Pancreatic ductal adenocarcinoma (PDAC) is defined by spatially organized biology—neural invasion niches, desmoplastic stroma barriers, immune exclusion, therapy-driven remodeling, and organ-specific metastatic ecosystems. This curated literature roundup highlights landmark spatial (multi-)omics studies and translates each paper into a copy-ready blueprint you can adapt for your own cohort.

What you'll get

• A fast-scan summary of each landmark paper: cohort, platforms, and validation stack
• Key spatial findings distilled into actionable bullets (what changed, where, and in which cell states)
• A practical "study blueprint" per paper: design, minimum assays, core readouts, and pitfalls to avoid

For research use only. This page summarizes published studies and is not medical advice.

Why spatial multi-omics matters in PDAC

Many PDAC mechanisms are location dependent. A tumor's invasive edge, nerve-adjacent interface, fibroblast-rich barrier, immune-permissive pockets, and metastatic organ niches can show distinct programs—even when bulk averages look similar. Spatial omics adds "where" to "what," enabling defensible hypotheses about microenvironmental control of invasion, immune exclusion, and therapy resistance.

If you're new to cancer-oriented spatial study design, start with this Resource article:

Spatial Transcriptomics Cancer Research: 4 Core Strategies

A practical platform map for PDAC studies

Most successful PDAC spatial projects follow a "discovery → resolution → validation" stack. The goal is not to run every assay, but to choose a minimal set that can (1) map tissue architecture, (2) resolve key niches and neighbors, and (3) validate one or two mechanisms credibly.

Recommended assay stack (common and defensible)

• Discovery (architecture + broad programs): spot-based whole-transcriptome spatial profiling (e.g., Visium-class approaches)
• Cell-resolved confirmation (niches + neighbors): high-plex imaging readouts (e.g., CosMx/Xenium/GeoMx-class, depending on aim)
• Reference atlas: scRNA-seq or snRNA-seq to define cell states and enable deconvolution/cell-type mapping
• Orthogonal validation: IF/IHC/RNA-ISH ± targeted functional assays for top hypotheses only

Method-selection guides (Resource hub):

• Need help selecting the right platform mix for your cohort and readouts? Use: Guide to Spatial Transcriptomics Platforms: Sequencing vs Imaging and How to Choose
• For an experimental-method comparison (Visium vs DBiT-seq vs Slide-seq), see: Spatial Transcriptomics vs DBiT-seq vs Slide-seq Comparison Guide
• If you want a quick decision refresher for tissue studies: Bulk RNA-seq vs Single-Cell RNA-seq vs Spatial Transcriptomics: How to Choose

Landmark papers (6 study cards)

Paper 1 — Neural invasion niches mapped by integrated single-cell and spatial transcriptomics

Integrated single-cell and spatial transcriptomics uncover distinct cellular subtypes involved in neural invasion in pancreatic cancer.

Cancer Cell(2025). DOI: 10.1016/j.ccell.2025.06.020

Why it matters

Neural invasion (NI) is spatial by definition. This study shows how tumor, immune, stromal, and Schwann-cell programs organize around invaded vs non-invaded nerves—and why nerve-adjacent niches can drive aggressive behavior.

Cohort & samples

• 62 samples from 25 PDAC patients (tumor, adjacent tissue, peripheral blood)
• Stratified into NI-low vs NI-high states based on nerve invasion and microenvironment features

Tech stack

• sc/snRNA-seq + spatial transcriptomics
• Validation: immunofluorescence, survival association, in vitro migration assays

Key spatial findings

• NI-front Schwann program: a TGFBI+ Schwann cell subtype localizes at the neural invasion front and aligns with pro-migratory signaling.
• Immune–nerve spatial contrast: NI-low tumors show tertiary lymphoid structures (TLS) co-localizing with non-invaded nerves; NI-high tumors show inflammatory/stromal enrichment around invaded nerves.
• Myeloid + myCAF "ring" around invaded nerves: NLRP3+ macrophages and myCAFs are spatially enriched around invaded nerves in NI-high regions.
• Malignant-state geography: basal-like and "neural-reactive" malignant subpopulations show distinct morphologies and enhanced NI-associated potential.

Figure 2. Graphical abstract from Chen et al., Cancer Cell (2025): spatial single-cell integration maps nerve-adjacent niches and identifies TGFBI+ Schwann cells linked to neural invasion.

