Integrative Transcriptomic Profiling of Perineural Invasion (PNI) Signatures in Pancreatic Ductal Adenocarcinoma (PDAC): a Multi-Contrast Bioinformatics Study
Author: Yosia Jose Rasdiva Manurung
Affiliation: Biotechnology Study Program, Faculty of Science and Mathematics, Diponegoro University
PDAC-PNI-Transcriptomics-Analysis/
├── Affiliation/ # Institutional branding banners
├── Dataset/ # Curated expression matrix and metadata from GSE102238
├── Results/
│ ├── Data_Tables/ # Statistical output of DEGs (CSV/Excel tables)
│ └── Plots/ # Visualizations (box plot, density plot, UMAP plot, volcano plots, heatmaps, scatter plot, venn diagram, dot plot, and bar plot)
├── Script/ # Core analytical engine containing end-to-end R scripts for the entire bioinformatics workflow, from raw GEO data to biological interpretation
├── .gitignore # Git exclusion rules for temporary/system files
├── CITATION.cff # Academic citation metadata for the repository and DOI
├── LICENSE # MIT open-source licensing agreement
└── README.md # Comprehensive project documentation and execution guide
Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive malignancies worldwide, characterized by late-stage diagnosis and high therapeutic resistance. A defining hallmark of its progression is perineural invasion (PNI) a process where cancer cells infiltrate the neural network, driving debilitating pain and clinical recurrence.
Rather than simple physical infiltration, PNI represents a complex molecular reprogramming of the tumor microenvironment (TME) (Chen et al., 2023; Sarantis et al., 2020; Sun et al., 2024).
This study utilizes the GSE102238 dataset (Yang et al., 2020) to systematically map the bidirectional signaling loop between malignant cells and the peripheral nervous system, aiming to identify unique transcriptomic signatures that could serve as novel therapeutic targets.
The primary goal of this study is to identify Differentially Expressed Genes (DEGs) across 6 PNI-related clinical contrasts. This systematic approach allows me to isolate the specific transcriptomic signatures driven by neural invasion versus those driven by general tumorigenesis.
- PNI Effect (Tumor): PNI-positive Tumor vs. PNI-negative Tumor
- PNI Effect (Normal): PNI-positive Normal vs. PNI-negative Normal
- Tumor vs. Normal (+): PNI-positive Tumor vs. PNI-positive Normal
- Tumor vs. Normal (-): PNI-negative Tumor vs. PNI-negative Normal
- Extreme Contrast: PNI-positive Tumor vs. PNI-negative Normal
- Reverse Contrast: PNI-negative Tumor vs. PNI-positive Normal
The analysis was conducted using R (v4.5.2). The integrated pipeline combines data acquisition, rigorous preprocessing, and functional interpretation as follows:
-
Data Acquisition: Retrieval of raw data via
GEOquery. -
Data Preprocessing: Conditional
$\log_2$ transformation usinglog2()and quantile distribution evaluation viaquantile()to stabilize expression signal ranges. -
Data Analysis: Modeled using the
limmapackage across 6 distinct clinical contrasts. -
Data Annotation: Systematic mapping of probes to HGNC Symbols via
biomaRtand relational merging with platform metadata. -
Data Visualization: Generation of high-fidelity plots to assess data distribution and DEGs significance:
-
Box Plot:
ggplot()withstat_boxplot()andgeom_boxplot()to evaluate cross-sample intensity distribution. -
Density Plot:
ggplot()withgeom_density()to inspect overall expression profile shapes and normalization symmetry. -
UMAP Plot:
umap()algorithm followed bygeom_point()to visualize 2D sample clustering. -
Volcano Plots: Custom function
make_volcano()mapping$\log_2\text{FC}$ vs.$-\log_{10}(\text{Adjusted } P\text{-value})$ across 6 distinct clinical contrasts (v1–v6). -
Heatmaps:
pheatmap()with Ward.D2 hierarchical clustering for top 50 DEGs, featuring 1 Global ANOVA overview across 100 samples (H0) and 6 contrast-specific subsets (H1–H6). -
Scatter Plot:
ggplot()withgeom_smooth(method = "gam")to profile gene expression stability (Mean vs. SD) across clinical cohorts. -
Venn Diagram:
ggVennDiagram()with a 6-set elliptical layout (shape_id = "601") to identify core biomarkers across all clinical contrasts.
