Single-cell atlas of tumor clonal evolution in liver cancer

Lichun Ma, Limin Wang, Ching-Wen Chang, S. Heinrich, Dana A. Dominguez, Marshonna Forgues, Julián Candia, Maria O. Hernandez, Michael C. Kelly, Yongmei Zhao, Bao Tran, Jonathan M. Hernandez, Jeremy L. Davis, David E. Kleiner, Bradford J. Wood, Tim F. Greten, Xin Wei WangbioRxiv 2020 · DOI

TL;DR

This study presents a single-cell atlas of liver cancer, revealing the complex evolutionary trajectories of tumor cells and their impact on patient prognosis. By analyzing ~57,000 cells, researchers identified osteopontin (SPP1) as a key driver of tumor evolution and immune microenvironment reprogramming, offering new avenues for therapeutic intervention.

What Problem Does This Paper Address?

Liver cancer, a leading cause of cancer-related deaths, is characterized by significant intratumor heterogeneity (ITH). This heterogeneity poses a major challenge for effective treatment, often leading to drug resistance and treatment failure. Understanding the evolutionary dynamics of tumor cell populations and identifying the drivers of this evolution are crucial for developing better diagnostic and therapeutic strategies. While bulk sequencing has provided evidence of ITH, it lacks the resolution to define distinct tumor subclones and their functional consequences. This paper aims to address this gap by creating a detailed single-cell map of liver cancer to unravel tumor cell clonality and evolutionary pathways.

Methods

The researchers employed single-cell RNA sequencing (scRNA-seq) to profile liver tumors from 37 patients. Key methods include:

  • Machine Learning-based Consensus Clustering: Developed to determine tumor cell clonality, accounting for variations in the number of malignant cells per tumor.
  • Hierarchical Clustering: Used to construct tumor branching architectures, visualizing evolutionary relationships between tumor clones.
  • RNA Velocity and Reverse Graph Embedding: These computational methods were utilized to infer single-cell trajectories and lineage relationships among malignant cells, providing dynamic insights into tumor evolution.
  • CellphoneDB: Applied to analyze ligand-receptor interactions between malignant cells and T cells, shedding light on cell-cell communication within the tumor microenvironment.
  • Genome-wide Copy Number Variation (CNV) Inference: Used to identify malignant cells and assess genomic similarity within and between tumor samples.
  • Gene Set Enrichment Analysis (GSEA): Employed to identify enriched pathways associated with tumor aggressiveness and T-cell responses.

Datasets

The study utilized single-cell transcriptomic data from 46 primary liver tumors collected from 37 patients enrolled in interventional studies at the NIH Clinical Center. Additionally, bulk transcriptomic data from 488 HCC patients and 277 iCCA patients from independent cohorts (ICGC, TCGA, LCI, and Japan cohorts) were used for validation.

Key Insights and Ideas

  • Tumor Clonality as a Prognostic Indicator: The study found a strong correlation between increased tumor cell clonality and poorer patient prognosis, suggesting that a more diverse tumor cell population is associated with increased aggressiveness.
  • Osteopontin (SPP1) as a Key Driver: SPP1 emerged as a critical molecule in driving tumor branching evolution and reprogramming the tumor microenvironment. Its expression patterns followed hierarchical evolutionary relationships, and it was significantly elevated in more aggressive tumor branches and after treatment in some cases.
  • Immune Microenvironment Polarization: The study revealed a distinct polarization of the T-cell landscape between different tumor branches (BI-A vs. BI-B&C), indicating that tumor evolution can significantly alter the immune response.
  • Treatment-Induced Evolutionary Shifts: Analysis of paired biopsies before and after treatment showed that tumor evolutionary trajectories can change, with some cases exhibiting distinct lineage shifts while others maintained a stable population.

Key Results

  • Single-cell Atlas: The researchers generated a comprehensive single-cell atlas of ~57,000 malignant and non-malignant cells from 46 liver tumors.
  • Branching Evolution: Evidence of tumor cell branching evolution was identified using hierarchical clustering, RNA velocity, and reverse graph embedding.
  • Prognostic Link: A higher number of tumor clones was significantly associated with shorter patient survival.
  • SPP1 Enrichment: SPP1 expression was significantly higher in more aggressive tumor branches (BI-B/C) compared to less aggressive ones (BI-A) and was elevated in tumor cells after treatment in specific cases.
  • T-cell Landscape Polarization: Distinct T-cell subtype compositions and functional profiles were observed between different tumor branches, suggesting immune reprogramming.
  • Validation: Findings were validated using bulk transcriptomic data from independent patient cohorts, reinforcing the association between tumor clonality and prognosis.

Limitations and Future Work

  • Limited Sample Size for Temporal Analysis: While temporal analysis was performed on 4 cases with biopsies before and after treatment, the authors note that this is a limited number. Future work could expand on this by analyzing more patients with longitudinal biopsies to better understand treatment responses.
  • Functional Significance of T-cell Subtypes: The functional significance of certain T-cell subtypes, particularly those expressing HSPA1A, remains unclear and requires further investigation.
  • Experimental Validation of SPP1 Role: While SPP1 was identified as a key player, its functional role in driving tumor cell evolution and reprogramming tumor immunity needs to be tested in experimental model systems. The development of neutralizing antibodies against specific OPN isoforms is suggested as a potential therapeutic strategy.
  • Non-invasive Biomarkers: The authors suggest the discovery and validation of non-invasive imaging biomarkers for clonality density could help pinpoint evolutionary trends earlier, enabling more timely interventions.
  • Precision Medicine Applications: The study highlights the potential for developing classification systems based on tumor and cellular heterogeneity to guide personalized treatment strategies, such as targeted ablation or immunomodulation.

Summary updated

Abstract

SUMMARY Tumor evolution is a key feature of tumorigenesis and plays a pivotal role in driving intratumor heterogeneity, treatment failure and patients’ prognosis. Here we performed single-cell transcriptome profiling of 46 primary liver cancers from 37 patients enrolled for interventional studies. We surveyed the landscape of ~57,000 malignant and non-malignant cells and determined tumor cell clonality by developing a machine learning-based consensus clustering method. We found evidence of tumor cell branching evolution using hierarchical clustering, RNA velocity as well as reverse graph embedding methods. Interestingly, an increasing tumor cell clonality was tightly linked to patients’ prognosis, accompanied by a polarized immune cell landscape. We identified osteopontin as a key player for tumor cell evolution and microenvironmental reprogramming. Our study offers insight into the collective behavior of tumor cell communities in liver cancer as well as potential drivers for tumor evolution in response to therapy.
BibTeX
@misc{ma2020single,
  author = {Lichun Ma and Limin Wang and Ching-Wen Chang and S. Heinrich and Dana A. Dominguez and Marshonna Forgues and Julián Candia and Maria O. Hernandez and Michael C. Kelly and Yongmei Zhao and Bao Tran and Jonathan M. Hernandez and Jeremy L. Davis and David E. Kleiner and Bradford J. Wood and Tim F. Greten and Xin Wei Wang},
  title = {{Single-cell atlas of tumor clonal evolution in liver cancer}},
  year = {2020},
  doi = {10.1101/2020.08.18.254748},
  howpublished = {bioRxiv},
}
Figure