challenge as they leverage natural selection to the detriment of patients, exhibiting high adaptability and resistance to targeted therapies. Overcoming these obstacles requires new therapeutic approaches or combinations that can effectively address the continuous evolution of these tumors.

However, replicating the complexity of highly heterogeneous tumors in laboratory settings proves challenging. Tumor heterogeneity extends beyond cell-to-cell variability to encompass the tumor’s microenvironment, encompassing factors like infiltrating immune cells and vasculature. Existing mouse models, including patient-derived xenografts, often fall short in fully capturing this heterogeneity within and around tumors.

Single-cell RNA sequencing (scRNA-seq) has transformed the study of tumor heterogeneity using patient-derived samples, revealing intricate cellular and genetic diversity within tumors and their microenvironments. However, scRNA-seq methods typically involve tissue dissociation, disrupting natural cell interactions and the tumor microenvironment (TME).

Spatial transcriptomics offers a solution by allowing researchers to visualize biomarkers, spatial interactions, and tumor transcriptomes in their native architecture. This capability enhances the identification of potential drug targets and resistance mechanisms. Investigators have used our Visium Spatial Gene Expression technology to develop reproducible workflows for oncology drug discovery and identify mechanisms of resistance and recurrence in post-treatment patient tumor samples.

Exploring the archipelago of cancer

Researchers from Bristol Myers Squibb demonstrated that spatial transcriptomics can accurately capture the various compartments and interactions within the tumor microenvironment (TME) of heterogeneous tumors using fresh, frozen, and FFPE samples. They analyzed 40 tissue samples from mouse tumor models and patient tumors with varying heterogeneity using Visium Spatial Gene Expression, identifying tumor biomarkers and validating gene expression data through a spatial validation framework integrating gene expression and digital pathology.

A deep learning model was created to annotate the morphology of H&E-stained tissue samples, which was validated by pathologists. The researchers combined spatial transcriptomics analyses to assign each spot to a digital pathology-defined tissue compartment. This colocalization helped validate known biomarkers, identify new biomarkers, and map receptor–ligand pairs within distinct tissue regions.

They first analyzed fresh frozen colon tissue samples from a healthy rat, ensuring the gene expression profiles matched histologically well-organized compartments. This analysis captured the gene expression profiles of hallmark compartments and rare cells. Their approach accurately represented tumor tissue architecture, identifying distinct compartments and aligning gene expression clusters with known biomarkers.

The researchers aimed to show spatial transcriptomics’ effectiveness in characterizing complex, heterogeneous cancer tissue. Using rat and patient tumor samples, they consistently identified biomarkers and interactions. For example, in patient-derived pancreatic ductal adenocarcinoma (PDAC) tumors, they identified known biomarkers like MIF and pairwise interactions such as CXCR4 and its ligand CXCL12.

They developed the BMS Spatial Portal, a visualization application to explore spatial genomics data, allowing others to analyze and gain new insights into cancer and potential therapies based on their methods and findings.

 Surveying the effects of natural selection

Researchers from Johns Hopkins University School of Medicine used Visium Spatial Gene Expression analysis to identify differences in hepatocellular carcinoma (HCC) tumors from patients who did and did not respond to a combined therapy of cabozantinib (an anti-angiogenic) and nivolumab (a PD-1 inhibitor) in a phase 1b clinical trial. HCC, the most common primary liver cancer, is deadly and often resistant to current treatments, with no known biomarkers to predict patient response.

In the trial, five of fifteen patients had a major pathologic response to CABO/NIVO, showing significant tumor necrosis. These responders’ tumors exhibited increased immune infiltration and fewer immunosuppressive macrophages. By analyzing seven HCC samples (four responders, three non-responders) with spatial transcriptomics, researchers found that responders had more immune cells and fibroblasts, with higher expression of immune-related genes (CCL19, CXCL14, CXCL6). Non-responders primarily had cancer cells expressing tumor markers (AFP, IGF2) and genes related to metabolism and proliferation.

Researchers identified critical cell-to-cell interactions and signaling pathways, noting high PAX5 expression in immune cell regions next to tumors in responders, and increased collagen gene expression in areas with cancer-associated fibroblasts. These insights suggest that enhancing B cell activity and inhibiting cancer-associated fibroblasts could improve immunotherapy efficacy for HCC.

A critical finding was that one responder with recurrent disease had a cancer stem cell (CSC) molecular signature, indicating that CSCs, which are resistant to therapies, could be responsible for recurrence. Tumors with CSC signatures had high infiltration of immunosuppressive T cells, not immune-activating B cells. This distinction between responders and non-responders, including identifying markers for resistance and recurrence, aids in refining and personalizing HCC treatments.

Outsmarting evolution

Spatial transcriptomics enables scientists to understand the molecular underpinnings of tumor response and resistance in a cellular context. As evidenced by the studies above, it is particularly useful for interrogating the role of cell-to-cell interactions and intercellular signaling, which plays a key role in the tumor’s interaction with its microenvironment. We’re excited to see how spatial transcriptomics continues to help investigators analyze patient samples, identify new biomarkers and drug targets, and move closer to truly personalized medicine.

References:

• Tammela T and Sage J. Investigating tumor heterogeneity in mouse models. Ann Rev of Cancer Biol 4: 99–119 (2020). doi: 10.1146/annurev-cancerbio-030419-033413
• Yuan Z, et al. Presence of complete murine viral genome sequences in patient-derived xenografts. Nat Commun 12: 2031 (2021). doi: 10.1038/s41467-021-22200-5
• Gonzalez-Silva L, et al. Tumor functional heterogeneity unraveled by scRNA-seq technologies. Trends in Cancer 6: 13–19 (2020). doi: 10.1016/j.trecan.2021.02.001
• Lyubetskaya A, et al. Assessment of spatial transcriptomics for oncology discovery. Cell Reports Methods 2: 100340 (2022). doi: 10.1016/j.crmeth.2022.100340
• Zhang S, et al. Spatial transcriptomics analysis of neoadjuvant cabozantinib and nivolumab in advanced hepatocellular carcinoma identifies independent mechanisms of resistance and recurrence. bioRxiv (2023). doi: 10.1101/2023.01.10.523481
• Sung H, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 71: 209–249 (2021). doi: 10.3322/caac.21660
• Ho WJ, et al. Neoadjuvant cabozantinib and nivolumab convert locally advanced hepatocellular carcinoma into resectable disease with enhanced antitumor immunity. Nat Cancer 2: 891–903 (2021). doi: 10.1038/s43018-021-00234-4
• Cherry C, et al. Computational reconstruction of the signalling networks surrounding implanted biomaterials from single-cell transcriptomics. Nat Biomed Eng 5: 1228–1238 (2021). doi:10.1038/s41551-021-00770-5

Source: Spatial transcriptomics—the next evolution of cancer drug discovery