-
Box Plot:
-
Data Interpretation: Functional enrichment analysis and visual mapping of biological mechanisms:
-
ID Conversion:
bitr()fromclusterProfilerusingorg.Hs.eg.dbto map HGNC Symbols to Entrez IDs. -
Gene Ontology (GO) Analysis:
enrichGO()focusing on Biological Processes (ont = "BP") evaluated withdotplot()for top 15 enriched terms. -
Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway Analysis:
enrichKEGG()for Homo sapiens (organism = 'hsa') visualized viabarplot()based on gene count.
-
ID Conversion:
Below is the visual representation of the analytical steps performed in this project:
graph TD
%% Node Definitions with Inline Padding for Uniform Box Sizes
S1[" 1. Data Acquisition "]
S2[" 2. Data Preprocessing "]
S3[" 3. Data Analysis "]
S4[" 4. Data Annotation "]
S5[" 5. Data Visualization "]
S6[" 6. Data Interpretation "]
%% Sequential Flow Connections
S1 --> S2
S2 --> S3
S3 --> S4
S4 --> S5
S5 --> S6
%% Color Palette and Styling Definitions
style S1 fill:#3f72af,color:#fff,stroke:#112d4e,stroke-width:2px
style S2 fill:#00adb5,color:#fff,stroke:#393e46,stroke-width:2px
style S3 fill:#ff5722,color:#fff,stroke:#b23b00,stroke-width:2px
style S4 fill:#9c27b0,color:#fff,stroke:#4a148c,stroke-width:2px
style S5 fill:#e91e63,color:#fff,stroke:#880e4f,stroke-width:2px
style S6 fill:#ff9800,color:#fff,stroke:#e65100,stroke-width:2px
To illustrate the extreme variance in gene expression, this analysis compares highly stable transcriptomic profiles against those undergoing massive dysregulation ("eruption"):
| Condition: Stability (Normal Tissue) | Condition: Eruption (Extreme Contrast) |
|---|---|
 |
 |
| Stability Example: The PNI effect within normal tissues shows minimal differential expression. | Eruption Example: The Extreme Contrast reveals massive gene activation and suppression. |
- Key Insight: PDAC maintains high transcriptomic stability at the tissue level, particularly within normal cohorts. However, it undergoes a massive "expression eruption" when transitioning from a basal normal state to a malignant, nerve-involved (PNI-positive) state.
- Biological Significance: This suggests that while Neural Invasion is a critical clinical marker, the most profound molecular shifts are driven by the synergy between malignancy and the perineural environment.
The heatmaps illustrate the contrast between homeostatic stability and significant clinical divergence across the 100-sample cohort:
| Condition: Stability (Normal Tissue) | Condition: Eruption (Extreme Contrast) |
|---|---|
 |
 |
| H2: Minimal expression variance in normal tissues regardless of PNI status. | H5: Distinct bifurcated expression patterns in PNI-Pos Tumor vs PNI-Neg Normal. |
- Clustering Insight: Hierarchical clustering in H2 confirms that PNI status does not disrupt the basal transcriptomic state of normal tissues.
- Malignancy Signature: H5 demonstrates a clear molecular signature that separates aggressive PNI-positive tumors from healthy controls, highlighting the "eruption" of differentially expressed genes.
To isolate the core genetic drivers, I intersected multiple clinical contrasts and verified expression stability:
| Core Biomarker Intersection | Expression Stability Profile |
|---|---|
 |
 |
| Venn diagram identifying 9,750 unique DEGs across all 6 clinical contrasts. | Mean vs SD Scatter plot visualizing gene stability with GAM smoothing. |
- Robust Intersection: The 6-set Venn diagram allows for the identification of consistently dysregulated genes across all clinical scenarios.
- High-Confidence Biomarkers: Key upregulated genes identified through this pipeline include CEACAM5, S100P, CST2, and TMPRSS4.
