SPATIAL TRANSCRIPTOMIC PROFILING OF ADVANCED OVARIAN CLEAR CELL CARCINOMA REVEALS INTRA-TUMOR HETEROGENEITY IN EPITHELIAL-MESENCHYMAL GRADIENT

Thang Truong Le, Duncan Yi-Te Wang, Ko-Chen Chen, Simin Yan, Pei-Yu Chu, Tuan Zea Tan, Yi-Chia Chiu, Jia-Yuh Sheu, Caizhi Huang, Jieru Ye, Ya-Ting Tai, Hsueh‐Fen Juan, Fu-Chieh Tsai, Wei‐Chou Lin, Lin‐Hung Wei, Ruby Yun‐Ju HuangbioRxiv 2024 · DOI

TL;DR

This study uses advanced spatial transcriptomics to map the intricate cellular and molecular landscape of ovarian clear cell carcinoma (OCCC), revealing significant intra-tumor heterogeneity. The findings highlight a gradient of epithelial-mesenchymal transition (EMT) that is spatially organized and linked to tumor progression and metastatic potential, offering new avenues for understanding and treating this aggressive cancer.

What Problem Does This Paper Address?

Ovarian clear cell carcinoma (OCCC) is a rare but aggressive subtype of ovarian cancer, particularly prevalent in East Asian women, and notoriously difficult to treat due to its complex molecular features and resistance to chemotherapy. A key challenge in understanding and treating such cancers is intratumoral heterogeneity (ITH) – the presence of diverse cell populations within a single tumor. This heterogeneity isn't just about genetic differences; it also involves how tumor cells interact with their surrounding microenvironment (TME) and how these interactions change as the cancer progresses and metastasizes. Previous studies often lacked the spatial resolution to untangle these complex relationships, leaving a gap in our understanding of how ITH drives OCCC's aggressiveness and metastatic potential. This paper aims to address this by providing a high-resolution, spatially resolved view of OCCC's cellular and molecular diversity.

Methods

The researchers employed a multi-platform spatial transcriptomics (ST) approach. They utilized:

  • NanoString GeoMx Digital Spatial Profiling (DSP): This technology allows for the simultaneous measurement of RNA and protein targets within specific regions of interest (ROIs) on a tissue slide. They used the Cancer Transcriptome Atlas (CTA) panel, which covers a broad range of genes. This enabled them to identify distinct tumor cell subclusters and their associated biological pathways.
  • 10x Genomics Visium Spatial Gene Expression: This platform provides whole-transcriptome spatial profiling at near-single-cell resolution, allowing for a more detailed mapping of gene expression across tissue sections.
  • CosMx Spatial Molecular Imaging (SMI): This single-cell ST technology offers even higher resolution, enabling the identification and characterization of individual cell types and their spatial distribution.
  • Gene Set Enrichment Analysis (GSEA) and Gene Set Variation Analysis (GSVA): These computational methods were used to analyze the functional enrichment of gene sets within different cell populations and spatial locations, revealing key biological processes and pathways at play.
  • Trajectory analysis: This computational approach was used to infer the evolutionary paths and developmental trajectories of different cancer cell populations, suggesting potential founder clones and their progression.
  • In vitro cell line experiments and Chorioallantoic Membrane-xenograft (CAM-X) assay: These experimental models were used to validate findings related to EMT, cellular metabolism, and tumor growth/dissemination.

Datasets

The study utilized:

  • 21 tumors from matched ovarian and metastatic sites: These samples were collected from advanced-stage OCCC patients.
  • In-house and public OCCC cohorts: Used for validation of findings.
  • PanCancer datasets: Used to correlate specific OCCC subclone features with clinical outcomes.
  • High-grade serous carcinoma (HGSC) tumor sections: Included for comparative analysis.
  • Published scRNA-seq datasets: Used for cell type superimposition analyses.
  • Visium data from various cancer types: Used to assess the generalizability of identified features.

Key Insights and Ideas

  • Spatially organized EMT gradient: The study revealed that the epithelial-mesenchymal transition (EMT) isn't uniform within OCCC tumors. Instead, it forms a spatial gradient, with more epithelial cells often found in the tumor center and more mesenchymal cells at the invasive margins and tumor-stromal interfaces. This suggests that the tumor microenvironment actively influences EMT.
  • Founder clone with metastatic potential: A specific cancer cell subtype (subtype f) was identified as a potential "founder clone." This clone exhibited intermediate EMT traits and a tendency to metastasize from the tumor center to invasion/stroma regions, then to the peritoneum, and finally to the colon. This highlights a specific evolutionary trajectory for metastasis.
  • Organ-specific tropism: Different cancer cell subtypes showed a preference for specific organ sites, suggesting that cancer cells adapt to and exploit the unique microenvironments of different metastatic locations.
  • Inverse relationship between OXPHOS and EMT: The study found an inverse correlation between oxidative phosphorylation (OXPHOS) activity and EMT. As cells become more mesenchymal, their reliance on OXPHOS decreases, suggesting a metabolic shift that supports invasion and metastasis.
  • NEAT1 as a potential prognostic marker: High expression of the long non-coding RNA NEAT1 was associated with poorer survival outcomes in OCCC and certain other cancers, suggesting its potential as a prognostic marker.

