Abstract NG03: Spatial RNA-seq reveals intratumor heterogeneity and transcriptional substructure in high-grade ovarian cancers
Lee D. Gibbs,Stephen R. Williams,Neil I. Weisenfeld,David W. Craig,Rania Bassiouni,Nigel F. Delaney,Diane Da Silva,Yifeng Yin,Solomon Rotimi,Jennifer Chew,Meghan Frey,Michelle Webb,Jing Qian,Heather Miller,Laila Murderspach,Troy McEachron,Lynda Roman,John D. Carpten
DOI: https://doi.org/10.1158/1538-7445.am2020-ng03
2020-08-13
Tumor Biology
Abstract:Abstract Ovarian cancer is the fifth leading cause of cancer-related deaths in women and is the deadliest gynecological malignancy in the United States. The standard treatment of ovarian cancer is based on debulking surgery followed by platinum- and taxane-based chemotherapy, and has remained the same over the past three decades. Over those years, molecular targeted and combination therapies have been developed and clinically approved, however the overall survival rate has not improved significantly due to chemo-resistance. Further, the majority of patients experience recurrence of treatment-resistant tumors. The genomic diversity within a tumor and the varying cell types within its microenvironment has placed significant importance on heterogeneity and its clinical implications. Intra-tumor heterogeneity has often been blamed for treatment failure in ovarian carcinoma. Consequently, intra-tumor heterogeneity is a key factor driving drug resistance, therapeutic failure, and poor outcomes and poses a significant challenge to personalized cancer medicine. Intra-tumor heterogeneity is a hallmark of cancer where the molecular and cellular interactions within the tumor microenvironment can dictate a cancer's fate. Molecular profiling of bulk tissue specimens using methods such as whole-transcriptome sequencing are limited in their ability to resolve fine grain molecular signatures and hinder our utility to dissect the underlying biology of individual tumors. Although informatics approaches are available that attempt to disentangle tissue heterogeneity from bulk tumor data, emerging spatial whole transcriptome sequencing technologies allow a more precise delineation of cellular and molecular substructure in a comprehensive unbiased approach. Spatial Transcriptomics (ST) enables high-throughput whole transcriptomic sequencing within a single intact tissue by using a glass slide arrayed with barcoded cDNA primers at a resolution of 100um (3-30 cells). This workflow requires no tissue dissociation keeping fragile cell types intact. Resulting data from this workflow is overlaid on the tissue, displayed as a “cluster reference map”, providing comprehensive unbiased transcriptional substructure and unique possibilities for subsequent in situ analysis. The overall goal of this study is to utilize ST to unveil the unexplored landscape of intra-tumor heterogeneity in ovarian cancer and determine its translational relevance. Here, we applied ST to profile gene expression in fresh frozen OCT embedded sections from nine high grade ovarian patients. Three serial 5-micron frozen sections were placed on proprietary 10x Genomics Spatial Transcriptomics (ST) slides and processed using manufacturer specifications. Libraries were sequenced on the Illumina NovaSeq 6000 system and data was processed using 10X Genomics analytical tools. Next, we aggregated the transcriptional profiles of serial sections from each case increasing our power to cluster similar regions and identify differentially expressed genes within these tissues. Using serial sections of solid tumors from each subject we were not only able to profile each section at 100um resolution but also spatially resolve gene expression signatures and cluster regions of tissue based on these signatures. This ST workflow reliably quantitated an average spatial distribution of 19,285 genes per section in our ovarian cancer cohort. Interestingly, within our cohort we have extreme outliers that include primary tumors that did not have any response to standard adjuvant chemotherapy, paclitaxel and carboplatin, and patients that sustained a durable response to standard adjuvant chemotherapy and diagnosed as disease free ≥3 years and then presented with recurrent disease. ST revealed cellular heterogeneity across the entirety of the tumor microenvironment in an anatomically resolved manner. Bioinformatic tools and molecular pathway enrichment were used to annotate tumor, stromal, and immune tissue regions. Our analyses blindly recapitulate specific histopathological annotations spatially through unsupervised gene expression clustering. These data also revealed clear transcriptional substructure, where we identified transcriptionally diverse tumor clones that have unique gene sets associated with distinct molecular processes that relate to oncogenesis (BRCA1, P53, and PTEN Pathway). and potential targetable pathways (EGFR, HER2, PI3K/AKT/MAPK signaling pathways). Additionally, we were able to create spatial heat maps of the inflammatory microenvironment at the single gene level and for gene profiles that define specific inflammatory signals and immune cell profiles. Using this approach, we went on to conduct a refined analysis of key inflammatory genes that have been previously known to play an important role in the immune response of ovarian cancer and developed a spatially resolved inflammation score for each section from each subject. Not only did we observe statistically significant regionalized inflammation scores within each subject, we also found inter-subject differences which may give clinical insight into response to treatment. Surprisingly, this technology also enabled the ability to analyze conventional and emerging immune checkpoint gene expression while accessing the cells and regions of the tissue where they are expressed. Further, our analysis of spatial maps and immune cell quantitation within a primary and recurrent tumor from the same individual revealed dynamic longitudinal changes in inflammation scores and immune infiltration between tumors and spatially within each tumor. These approaches highlight the power of spatial whole-transcriptomic sequencing in solid tumor studies to help unravel the complexity of heterogeneous cancers and provide a comprehensive characterization of transcriptional substructure within a single tissue section. Citation Format: Lee D. Gibbs, Stephen R. Williams, Neil I. Weisenfeld, David W. Craig, Rania Bassiouni, Nigel F. Delaney, Diane Da Silva, Yifeng Yin, Solomon Rotimi, Jennifer Chew, Meghan Frey, Michelle Webb, Jing Qian, Heather Miller, Laila Murderspach, Troy McEachron, Lynda Roman, John D. Carpten. Spatial RNA-seq reveals intratumor heterogeneity and transcriptional substructure in high-grade ovarian cancers [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr NG03.