Periodic Reporting for period 2 - Precision Motifs (Analysis of functional tissue motifs for precision medicine in metastatic breast cancer)
Reporting period: 2022-03-01 to 2023-08-31
Cancer precision medicine aims to match a person's tumor with the right drug or drug combination. To achieve this, the molecular cause of an individual’s disease must be identified, and a treatment chosen that targets this cause to correct the dysfunction. Tumors are dynamic ecosystems with high heterogeneity in tumor, immune, and stromal cell compartments. This complexity is the main obstacle to implementing precision medicine. Consequently, approaches that rationalize and utilize tumor ecosystems for precision medicine are needed. To this end, the project proposed here aims to establish concepts providing a single-cell and spatially resolved representation of multi-cellular assemblies that execute definable sets of anti-tumor or pro-tumor functional outputs. The utility of this approach for precision medicine will be demonstrated in the context of metastatic breast cancer, responsible for nearly 450,000 cancer-associated deaths annually. We will develop a highly multiplexed 3D tissue imaging approach with unprecedented throughput, providing subcellular resolution to analyze a large cohort of metastatic breast cancer tumors. Multi-cellular structures and their function will be identified using community and geostatistical methods, along with probabilistic network modeling. We will then systematically perturb intra-cellular and inter-cellular network of these multi-cellular structures using approved drugs on viable tumor samples ex vivo, measuring drug effects through highly multiplexed imaging. Analyzing these structures in light of clinical data will help derive a novel classification system and identify vulnerabilities in metastatic breast tumors. Ultimately, these relationships and vulnerabilities will be validated in follow-up experiments and eventually in a clinical context. Given their focus on functional outputs, information content, and versatility, the multi-cellular structures we will derive promise to be a powerful tool, potentially guiding appropriate treatments and informing the development of new cancer therapies.
During the reporting period, we achieved several milestones as outlined in the research proposal: On the technology side, we developed a highly multiplexed X-ray imaging method, multi-element Z-tag X-ray fluorescence (MEZ-XRF) imaging. This technique enabled parallel imaging of 20 Z-tag or SABER Z-tag antibodies at subcellular resolution in various human tissues, including breast cancer. We showcased the nondestructive, multiscale repeat imaging capabilities of MEZ-XRF, starting with rapid tissue overview scans followed by more sensitive imaging of low-abundance markers like immune checkpoint proteins. The multiscale, nondestructive nature of MEZ-XRF facilitates highly multiplexed bioimaging across biological scales. To achieve kHz scanning speeds, we combined MEZ-XRF with signal amplification by exchange reaction (SABER)-amplified Z-tag reagents. This method is also highly useful for other metal-based imaging techniques, such as imaging mass cytometry.
On the data analysis side we realized that no multiplexed tissue imaging method is fast enough to analyze large tissue sections across extensive patient cohorts. We therefore developed a statistical framework to determine the number and area necessary to accurately identify all cell phenotypes in a tissue for a given biological or clincial question. We introduced a measure called tissue spatial segregation to improve the design of multiplexed imaging studies. Furthermore, we developed a full data analysis pipeline, automating multiplexed tissue imaging processing and downstream analysis.
On the biomedical side, our proposal aims to describe complex (metastatic) breast cancer ecosystems as multi-cellular assemblies (functional tissue motifs) executing defined functions, and relate these motifs to clinical parameters. To identify such multi-cellular assemblies, multiscale tissue analysis from molecules to single-cell phenotypes, cell-to-cell interactions, and morphology is necessary. We first conducted single-cell RNA sequencing and imaging mass cytometry to comprehensively determine single-cell phenotypes and states in primary and metastatic breast cancer. Comparing single-cell phenotypes of primary breast tumors and matched lymph node metastases in 205 patients, we identified phenotypes and spatial organizations of disseminated tumor cells associated with patient survival. Notably, using the lymph node metastasis tumor cells, we identified high-risk patients within a patient group normally considered low risk. This suggests that molecular characterization of locally disseminated tumor cells is a valuable source of clinically relevant prognostic information for breast cancer. We also examined matched pairs of primary breast tumors and distant metastases using 75 unique antibody targets. Our findings indicate that primary tumors and metastatic sites typically share tumor cell phenotypes, supporting a linear spread of metastases. Except for brain metastases, the metastatic site didn’t significantly influence tumor phenotype composition. We observed higher concentrations of myeloid cells, exhausted and cytotoxic T cells in metastatic sites. This analysis of tumor and immune cell phenotypic composition in distant metastatic breast cancer underscores the heterogeneity of metastatic disease within patients and across distant sites, highlighting myeloid cells as key immune modulators that could be targeted in metastatic sites. We aim to validate some of these findings in viable ex vivo breast cancer samples and hopefully in the future in a clinical study.
