Self-assembled primary tumor clusters for precision cancer therapy

Shenyi Yin, Ruibin Xi, Aiwen Wu, Jiafu Ji, Jianzhong XibioRxiv 2019 · DOI

TL;DR

This paper introduces a novel "patient-derived tumor-like cell clusters" (PTC) model that can be generated from patient tumor samples within 10-20 days, enabling rapid personalized drug testing. The PTC model accurately recapitulates the original tumor's structure and genetics, offering a promising new tool to guide precision cancer therapy.

What Problem Does This Paper Address?

Precision oncology aims to tailor cancer treatments to individual patients based on their unique tumor characteristics. While various patient-derived models exist (like xenografts and organoids), they often take too long to generate (months) for timely clinical decision-making, especially for aggressive cancers like gastric and colorectal. Furthermore, some models have limitations in fully recapitulating the tumor microenvironment or are difficult to establish from certain tumor types. This gap leaves a critical need for a preclinical model that is fast, accurate, and versatile for personalized drug testing.

Methods

The researchers developed a new in vitro model called patient-derived tumor-like cell clusters (PTC). Unlike traditional 3D cultures that rely on matrices like Matrigel, PTCs are formed through the self-assembly and migration of dissociated primary tumor cells, including epithelial cells, fibroblasts, and macrophages. This approach was optimized to maintain the integrity of these diverse cell types and generate sufficient material for drug testing within two weeks. They characterized the PTCs using immunofluorescence and flow cytometry to confirm their cellular composition and compared their genomic profiles (somatic mutations and copy number variations) to the original patient tumors using target sequencing and low-pass whole genome sequencing. For drug testing, PTCs were cultured in multi-well chips, and drug efficacy was assessed by measuring changes in cluster area over seven days, using a formula adapted from RECIST criteria for solid tumor response.

Datasets

The study utilized 190 fresh gastric or colorectal tumor samples, including 153 surgically resected tissues, 31 endoscopic biopsies, and 6 ascites samples, for establishing PTC models. Genomic data (somatic mutations and copy number variations) were generated from these PTCs and their matched original tumor samples.

Key Insights and Ideas

  • Rapid Generation: The PTC model can be established and ready for drug testing within 10-20 days, a significant improvement over existing models that take months.
  • Self-Assembly is Key: The formation of PTCs relies on the natural migration and self-assembly of tumor cells, including stromal components, which is crucial for recapitulating the tumor microenvironment.
  • Comprehensive Recapitulation: PTCs not only mimic the morphology of original tumors but also maintain their genomic fidelity, showing high concordance in somatic mutations and copy number variations.
  • Versatility: The PTC platform is adaptable to a wide range of cancer subtypes, including those historically difficult to model with organoids, such as poorly differentiated adenocarcinomas and metastatic samples.
  • Stromal Cell Importance: The inclusion of stromal cells (macrophages and fibroblasts) in PTCs is highlighted as essential for assembling the clusters and accurately recapitulating drug responses, a feature often lacking in epithelial-only organoid models.

Key Results

  • High Establishment Rate: An overall success rate of 74.7% (142/190) was achieved for PTC establishment. Interestingly, biopsy samples had a higher success rate (83.9%) than surgically resected samples (72.5%).
  • Genomic Concordance: PTCs showed high genomic similarity to original tumors, with mean concordance for somatic mutations at 0.86 (s.d. 0.13) and for CNVs at 0.78 (s.d. 0.13).
  • Drug Efficacy Consistency: In drug efficacy tests, 91.74% of CVs were less than 0.5, indicating high consistency across repeated experiments for the same drug-sample pair. Pearson correlation coefficients between experiments were consistently high (mean 0.84, s.d. 0.036).
  • Genomic Consistency Across Wells: Replicated PTCs from the same sample also showed high genomic consistency, with pairwise overlap for somatic mutations greater than 0.8 and CNV correlations averaging 0.91.

Limitations and Future Work

The paper acknowledges that while PTCs show high concordance, minor differences in genomic alterations can occur, potentially due to detection power limitations. The authors suggest that the PTC platform's seamless integration into neoadjuvant, chemotherapy, or target therapy programs could revolutionize precision oncology. Future work could focus on further validating the predictive power of PTCs across a broader range of cancer types and therapeutic agents, and potentially standardizing the protocol for wider clinical adoption.

Summary updated

Abstract

Abstract Several patient-derived tumor models emerged recently as robust preclinical models. However, their potential to guide clinical therapy remained unclear. We report a novel model for personalized drug testing called patient-derived tumor-like cell clusters (PTC) that enabled us to accomplish personalized drug tests within 10-20 days. Mechanistically, PTCs result from the migration and aggregation of CD8+/CD44+ primary epithelial cells in a mixture of fibroblasts and macrophages, and form spheroid cultures. Phenotypic and genotypic profiling of PTCs showed a high degree of similarity with the original patient tumors, indicating that PTC structurally and functionally recapitulated original tumors. Over 200 PTC models have been established from at least six cancer types and diverse sampling approaches. Given its seamless integration with current clinical sampling approaches, the PTC platform will revolutionize chemotherapy programs for cancer patients.
BibTeX
@misc{yin2019self,
  author = {Shenyi Yin and Ruibin Xi and Aiwen Wu and Jiafu Ji and Jianzhong Xi},
  title = {{Self-assembled primary tumor clusters for precision cancer therapy}},
  year = {2019},
  doi = {10.1101/700724},
  howpublished = {bioRxiv},
}

Loading related content...

Figure