The Biological Basis of Regional ECM Breakdown and Nerve Invasion in Pancreatic Cancer

Background

Pancreatic Ductal Adenocarcinoma (PDAC) is one of the most aggressive types of cancer, typically associated with a highly fibrotic stroma, which can account for about 90% of the tumor mass. Although morphological heterogeneity is not addressed in routine pathological reports, it has prognostic significance, indicating underlying tumor biology. The heterogeneity within PDAC tumors has been revealed at the molecular level through single-cell RNA sequencing (scRNA-seq), showing diversity in gene expression programs, and high expression heterogeneity is associated with poorer overall survival. This suggests that the heterogeneity of tumor cells may allow them to quickly adapt to therapy or select for aggressive and treatment-resistant tumor cells.

However, scRNA-seq does not elucidate the relationship between gene expression programs and morphological patterns, limiting our understanding of the molecular circuits related to morphological heterogeneity. To bridge this knowledge gap, this study aims to map transcriptional and functional variations to different morphological patterns in individual PDAC, revealing their relationship with extracellular matrix (ECM) and nerve invasion.

Source of the Paper

The study was completed by Pierluigi Di Chiaro, Lucia Nacci, Fabiana Arco, and the Mediterranean Italian-German Laboratory (IRCCS), led by Gioacchino Natoli, and the related manuscript was published in the journal Cancer Cell on April 8, 2024.

Research Process

Process

This study used Laser Microdissection (LMD) technology to isolate multiple morphologically distinguishable regions from untreated PDAC samples, each containing approximately 200-500 cells. These regions were then subjected to RNA-seq for gene expression analysis. RNA was collected from distinct morphological region samples (N=102, 1-7 samples per person), with each sample expressing an average of over 9513 genes. Four morphologically distinct gene expression clusters were identified through consensus clustering and Principal Component Analysis (PCA) and corresponded to morphological patterns.

Experiments and Methods

1. Sample Dissection and Analysis: Using Laser Microdissection (LMD) technology, morphologically distinguishable regions were isolated from untreated PDAC samples.

2. RNA Extraction and Sequencing: Gene expression analysis was conducted on the dissected regions through RNA-seq to reveal gene expression patterns.

3. Data Analysis: Unsupervised classification was performed using consensus clustering and PCA to identify four gene expression clusters, which were retrospectively categorized into three groups related to morphological patterns.

4. Extracellular Matrix Analysis: The study focused on the changes in basement membrane and stromal matrix, analyzing the expression of basement membrane components in different morphological forms.

5. Clinical Sample Validation: Machine learning methods were used to construct a Random Forest (RF) classifier and employ Recursive Feature Elimination (RFE) to find the best gene set.

6. Single-cell RNA Sequencing Validation: Existing single-cell RNA sequencing datasets were combined to verify the enrichment of gene signatures of fragmented biotypes, confirming heterogeneity at the single-cell level.

Results

Three Major Types of Morphological and Functional Variations

1. Glandular Variation: Displays classic ductal features and expresses pancreatic and endodermal genes.

2. Transitional Variation: Shows undifferentiated ductal structures, exhibiting a mix of endodermal and myofibroblast-like gene expressions.

3. Differentiation-deficient Variation: Lacks ductal characteristics and basement membrane, expressing neuro-lineage guide genes.

Role of ECM in Tumor Invasion

The study found that transitional variant PDAC cells disrupt local ECM organization, impacting local mechanical stress. Transitional PDAC cells display a myofibroblast-like gene expression program, leading to the formation of thick collagen fibers and causing local fibrosis. Fibrotic tissue also promotes tumor invasiveness and dissemination through various mechanisms, such as active migration along established gradients of fibrin and collagen fibers (durotaxis).

Prognostic Relevance

Using a Random Forest classifier built from LMD-seq sample data, the study effectively categorized patients into different risk groups, finding that high-risk group patients had significantly reduced overall survival. Elastic net regression was used to extract 23 genes from 457 predictive genes, calculating a risk score for each patient, dividing them into high-risk and low-risk groups, showing significant differences in overall survival between the groups.

Single-cell RNA Sequencing Validation Results

Single-cell RNA sequencing data validated the enrichment of the gene expression program in transitional cells, discovering that this cell type shows a significant myofibroblast-like gene signature expression, with the co-existence of high expression of endodermal genes and ECM components.

Conclusions and Significance

By mapping the transcriptional programs of pancreatic cancer cells to morphological patterns, this study revealed three main types of morphological and functional variations and their impact on tumor invasion and prognosis. Particularly, transitional variant cells disrupt local ECM organization, causing fibrosis and promoting tumor invasion. The results indicate that the coexistence and proportion of different morphological biotypes in PDAC hold significant value for clinical assessment of patients with different gene expression characteristics.

Highlights

1. Three Morphological and Functional Variations Coexist in All PDACs: Using an integrated transcriptomic and morphological map, three main morphological variations coexist in all PDACs.

2. Combined Transcriptional and Tissue Characteristics Have High Prognostic Impact: The combination of transcriptional and tissue characteristics has a highly significant impact on predicting patient prognosis.

3. Local ECM Breakdown Affects Tissue Mechanics and Nerve Invasion: The study found that some myofibroblast-like variations disrupt ECM organization, affecting the state and nerve invasion of PDAC cells.

4. PDAC Cells Show Different Levels of Gene Expression Programs Based on Differentiation Status: Single-cell level analysis indicates that different morphological biotype cells exhibit a gradual change in gene expression from endodermal to myofibroblast-like.

Significance and Value

This study provides a valuable resource for translational research, revealing the association between different gene expression programs and morphological changes in tumor cells, and elucidating their roles in local environmental foundation and invasive characteristics. The research data assist in developing AI-driven methods to link histological patterns and gene expression signatures, providing more personalized treatment options for each patient.