Pediatric cancer is an urgent and important problem that requires highly precise diagnostics and therapeutic approaches, however many of the necessary tools have yet to be developed.
Around 60% of pediatric cancers are solid tumors, and the prognosis of these cancers is poor with mortality as high as 40%. Despite current therapies such as chemotherapy and radiation, relapse may still occur with mortality up to 66%.
Advanced solid tumors are highly complex and extremely successful at evading the immune response. They are characterized by aberrant cell proliferation and vascularization, resulting in a hypoxic environment inhospitable to most cells, as well as containing their own immunosuppressive immune cells.
Dr. David Steffin, a pediatric oncologist and assistant professor at Texas Children’s Hospital/Baylor College of Medicine, Houston, joined us for a webinar to discuss how he and his team have implemented CODEX® to aid his efforts in developing novel treatments for children with solid tumors.
Pushing the frontiers of CAR T-cell therapy using spatial analysis
While there has been significant progress in differentiating histologic characteristics and identifying genetic mutations in pediatric cancers, the interaction of tumor cells with stromal and infiltrating immune cell subsets is not broadly defined. In the context of refectory pediatric cancers, immunotherapies have become an attractive option. These therapies include but are not limited to checkpoint inhibitors, monoclonal antibodies, vaccines, and the focus of Dr. Steffin, adaptive cellular therapies, namely CAR T-cells.
To truly understand the complex mingling of cells Dr. Steffin and his team have decided a focus on spatial context as the key factor for unlocking the answers to many complex questions posed by cancer.
Spatial biology is the study of tissues or cells in their given microenvironment within the body. This is particularly important when we think about cancers. Solid tumors can be highly heterogeneous. For example, variability may exist between tumors within a single biopsy, a single patient, or between patients. Understanding how cancers differ between patients in addition to cell-cell interactions creates an opportunity to hone in on specific drug targets.
Immunohistochemistry– though useful for spatial assessment – is limited to one or a handful of markers at a time, making it challenging to evaluate the protein expression of tissues in tandem. Single-cell RNA sequencing on the other hand is useful to identify changes in gene expression but does not provide information on cellular organization.
How CODEX can help
To overcome the limitations of immunohistochemistry and single cell RNA sequencing, Dr. Steffin has chosen CO-Detection-by-indEXing (CODEX), a highly multiplexed imaging technique that builds on both the single-cell aspects of single-cell technologies enabling higher resolution with the spatial information of immunohistochemistry with more biomarkers. It uses a theoretically unlimited number of biomarkers to identify larger groups of cell subsets and characterize the overarching cellular interactions in the same space.
Dr. Steffin and his team set out to define novel immune escape mechanisms in solid tumors based on predictive biomarker expression, immune cell localization, and cellular neighborhood formation with the ultimate goal of identifying specific therapeutic targets. To do this, individual solid tumor tissue samples from nodular sclerosing and mixed cellularity tumors were compared to assess distribution of similar cell phenotypes within different regions of each tumor. Machine learning algorithms were used to delineate cell size and shape, followed by clustering and segmentation algorithms to identify cell-cell interactions.
Using various validated antibody panels, highly multiplexed images were generated using CODEX to assess the pediatric lymphomas. These images were extrapolated onto XY planes to annotate phenotypes and conduct distance calculation between cells.
Dr. Steffin also used CODEX to identify specific regions of tissue that have common cellular architecture and contrast these regions with other tissues.
Source: Dr. David Steffin
The above image shows heatmaps generated using sample phenotype outputs and Pearson correlations coefficients generated to assess cell colonization in an unsupervised fashion and cellular interactions. This type of analysis allowed the team to identify exhausted T cells in close proximity to PD-L1+ tumor cells and fibroblasts.
Source: Dr. David Steffin
The use of dimensionality reduction (above, including UMAPs and TSNE maps) allows annotation of regions in a given tissue which can then be contrasted with other tissue samples. These data can then be used to conduct nearest neighbor calculations. In the case of the mixed cellularity tumors the team was able to determine that there was a higher proportion of dysfunctional T cells in close proximity to PD-L1+ cells, suggesting PD-L1 positivity is exerting an immune suppressive effect on these types of tissues.
Several microenvironment factors inhibit the effects of immunotherapy for solid tumors and the question remains: “How do we identify what inhibitory factors are in place so that we can make a better treatment?” Using CODEX, we can designate individual cells with different phenotypes, such as expressing one or multiple antibodies at once, and in this case the presence and function of a CAR T-cell.
To learn more about spatial analysis of solid tumors and CAR T-Cell therapy, watch Dr. Steffin’s full presentation on demand