Study blueprint (what to replicate)

• Best-fit design: recruit PDAC cases spanning NI-low and NI-high; predefine nerve-adjacent ROIs plus matched non-nerve control ROIs.
• Minimum assays: (1) sc/snRNA-seq reference atlas, (2) whole-transcriptome spatial profiling for architecture, (3) IF/IHC panel for Schwann, macrophage, CAF, and malignant state markers.
• Core readouts: nerve-adjacent niche calling; neighborhood enrichment around invaded vs non-invaded nerves; spatially constrained ligand–receptor prioritization; survival association for NI-linked markers.
• Common pitfalls: inconsistent nerve annotation; ROI selection bias; attributing "nerve-associated" programs without matched non-nerve controls.

Internal method support (recommended Resource link): Spatial Transcriptomics Data Analysis: A Practical Introduction

Paper 2 — Spatial drivers of untreated vs chemo-resistant PDAC (transitional populations + microenvironment)

Spatially restricted drivers and transitional cell populations cooperate with the microenvironment in untreated and chemo-resistant pancreatic cancer.

Nature Genetics (2022). DOI: 10.1038/s41588-022-01157-1

Why it matters

Chemo-resistance is not only which cells exist, but where resistant programs emerge and which neighboring communities support them. This study integrates genomic events, spatial architecture, and microenvironment remodeling.

Cohort & samples

• 83 spatial samples from 31 patients
• 11 untreated vs 20 chemotherapy-exposed (chemo-resistant context)

Tech stack

• sc/snRNA-seq + proteomics + spatial transcriptomics + tissue imaging
• Augmentation/validation: IHC, RNA-seq, WES

Key spatial findings

• Spatially restricted drivers: mapping mutations/CNV patterns onto tissue helps separate malignant, normal, and transitional populations.
• Transitional geography: ADM and PanIN-like states occupy recognizable spatial contexts, supporting continuum models of progression.
• Therapy-linked stroma shift: chemo-resistant samples show strong enrichment of inflammatory CAF programs (iCAF) alongside metallothionein upregulation.
• Immune checkpoint spatial motifs: coordinated TIGIT-associated exhausted/regulatory T cell patterns align with tumor-side checkpoint ligands, suggesting spatial immune-evasion arrangements.

Study blueprint (what to replicate)

• Best-fit design: untreated vs post-therapy PDAC sections with matched pathology; include transitional lesions (ADM/PanIN fields) when available.
• Minimum assays: spatial transcriptomics plus basic genomic profiling (targeted DNA or WES on matched tissue) plus CAF/immune validation (IHC/IF).
• Core readouts: spatial CNV mapping; ecotype/domain detection; therapy-associated CAF program scoring; immune checkpoint spatial co-localization analyses.
• Common pitfalls: mixing treatment regimens without stratification; weak pathology alignment to spatial domains; batch effects between untreated and treated datasets.

Internal method support (recommended Resource link)

Guide to Spatial Transcriptomics Platforms: Sequencing vs Imaging and How to Choose

Paper 3 — Spatial mapping of transcriptomic plasticity in metastatic PDAC (organ-specific ecosystems)

Spatial mapping of transcriptomic plasticity in metastatic pancreatic cancer.

Nature (2025). DOI: 10.1038/s41586-025-08927-x

Why it matters

Metastasis is an ecosystem shift. This work uses spatial profiling to connect lineage-state plasticity, clonal architecture, and organ-specific microenvironments—key for designing metastasis cohorts and analyses that don't overgeneralize.

Cohort & samples

• 55 tumor samples from 13 PDAC patients
• Primary tumors plus matched metastases (liver, lung, peritoneum)

Tech stack

• Whole-transcriptome spatial profiling for architecture
• Cell-resolved high-plex imaging for niche validation (reported platforms include Visium-class discovery + CosMx-class validation)
• Validation: mouse models and immunofluorescence

Key spatial findings

• Organ-specific malignant geography: cancer lineage states shift with metastatic site; lineage phenotypes are not strictly tied to clonal origin.
• No single metastasis blueprint: clonal evolution patterns vary by patient; key metastatic clones can appear early in primary tumors.
• Microenvironment-linked state control: local stromal/immune context strongly associates with malignant phenotype; organ niches bias lineage-state distributions.
• Immune exclusion motif: myCAFs co-localize with basal-like malignant programs and align with immune remodeling consistent with CXCL12/CXCR4-type exclusion patterns.

