BPG is committed to discovery and dissemination of knowledge
Minireviews Open Access
Copyright: ©Author(s) 2026. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial (CC BY-NC 4.0) license. No commercial re-use. See permissions. Published by Baishideng Publishing Group Inc.
World J Gastrointest Oncol. Nov 15, 2026; 18(11): 124427
Published online Nov 15, 2026. doi: 10.4251/wjgo.124427
Perineural invasion in pancreatic ductal adenocarcinoma: Molecular mechanisms and clinical significance
Rong Hu, The Fourth School of Clinical Medicine, Zhejiang Chinese Medical University, Hangzhou 310000, Zhejiang Province, China
Rong Hu, Hong-Chen Zhang, Xiao-Feng Zhang, Department of Gastroenterology, Affiliated Hangzhou First People’s Hospital, School of Medicine, Westlake University; Hangzhou Institute of Digestive Diseases; Key Laboratory of Integrated Traditional Chinese and Western Medicine for Biliary and Pancreatic Diseases of Zhejiang Province; Key Laboratory of Clinical Cancer Pharmacology and Toxicology Research of Zhejiang Province, Affiliated Hangzhou First People’s Hospital, Hangzhou 310006, Zhejiang Province, China
ORCID number: Rong Hu (0009-0006-0577-5309); Hong-Chen Zhang (0000-0002-8956-0519); Xiao-Feng Zhang (0009-0008-7047-1058).
Co-first authors: Rong Hu and Hong-Chen Zhang.
Author contributions: Hu R and Zhang HC contributed equally to this article, they are the co-first authors of this manuscript; Hu R drafted most of the manuscript and prepared the initial tables; Zhang HC drafted the original manuscript, substantially rewrote and revised the manuscript after peer review, redesigned and revised the tables, prepared the figures, and drafted the responses to reviewers; Zhang XF designed the manuscript outline and coordinated its preparation; and all authors thoroughly reviewed and endorsed the final manuscript.
AI contribution statement: Some graphic elements in the Figure 1 come from the BioRender standard materials/icons library and were used in accordance with BioRender’s license.
Supported by the Construction Fund of Key Medical Disciplines of Hangzhou, No. 2025HZGF05; Zhejiang Provincial Key Research and Development Program, No. 2024C03048; and Zhejiang Provincial Key Discipline of Integrated Traditional Chinese and Western Medicine (Digestive Diseases), No. 2024-XK-61.
Conflict-of-interest statement: All the authors report no relevant conflicts of interest for this article.
Corresponding author: Xiao-Feng Zhang, MD, Chief Physician, Professor, Department of Gastroenterology, Affiliated Hangzhou First People’s Hospital, School of Medicine, Westlake University; Hangzhou Institute of Digestive Diseases; Key Laboratory of Integrated Traditional Chinese and Western Medicine for Biliary and Pancreatic Diseases of Zhejiang Province; Key Laboratory of Clinical Cancer Pharmacology and Toxicology Research of Zhejiang Province, Affiliated Hangzhou First People’s Hospital, No. 261 Huansha Road, Hangzhou 310006, Zhejiang Province, China. zhangxiaofeng@hospital.westlake.edu.cn
Received: June 15, 2026
Revised: August 12, 2026
Accepted: September 21, 2026
Published online: November 15, 2026
Processing time: 143 Days and 15.3 Hours

Abstract

Pancreatic ductal adenocarcinoma is an aggressive malignancy in which perineural invasion (PNI) is frequent and associated with adverse clinical outcomes. This article reframes PNI as an active, spatially organized invasive process within a neural-immune-stromal-metabolic niche rather than an isolated histopathological finding. It integrates interactions among tumor cells, nerves, Schwann cells, and immune and stromal compartments through neurotrophic tumor-nerve convergence, extracellular-matrix remodeling, metabolic and RNA regulation, and neural signaling across distinct evidence levels. PNI-associated pain reflects overlapping mechanical injury or compression and chemical and neuroimmune sensitization, although PNI does not explain all pain in pancreatic ductal adenocarcinoma. Histopathology remains the diagnostic reference, whereas quantitative pathology, radiomics, artificial intelligence-assisted pathology, and experimental imaging occupy different stages of maturity. No PNI-specific disease-modifying therapy has been clinically validated, and symptom control must remain distinct from interventions intended to modify neural invasion or disease outcomes. Spatial transcriptomics and multiplex imaging can refine biological characterization while preserving spatial context, but their application to PNI remains research-stage. Prospective validation, standardized annotation, independent cohorts, and PNI-specific endpoints are required before emerging diagnostic, prognostic, or therapeutic concepts can support routine care.