On the data analysis side we realized that no multiplexed tissue imaging method is fast enough to analyze large tissue sections across extensive patient cohorts. We therefore developed a statistical framework to determine the number and area necessary to accurately identify all cell phenotypes in a tissue for a given biological or clincial question. We introduced a measure called tissue spatial segregation to improve the design of multiplexed imaging studies. Furthermore, we developed a full data analysis pipeline, automating multiplexed tissue imaging processing and downstream analysis.
On the biomedical side, our proposal aims to describe complex (metastatic) breast cancer ecosystems as multi-cellular assemblies (functional tissue motifs) executing defined functions, and relate these motifs to clinical parameters. To identify such multi-cellular assemblies, multiscale tissue analysis from molecules to single-cell phenotypes, cell-to-cell interactions, and morphology is necessary. We first conducted single-cell RNA sequencing and imaging mass cytometry to comprehensively determine single-cell phenotypes and states in primary and metastatic breast cancer. Comparing single-cell phenotypes of primary breast tumors and matched lymph node metastases in 205 patients, we identified phenotypes and spatial organizations of disseminated tumor cells associated with patient survival. Notably, using the lymph node metastasis tumor cells, we identified high-risk patients within a patient group normally considered low risk. This suggests that molecular characterization of locally disseminated tumor cells is a valuable source of clinically relevant prognostic information for breast cancer. We also examined matched pairs of primary breast tumors and distant metastases using 75 unique antibody targets. Our findings indicate that primary tumors and metastatic sites typically share tumor cell phenotypes, supporting a linear spread of metastases. Except for brain metastases, the metastatic site didn’t significantly influence tumor phenotype composition. We observed higher concentrations of myeloid cells, exhausted and cytotoxic T cells in metastatic sites. This analysis of tumor and immune cell phenotypic composition in distant metastatic breast cancer underscores the heterogeneity of metastatic disease within patients and across distant sites, highlighting myeloid cells as key immune modulators that could be targeted in metastatic sites. We aim to validate some of these findings in viable ex vivo breast cancer samples and hopefully in the future in a clinical study.
We have made significant advancements beyond the current state of the art in several areas:
First, we developed high-speed MEZ-XRF imaging and SABER-IMC, establishing a novel, non-destructive approach for studying tissues. This method enables unprecedentedly fast, highly-multiplexed analysis in 2D. Moving forward, we aim to extend MEZ-XRF for ultrafast 3D tumor analysis, which would further enhance its applicability.
Second, we created a statistical framework to determine the number and area of regions that need to be quantitatively measured by highly-multiplexed tissue imaging. This framework is a crucial aid for the entire field of multiplexed tissue analysis methods, including spatial transcriptomics and proteomics. It guides researchers in designing their experiments more effectively, leading to results that are more reproducible and accurate.
Third, through the use of single-cell RNA sequencing and highly multiplexed tissue imaging, we have comprehensively analyzed tumor ecosystems at the cell phenotype and functional tissue motif levels in both local breast cancer metastasis to the lymph nodes and distant metastasis. This analysis has led to new insights, challenging some existing beliefs about the seeding and evolution of these metastases. For example, we found that the distant site has minimal influence on tumor phenotypes, with the primary tumor being the main determinant of the distant metastatic tumor ecosystem.
In our future work, we plan to investigate whether the identified phenotypes and tissue motifs can be targeted or reprogrammed using drugs. We also aim to assess the practical value of this knowledge about phenotypes and tissue motifs for the treatment of breast cancer patients. This could potentially lead to more effective, tailored treatment strategies.
First, we developed high-speed MEZ-XRF imaging and SABER-IMC, establishing a novel, non-destructive approach for studying tissues. This method enables unprecedentedly fast, highly-multiplexed analysis in 2D. Moving forward, we aim to extend MEZ-XRF for ultrafast 3D tumor analysis, which would further enhance its applicability.
Second, we created a statistical framework to determine the number and area of regions that need to be quantitatively measured by highly-multiplexed tissue imaging. This framework is a crucial aid for the entire field of multiplexed tissue analysis methods, including spatial transcriptomics and proteomics. It guides researchers in designing their experiments more effectively, leading to results that are more reproducible and accurate.
Third, through the use of single-cell RNA sequencing and highly multiplexed tissue imaging, we have comprehensively analyzed tumor ecosystems at the cell phenotype and functional tissue motif levels in both local breast cancer metastasis to the lymph nodes and distant metastasis. This analysis has led to new insights, challenging some existing beliefs about the seeding and evolution of these metastases. For example, we found that the distant site has minimal influence on tumor phenotypes, with the primary tumor being the main determinant of the distant metastatic tumor ecosystem.
In our future work, we plan to investigate whether the identified phenotypes and tissue motifs can be targeted or reprogrammed using drugs. We also aim to assess the practical value of this knowledge about phenotypes and tissue motifs for the treatment of breast cancer patients. This could potentially lead to more effective, tailored treatment strategies.