Study blueprint (what to replicate)

• Best-fit design: matched primary plus multi-organ metastases where feasible; at minimum primary + one metastatic site with standardized processing and clinical annotation.
• Minimum assays: spatial transcriptomics for ecotypes + targeted imaging for cell-resolved neighborhood validation in key niches (CAF–immune, tumor–immune).
• Core readouts: organ-stratified domain/ecotype detection; malignant-state scoring by niche; CAF–immune proximity metrics; patient-level heterogeneity reporting (per-patient plots).
• Common pitfalls: overgeneralizing across patients; underpowering organ groups; ignoring within-organ lesion heterogeneity.

Internal method support (recommended Resource link): Frontiers in Spatial Multi-Omics Integration: From Data to Mechanism

Paper 4 — Neoadjuvant treatment response decoded by single-nucleus + spatial transcriptomics

Single-nucleus and spatial transcriptome profiling of pancreatic cancer identifies multicellular dynamics associated with neoadjuvant treatment.

Nature Genetics (2022). DOI: 10.1038/s41588-022-01134-8

Why it matters

NAT response in PDAC is heterogeneous. This paper shows how multicellular communities differ in space between responders and non-responders, shifting the focus from single biomarkers to spatial community structure.

Cohort & samples

• 56 clinically annotated cases from 37 patients (NAT and untreated)
• 113 samples for snRNA-seq (~400k nuclei) and 24 tumor samples for Visium spatial profiling
• NAT responders vs non-responders; paired baseline/post-treatment specimens in a subset

Tech stack

• snRNA-seq + spatial transcriptomics (Visium-class)
• Augmentation: TCR/BCR, WES, IHC/mIF, flow cytometry, co-culture/Transwell, PDX models

Key spatial findings

• Community structure: integrated single-nucleus and spatial data reveal multicellular communities tied to malignant, fibroblast, and immune compositions.
• Responder geometry: responders show immune activation patterns with stronger tumor–CD8 proximity signatures (immune "contact" microenvironments).
• Non-responder barrier: fibroblast-rich shielding neighborhoods correlate with immune exclusion; CXCL12-high fibroblast programs are implicated in barrier formation.
• Predictive scoring: community scoring frameworks align with response and can be carried to validation cohorts.

Figure 3. Hwang et al., Nature Genetics (2022), Fig. 7: spatial community structure and treatment-associated cell–cell communication in PDAC after neoadjuvant therapy.

Study blueprint (what to replicate)

• Best-fit design: NAT baseline and resection pairs; stratify by response using standardized clinical endpoints (pathologic regression, imaging, or curated response criteria).
• Minimum assays: snRNA-seq reference + spatial transcriptomics on matched sections; mIF/IHC for immune contact vs barrier confirmation.
• Core readouts: community detection; tumor–immune distance metrics; CAF program scoring; "contact vs barrier" spatial phenotypes at the patient level.
• Common pitfalls: inconsistent response definitions; ROI selection that misses invasive border; comparing unpaired samples without patient-level adjustment.

Internal method support (recommended Resource link): Spatial Transcriptomics Data Analysis: A Practical Introduction

Paper 5 — Primary vs metastatic PDAC ecotypes highlight microenvironmental heterogeneity

Spatial transcriptomic analysis of primary and metastatic pancreatic cancers highlights tumor microenvironmental heterogeneity.

Nature Genetics (2024). DOI: 10.1038/s41588-024-01914-4

Why it matters

"Metastatic PDAC" is not one TME. This work emphasizes spatial ecotypes that vary between primary and metastatic sites and provides a framework for ecotype-aware cohort design and reporting.

Cohort & samples

• Clinically annotated primary and metastatic PDAC tissues profiled with spatial transcriptomics
• Designed to compare microenvironmental ecotypes across sites and within tumor regions (core vs border)

Tech stack

• Spot-based spatial transcriptomics + cell-type/ecotype enrichment modeling
• Validation: orthogonal assays as reported; for replication, plan a compact marker panel aligned to your ecotypes

Key spatial findings

• Ecotype continuum: spatial ecotypes span fibrotic, metabolic, inflammatory, and immunosuppressive programs beyond bulk averages.
• Site enrichment: ecotypes show distinct enrichment between primary and metastatic sites, consistent with local adaptation.
• Border vs core heterogeneity: tumor edges can mix immune-active and immune-suppressive signals, making "edge-aware" stratification essential.
• Heterogeneity as a measurable axis: TME diversity becomes a reportable dimension for comparing cohorts.

