Patient-specific modeling of stroma-mediated chemoresistance of pancreatic cancer using a three-dimensional organoid-fibroblast co-culture system

Sebastian Schuth, Solange Le Blanc, Teresa G. Krieger, Julia Jabs, Miriam Schenk, Nathalia A. Giese, Markus W. Büchler, Roland Eils, Christian Conrad, Oliver StrobelJournal of Experimental & Clinical Cancer Research 2022 · DOI

TL;DR

Researchers have developed a personalized 3D co-culture system that pairs patient-derived pancreatic cancer organoids with their own matching "helper" cells (fibroblasts). This model reveals that these fibroblasts don't just provide structural support; they actively shield cancer cells from chemotherapy by triggering a molecular "identity shift" known as epithelial-to-mesenchymal transition (EMT).

What Problem Does This Paper Address?

Pancreatic ductal adenocarcinoma (PDAC) is notoriously difficult to treat, largely due to extreme chemoresistance. A major reason for this is the tumor microenvironment: up to 90% of a pancreatic tumor's volume isn't actually cancer cells, but a dense "stroma" made of extracellular matrix and cancer-associated fibroblasts (CAFs).

While patient-derived organoids (PDOs)—tiny 3D clusters of tumor cells grown in a dish—have become popular "avatars" for testing drugs, most models ignore the stroma. By only testing the epithelial (cancer) cells, researchers miss the critical protective signals provided by the CAFs. This paper addresses the need for a more "complete" personalized model that accounts for tumor-stroma crosstalk to better predict how a specific patient will respond to therapy.

Methods

The authors established a sophisticated 3D co-culture pipeline:

  • Patient-Matched Pairing: They isolated both primary tumor organoids and CAFs from the same surgical specimens of five patients, ensuring the biological interactions were patient-specific.
  • 3D Environment: Cells were embedded in a mixture of Matrigel and Collagen I to mimic the physical density of a real tumor.
  • DeathPro Assay: To solve the problem of "who is dying?" in a mixed culture, they used an image-based drug assay called DeathPro. By labeling CAFs with a fluorescent cell tracker and using automated confocal microscopy, they could independently measure cell death and proliferation for the cancer cells versus the fibroblasts.
  • Single-Cell RNA Sequencing (scRNA-seq): They sequenced over 20,000 individual cells from mono-cultures and co-cultures to map the transcriptional "conversations" happening between the two cell types.

Datasets

The study primarily generated and analyzed its own primary datasets:

  • Primary PDO/CAF Lines: Five matched pairs of patient-derived lines.
  • scRNA-seq Data: Transcriptomes of 13,235 PDAC cells and 7,356 CAFs.
  • Reference Benchmarks: The researchers used the MSigDB database for gene set enrichment and the PurIST classifier to categorize tumor subtypes (Basal-like vs. Classical).

Key Insights and Ideas

  • CAFs as Bodyguards: The presence of CAFs significantly reduced the effectiveness of gemcitabine, 5-fluorouracil (5-FU), and paclitaxel. Even when the drugs successfully killed CAFs, the cancer cells remained more resistant than they were in isolation.
  • The Pro-Inflammatory Shift: When co-cultured with tumor cells, CAFs shifted toward an "inflammatory" (iCAF) phenotype, characterized by increased signaling of pathways like TNF$\alpha$ and TGF$\beta$.
  • Inducing EMT: The most striking finding was that CAFs induced Epithelial-to-Mesenchymal Transition (EMT) in the organoids. The cancer cells began expressing genes associated with migration and invasion (like VIM, FN1, and COL1A1), a state known to be highly resistant to conventional chemotherapy.
  • The CD44-HGF Axis: Using ligand-receptor interaction analysis (CellPhoneDB), the team identified specific "handshakes" between CAFs and tumor cells. They highlighted the interaction between the CAF-secreted ligand HGF and the tumor receptor CD44 as a key driver of this resistance, which they further validated in original patient tissue sections.

Key Results

  • Increased Proliferation: CAFs boosted organoid growth by an average of 1.3-fold in control conditions.
  • Quantifiable Resistance: Co-culture led to a significant decrease in the Area Under the Curve (AUC) for cell death across all three tested chemotherapies. For example, gemcitabine-induced death was consistently lower when CAFs were present.
  • Transcriptional Reprogramming: scRNA-seq confirmed that the "EMT Hallmark" was the most significantly enriched gene set in co-cultured organoids.
  • Validation: Immunofluorescence confirmed that CD44 and HGF co-localize in the actual patient tumors, suggesting the lab model faithfully replicates the in vivo environment.

Limitations and Future Work

The primary hurdle for clinical use is time. It took a median of 40 days to expand enough pure biomass for drug testing. Since adjuvant chemotherapy usually starts 6–12 weeks after surgery, this timeline is tight, though potentially feasible for selecting second-line treatments.

Additionally, the success rate for establishing these organoid lines was 50%, which needs improvement for universal clinical application. The authors also note that while this model adds fibroblasts, it still lacks other critical components like immune cells and blood vessels. Future iterations of this "mini-tumor" model will likely aim to incorporate these elements to further refine drug response predictions.

Summary updated

Abstract

Our results demonstrate the potential of personalized PDAC co-cultures models not only for drug response profiling but also for unraveling the molecular mechanisms involved in the chemoresistance-supporting role of the tumor stroma.
BibTeX
@article{schuth2022patient,
  author = {Sebastian Schuth and Solange Le Blanc and Teresa G. Krieger and Julia Jabs and Miriam Schenk and Nathalia A. Giese and Markus W. Büchler and Roland Eils and Christian Conrad and Oliver Strobel},
  title = {{Patient-specific modeling of stroma-mediated chemoresistance of pancreatic cancer using a three-dimensional organoid-fibroblast co-culture system}},
  journal = {Journal of Experimental \& Clinical Cancer Research},
  year = {2022},
  doi = {10.1186/s13046-022-02519-7},
}
Figure