- Stability Verification: The scatter plot confirms that while most genes remain stable (low SD), a subset of high-variance genes drives the clinical differences observed in PDAC.
The biological roles of the core biomarkers were analyzed to link gene expression to clinical phenotypes:
| Gene Ontology (GO) | Kyoto Encyclopedia of Genes and Genomes (KEGG) |
|---|---|
 |
 |
| Dot plot highlighting Leukocyte Adhesion and T-cell Activation. | Bar plot showcasing IgSF CAM signaling and Axon Guidance. |
- Immune Response & Adhesion: Significant enrichment in Leukocyte cell-cell adhesion and T-cell activation suggests a strong immune-modulatory component in the PDAC microenvironment.
- Neural & Signaling Links: KEGG analysis identifies Axon Guidance and IgSF CAM signaling as key pathways, providing a molecular basis for how tumor cells interact with neural structures during PNI.
This study provides a high-resolution transcriptomic map of PNI in PDAC. By systematically dissecting six clinicopathological contrasts, I have established several key conclusions:
- Malignancy Overrides Localization: The transcriptomic landscape is dominated by a robust malignant signal that remains consistent regardless of localized neural involvement. The primary oncogenic "engine" of PDAC is the main driver of the observed mRNA profiles.
- The "Eruption" Signature: While PNI-only contrasts show high homeostatic stability, the transition from normal to malignant PNI-positive states triggers a massive molecular "eruption." This is evidenced by a core consensus signature of 4,857 shared DEGs, including high-confidence biomarkers such as CEACAM5, S100P, CST2, and TMPRSS4.
- Molecular Hijacking: Functional enrichment confirms that tumor cells do not move randomly; they actively exploit Axon Guidance and IgSF CAM signaling to infiltrate the peripheral nervous system.
- Immune-Adhesion Crosstalk: The convergence of Leukocyte cell-cell adhesion and T-cell activation pathways suggests that PNI is an immune-active process, characterized by complex bidirectional crosstalk between malignant cells and the inflammatory TME.
This research establishes a computational foundation for discovering novel diagnostic markers and therapeutic targets. By identifying the molecular pillars driving both PDAC progression and neural recruitment, these findings offer a roadmap for future studies aimed at disrupting the pathways that drive clinical recurrence and patient morbidity.
Chen, Z., Fang, Y., & Jiang, W. (2023). Important Cells and Factors from Tumor Microenvironment Participated in Perineural Invasion. Cancers, 15(5), 1360. https://doi.org/10.3390/cancers15051360.
Sarantis, P., Koustas, E., Papadimitropoulou, A., Papavassiliou, A. G., & Karamouzis, M. V. (2020). Pancreatic ductal adenocarcinoma: Treatment hurdles, tumor microenvironment and immunotherapy. World journal of gastrointestinal oncology, 12(2), 173–181. https://doi.org/10.4251/wjgo.v12.i2.173.
Sun, Y., Jiang, W., Liao, X., & Wang, D. (2024). Hallmarks of perineural invasion in pancreatic ductal adenocarcinoma: new biological dimensions. Frontiers in Oncology, 14, 1421067. Sec. https://doi.org/10.3389/fonc.2024.1421067.
Yang, M. W., Tao, L. Y., Jiang, Y. S., Yang, J. Y., Huo, Y. M., Liu, D. J., ... & Sun, Y. W. (2020). Perineural invasion reprograms the immune microenvironment through cholinergic signaling in pancreatic ductal adenocarcinoma. Cancer research, 80(10), 1991-2003. https://doi.org/10.1158/0008-5472.CAN-19-2689.
If you use this repository, datasets, or analytical pipelines in your academic research, please cite it using the metadata provided below:
Manurung, Y. J. R. (2026). Integrative Transcriptomic Profiling of Perineural Invasion (PNI) Signatures in Pancreatic Ductal Adenocarcinoma (PDAC): a Multi-Contrast Bioinformatics Study (Version 1.0.3) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.21870889.
© 2026 Yosia Jose Rasdiva Manurung. All Rights Reserved.