Figure 4. Khaliq et al., Nature Genetics (2024), Fig. 5: invasive ecotypes and immune enrichment patterns across primary and metastatic PDAC.

Study blueprint (what to replicate)

• Best-fit design: matched primary vs metastasis sampling with standardized processing; predefine border vs core ROIs.
• Minimum assays: spatial transcriptomics + compact validation panel (immune + CAF + endothelial markers) to confirm ecotype interpretations.
• Core readouts: ecotype/domain calling; site enrichment stats; border vs core comparisons; heterogeneity metrics per patient.
• Common pitfalls: merging ecotypes across batches without calibration; under-sampling borders; reporting ecotypes without validation markers.

Internal method support (recommended Resource link)

Computational Strategies and Machine Learning for Spatial Genomics Data

Paper 6 — PanIN immune organization revealed by spatial proteomics + transcriptomics

Spatial proteomics and transcriptomics reveal early immune cell organization in pancreatic intraepithelial neoplasia.

JCI Insight (2025). DOI: 10.1172/jci.insight.191595

Why it matters

If you study early lesions (PanIN) or interception biology, spatial immune organization is the point—not just immune abundance. This paper uses spatial proteomics and transcriptomics to characterize how immune architecture emerges in premalignant niches.

Cohort & samples

• Treatment-naïve resected specimens with regions containing normal tissue and PanIN lesions (deep spatial profiling)
• Corroboration in inducible mouse models reported in the study

Tech stack

• Spatial proteomics + spatial transcriptomics with emphasis on immune architecture and neighborhood organization
• Corroboration in mouse models and spatial profiling validation

Key spatial findings

• Early lymphoid organization: lymphoid cells form immature TLS-like structures adjacent to PanIN lesions.
• TLS maturation differences: TLS-like structures near PanIN differ from more mature TLS patterns near PDAC borders, suggesting progression-linked architecture shifts.
• PanIN vs PDAC immune composition: PanIN regions show distinct T cell state patterns and fewer advanced immunosuppressive/exhausted features than PDAC regions.
• Architecture as a mechanistic readout: spatial arrangement (not only cell counts) captures early immune reprogramming relevant to interception hypotheses.

Figure 5. Lyman et al., JCI Insight (2025), Fig. 1: spatial immune organization around PanIN lesions, including TLS-like aggregates adjacent to premalignant regions.

Study blueprint (what to replicate)

• Best-fit design: include sections containing PanIN plus adjacent non-lesion tissue; if possible include PDAC-border regions to compare TLS maturation states.
• Minimum assays: spatial proteomics (immune marker coverage) + spatial transcriptomics (program confirmation) + pathology annotation of PanIN grade and lesion boundaries.
• Core readouts: TLS detection/typing; immune neighborhood graphs; PanIN-adjacent vs PDAC-border comparisons; marker localization maps.
• Common pitfalls: PanIN region mislabeling; insufficient immune marker coverage; equating "immune-rich" with TLS without confirming architecture.

Internal method support (recommended Resource link): Databases and Resources: How to Find Spatial Omics Datasets and Protocols

Cross-paper insights you can reuse (PDAC-specific)

Across these studies, several patterns repeat—and can be treated as reusable hypotheses rather than one-off observations:

• Neural invasion is a niche, not a label. Nerve-adjacent interfaces host distinct Schwann, myeloid, CAF, and malignant programs that are easy to miss in bulk averages.
• CAFs shape geometry. Barrier-like fibroblast neighborhoods often align with immune exclusion and therapy resistance—treat distance and adjacency as primary endpoints.
• Therapy rewires communities. Response can appear as a shift from immune-contact neighborhoods to fibroblast-shield neighborhoods; multicellular communities can be more stable than single markers.
• Metastasis is ecosystem-dependent. Malignant lineage states can be microenvironment-linked and organ-specific; plan analyses that respect patient-level heterogeneity.
• Early lesions have architecture. PanIN studies benefit from mapping TLS-like organization, not only immune abundance.

For a broader integration roadmap across modalities:

Frontiers in Spatial Multi-Omics Integration: From Data to Mechanism

Copy-ready PDAC spatial study blueprint (adaptable template)

Use this as a minimal, publishable scaffold. Choose one primary contrast, then build the smallest assay and analysis stack that can answer it convincingly.

