For many cancer types, the wide-spread adoption of cancer screening has increased the detection of pre-malignant lesions (PML). Despite such efforts, screening has had limited impact on overall survival. Clinical guidelines vary widely from watchful waiting (e.g., prostate) to radical surgery and adjuvant treatment (e.g., breast). In absence of reliable progression risk biomarkers and models, these interventions may have deleterious consequences at the two clinical extremes: delay in life-saving treatment or overtreatment. The study of pre-malignant lesions (PML) at molecular level present significant challenges: PML are small, generally archived in formalin. Moreover, the clinical significance of any identified marker can only be assessed after long follow- up, limiting the translational studies to retrospective collections. These hurdles have prevented the development and application of precision medicine approaches and unbiased biomarker to develop models of progression. The current proposal will extend the MCL Pre-Cancer Atlas Pilot Project (PCAPP initiated in September 2017) with the goal to build multi-modal profiles of highly characterized pre-malignant lesions from the 4 target organs (Lung, Breast, Prostate and Pancreas). The four organs included represent a diverse spectrum of histology - pure histology or mixed with invasive lesions - and clinical settings - treatment or active surveillance. Similarly, the selected profiling methods are as comprehensiveas for invasive tumors atlas (whole transcriptome gene expression or DNA mutations) but also innovative, focusing on micro- environment and exploring spatial heterogeneity (multiplex IHC) and, for a few cases, cellular heterogeneity (single-nuclei sequencing). The propose extension will support the completion of the PCAPP and enable a uniform data analysis and sharing with the community.
UCSD Statement of Work The following resources or laboratories will be involved in the work specified in this subcontract. The Oncogenomics Laboratory at Moores Cancer Center: Lead by Dr Harismendy, the Oncogenomics laboratory is located in the Moores Cancer center. Dr Harismendy has established his laboratory to be able to develop molecular assays, sequence in high throughput at the UCSD-IGM core facility and analyzed the results on a compute-cloud compliant with HIPAA regulation. The laboratory includes 1 senior research associate, 3 computational biology students and one programmer. It further contracts with systems and databases administrators located in the division of biomedical informatics or at the San Diego Super Computer Center. The IGM Genomics Center at UCSD: The UCSD IGM Genomics Center offers full services in high throughput sequencing as well as whole-genome genotyping and copy number variation analysis. The IGM Genomics Center offers assay design consultation, sequencing library preparation (DNA/RNA/smallRNA/targeted sequencing), single cell RNA sequencing (10X Genomics), lllumina platform sequencing (MiSeq, HiSeq2500, HiSeq4000 and NovaSeq 6000), and Illumina genotyping and methylation arrays. The UC Davis Center for Genomic Pathology Laboratory provides expert histology and pathology services on a recharge basis. We provide tissue processing/sectioning, routine and special tissue staining, immunohistochemistry, whole slide scanning, database hosting, help with quantitative image analysis and pathology consulting services. Proposed Work: For the budget period we will perform the following work a. PCAPP project management x Continue to organize PCAPP monthly calls: Until December 2020, these calls will be used to keep track on progress. The goal will be to ensure that data is being generated and shared and to address any logistical or technical question that may delay progress. In 2021, the focus of the calls will be slowly transitioning to discuss standardized analysis and present preliminary analysis across teams and organs. x Continue to attend and collaborate with the Human Tumor Atlas Network (HTAN). In particular identify immediate opportunities for collaboration and data sharing. x Complete the data sharing via JPL LabCAS and dbGaP. Data standardization, upload and registration is a complex process and every member will need to be assisted to accomplish these goals. b. PCAPP data generation and analysis support x The UCSD IGM genomic center will be contracted to perfrom Exome library preparation and sequencing of any supplemental DNA samples that need additional coverage or replication x The UCSD team will offer to assist PCAPP teams to analyze and QC their raw data. They will determine whether the JPL cloud can be used to scale up this analysis and will provide forums and discussion board for data generator and analyst to discuss the results and suggest novel iterations. x The UC Davis histology core facility will be contracted to perform high resolution scanning, cell counting and analysis of the multiplex immuno-histochemistry data from the other PCAPP teams. c. Pan-PCAPP analysis, data enrichment and integration Unified data processing: The raw sequencing data (RNA and DNA) from all PCAPP teams will be re-processed at once to generate a uniform call of gene expression, mutations and copy number aberrations. Similarly, the multiplex IHC primary analysis results will be aggregated and processed at once to ensure that no team or organ specific technical differences occurs. The results of the Pan-PCAPP unified analysis will be made available in dbGaP through a new version of the data. Pan-expression analysis: Mitotic grade, immune-infiltration, presence of necrosis, or lesion morphology (papillary, cribriform, tubular, solid) may reveal commonalities across organs. Each component, or expression state, driving these differences will be annotated via gene set enrichment analysis to understand which biological signal and processes may be responsible Pan-mutational analysis: from a set of known driver genes in cancer we will build the oncomap of driver events (mutations or copy number gains and losses), annotate them for their known, likely or unknown pathogenic consequences and determine how their prevalence differs from the drivers identified in more advanced tumors. Pan-spatial immune analysis; across all PML and organs we will determine the relative number of each cell types in the stroma, defining hot, warm and cold immune states and determine their association with expression states or mutational status, after correcting for organ type. We will contract with Dr. Cambell (UCSF) to calculate the EcoScore for each PML, which will summarize a more context depend micro- environmental status, accounting for the clustering and distance between cells. We will determine whether the EcoScore is associated with specific histological features or epithelial expression states, independently of the overall immune-infiltration. Mutational burden and signature analysis: The mutational load (substitution and copy number) will be calculated for all PMLs and tested for association with expression state and histology features. Similarly for PML with sufficient number of mutations, we will determine the contribution of each of the COSMIC mutational signatures [4] to the mutational load and determine whether the presence of any signature is associated with expression state, histology features, or stromal-immune states. Stromal-Epithelial interactions: we will perform an unsupervised integration of the data to investigate stromal vs epithelial interaction. We will first select and abstract multiple epithelial features derived for RNA and DNA analysis above, such as proliferation, burden, presence of key mutational signature, organ-independent expression state (NMF cluster membership), common mutation drivers. We will then use random forest regression with 5 fold-cross validation to determine which features associate with specific stromal immune features derived from the mIHC experiments: hot / cold stroma, immuno-suppressive, proliferative T-cells, EcoScore, etc.