Key Results

  • Three distinct GeoMx tumor cell subclusters (C1, C2, TME) were identified, enriched in pathways like oxidative phosphorylation, EMT, and inflammation.
  • PanCK-positive tumor cells showed preferential spatial distributions across these subclusters, with Cla cells enriched in the tumor center, and Clb and C2 cells in the tumor periphery, stroma, and necrotic areas.
  • Generic EMT scores revealed that tumor cells in metastatic sites were more mesenchymal than those in primary sites.
  • CosMx single-cell ST identified 9 distinct cancer cell subtypes, with varying organ-specific distributions.
  • Trajectory analysis suggested cancer cell subtype f as a founder clone, with a clear progression from primary to metastatic sites.
  • In vitro studies confirmed that EMT gradients influenced tumor growth and metastatic potential, with more mesenchymal cells showing increased tumor growth and dissemination.
  • High NEAT1 expression was significantly associated with worse PFS and OS in OCCC and lower-grade gliomas/glioblastomas.

Limitations and Future Work

  • Small sample size: The study used a relatively small number of patients (21 tumors) for the initial discovery phase, which the authors acknowledge as a limitation compared to larger-scale studies.
  • Limited resolution for TME interactions: While CosMx provided single-cell resolution for tumor cells, the resolution for TME interactions was limited by the technology used.
  • NEAT1 distribution assessment: The distribution patterns of NEAT1 could not be fully assessed in the Visium dataset due to its exclusion.

Future research directions suggested include:

  • Investigating the functional significance of observed cellular interactions and spatially-distinct gene expression patterns.
  • Developing more comprehensive EMT spectrum models with additional cell lines and patient samples to further elucidate the interplay between EMT and metabolism.
  • Exploring the potential of NEAT1 as a prognostic marker and its role in driving tumor progression in specific niches.
  • Further investigating the overlaps between cancer cell f niches and invasive niche-specific ecosystems in other cancers like glioma.

Summary updated

Abstract

Abstract Intratumoral heterogeneity is intrinsically comprised of molecular alterations of tumor cells and extrinsically from interconnections with microenvironments. This study explores the spatial heterogeneity of ovarian clear cell carcinoma (OCCC), a rare cancer with significance to East Asian women. We profiled 21 tumors from matched ovarian and metastatic sites with spatial transcriptomic (ST) platforms including GeoMx and Visium CytAssist. Three subclusters of OCCC tumor cells enriched in oxidative phosphorylation, epithelial-mesenchymal transition, and inflammation pathways were identified with preferential geospatial localizations in tumor centers and invasive margins/tumor-stromal interfaces. These ST subclusters correlated with epithelial-mesenchymal gradients and prognostic features in larger cohorts. Single-cell ST in CosMx identified 9 cancer cell populations with organ-specific tropism. Trajectory analysis indicated a NEAT1-high subclone having founder characteristics, progressively dominating the metastatic niche, displaying mesenchymal traits. This founder feature correlated with clinical outcomes in PanCancer datasets. Our findings highlight the adaptive and context-specific interactions between cancer cells and microenvironments within spatial complexities.
BibTeX
@misc{le2024spatial,
  author = {Thang Truong Le and Duncan Yi-Te Wang and Ko-Chen Chen and Simin Yan and Pei-Yu Chu and Tuan Zea Tan and Yi-Chia Chiu and Jia-Yuh Sheu and Caizhi Huang and Jieru Ye and Ya-Ting Tai and Hsueh‐Fen Juan and Fu-Chieh Tsai and Wei‐Chou Lin and Lin‐Hung Wei and Ruby Yun‐Ju Huang},
  title = {{SPATIAL TRANSCRIPTOMIC PROFILING OF ADVANCED OVARIAN CLEAR CELL CARCINOMA REVEALS INTRA-TUMOR HETEROGENEITY IN EPITHELIAL-MESENCHYMAL GRADIENT}},
  year = {2024},
  doi = {10.1101/2024.10.19.619181},
  howpublished = {bioRxiv},
}
Figure