Step 1 — Choose your primary contrast (match the biology)

• Neural invasion: NI-low vs NI-high; nerve-adjacent ROIs vs matched non-nerve ROIs
• Treatment response: NAT responders vs non-responders; paired baseline → resection when possible
• Metastasis: matched primary + liver/lung/peritoneum; organ-stratified analysis
• Early lesions: PanIN + adjacent tissue; PanIN vs PDAC-border immune architecture comparison

Step 2 — Minimum assay stack (pragmatic and reviewer-friendly)

• Spatial discovery: whole-transcriptome spatial profiling to define ecotypes/domains
• Reference atlas: scRNA-seq or snRNA-seq to refine immune and fibroblast states
• Validation: IF/IHC/RNA-ISH for 2–3 key markers plus one key spatial relationship (adjacency/distance)
• Optional: targeted functional assays only for the top 1–2 hypotheses (avoid over-expanding scope)

Step 3 — Analysis stack (what reviewers expect)

• QC + histology alignment; domain/ecotype detection; cell-type mapping/deconvolution
• Neighborhood enrichment / proximity analysis (tumor–immune, CAF–immune, nerve–tumor)
• Spatially constrained ligand–receptor prioritization within defined niches
• Patient- and site-stratified statistics (avoid over-claiming across heterogeneous PDAC)

Step 4 — Reporting stack (make the story easy to trust)

• Representative tissue images with overlays (domains, key markers, distance fields)
• Per-patient plots alongside pooled summaries
• Validation of key markers and at least one spatial relationship for each major claim

If you need a step-by-step analysis walkthrough: Spatial Transcriptomics Data Analysis: A Practical Introduction

FAQs

1. What is "spatial multi-omics" in PDAC research?

   It combines spatial maps (where cells are) with molecular profiles (what they express) to link PDAC biology to specific tissue niches.

2. Which platform should I choose: Visium or an imaging-based method (e.g., CosMx/Xenium/GeoMx)?

   Use Visium-like methods for discovery and architecture; add imaging when you need cell-level neighborhoods and niche validation.

3. How many samples do I need for a publishable PDAC spatial study?

   Plan for multiple patients per group and report results per patient; paired designs (e.g., pre/post NAT, primary/metastasis) strengthen conclusions.

4. What are the most common PDAC spatial study endpoints?

   Niche/domain maps, tumor–immune/CAF proximity metrics, spatially constrained ligand–receptor signals, and marker validation (IF/IHC/ISH).

5. How do I avoid over-interpreting spatial results?

   Predefine ROIs, include matched controls (e.g., nerve-adjacent vs non-nerve), control batch effects, and validate key spatial relationships with orthogonal assays.

Conclusion

Spatial multi-omics is transforming PDAC research by turning "cell states" into "cell states in place"—revealing nerve-adjacent invasion niches, fibroblast-built barriers, therapy-rewired communities, and organ-specific metastatic ecosystems. Across the six landmark studies summarized here, the most reproducible strategy is a focused contrast (e.g., NI-high vs NI-low, responder vs non-responder, primary vs metastasis) plus a minimal discovery-to-validation stack that reports patient-level spatial patterns. Use the study blueprints in each card to design a defensible cohort, select platforms efficiently, and prioritize one or two spatial mechanisms that can be validated with orthogonal assays.

Back to Resource hub: https://www.spatial-omicslab.com/resource.html

References

1. Chen, Min-Min, et al. "Integrated single-cell and spatial transcriptomics uncover distinct cellular subtypes involved in neural invasion in pancreatic cancer." Cancer Cell, 2025.
2. Hwang, William L., et al. "Single-nucleus and spatial transcriptome profiling of pancreatic cancer identifies multicellular dynamics associated with neoadjuvant treatment." Nature Genetics, 2022.
3. Khaliq, Ateeq M., et al. "Spatial transcriptomic analysis of primary and metastatic pancreatic cancers highlights tumor microenvironmental heterogeneity." Nature Genetics, 2024.
4. Lyman, Melissa R., et al. "Spatial proteomics and transcriptomics reveal early immune cell organization in pancreatic intraepithelial neoplasia." JCI Insight, 2025.
5. Pei, Guangsheng, et al. "Spatial mapping of transcriptomic plasticity in metastatic pancreatic cancer." Nature, 2025.
6. Zhou, Daniel Cui, et al. "Spatially restricted drivers and transitional cell populations cooperate with the microenvironment in untreated and chemo-resistant pancreatic cancer." Nature Genetics, 2022.