Key Words: Pancreatic ductal adenocarcinoma; Perineural invasion; Tumor microenvironment; Neurotrophic factors; Spatial omics; Targeted therapy

Core Tip: This article organizes perineural invasion in pancreatic ductal adenocarcinoma around a spatially organized neural-invasion niche, integrating neural-immune-stromal-metabolic crosstalk while preserving differences in evidence strength. Its key contribution is to connect this niche framework with a dual pain model that distinguishes mechanical injury and compression from chemical and neuroimmune sensitization. Clinical implications remain evidence-bounded: Spatial omics is research-stage, and diagnostic or therapeutic translation requires prospective validation with perineural invasion-specific endpoints.



INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy characterized by difficulty in early diagnosis, rapid progression, limited responsiveness to currently available therapies, and consequently poor overall prognosis; its 5-year survival rate remains low[1]. Among the various invasive patterns of PDAC, perineural invasion (PNI), in which tumor cells spread along nerve bundles, is one of the most characteristic pathological manifestations, with an incidence as high as 70%-100%[2,3]. PNI is not only an important route for local tumor dissemination, but is also closely associated with retroperitoneal margin positivity and postoperative local recurrence[4]. In addition, PNI is a major contributor to refractory cancer-related abdominal pain in patients with PDAC and seriously impairs quality of life[2,5].

Accumulating evidence associates PNI with adverse clinical outcomes, including higher rates of local recurrence and shorter survival[6,7]. In a large retrospective study of patients with PDAC who underwent curative-intent (R0) resection, patients without PNI (Pn0) had significantly longer disease-free survival (DFS) and overall survival (OS) than those with PNI (Pn1)[4]. PNI also remained clinically relevant among patients with early-stage PDAC, defined by a tumor diameter of ≤ 20 mm or the absence of lymph-node metastasis, in relation to disease progression and recurrence[3].

With increasing understanding of the tumor microenvironment (TME), PNI has increasingly been recognized as an active invasive process involving complex molecular signaling networks. This process is characterized by dynamic interactions among tumor cells, neurons, Schwann cells, and diverse immune and stromal populations[2,8]. These interactions involve neurotrophic factors, cytokines, chemokines, neurotransmitters, metabolic reprogramming, and changes in the immune microenvironment[8]. For example, tumor-associated macrophages (TAMs) and Schwann cells can form a positive-feedback loop through the basic fibroblast growth factor (bFGF)/interleukin (IL)-33 signaling axis, thereby promoting PNI[9]. Cholinergic signaling has also been implicated in tumor growth through remodeling of the immune microenvironment and suppression of CD8+ T-cell infiltration[10]. In addition, cancer-associated fibroblasts (CAFs) can promote the expression of PNI-related genes, including L1 cell adhesion molecule and slit guidance ligand 1, by regulating lactate metabolism and associated epigenetic changes[11].

Despite substantial progress, several important gaps remain. First, reliable biomarkers for the early prediction of PNI are lacking, limiting the accuracy of preoperative risk assessment. Second, although numerous molecular pathways associated with neural invasion have been identified, few have been translated into clinically effective therapeutic strategies. Third, the spatial organization and molecular interactions of the neural-invasion niche remain incompletely characterized. Emerging technologies, including spatial transcriptomics and other spatial omics approaches, may enable high-resolution characterization of the tumor-neural microenvironment and support the identification of candidate therapeutic targets.

In recent years, reviews focused on PNI have summarized the clinicopathological significance, biological characteristics, molecular mechanisms, and therapeutic prospects of PNI related to PDAC[1,2,7,8]. This article aims to provide different tissue frameworks, examining PNI in PDAC through the integrative concept of a neural-invasion niche rather than as an isolated pathological finding. We first establish the pathological and clinical context of PNI and then organize its biology around neurotrophic tumor–nerve convergence, Schwann cell-mediated remodeling, immune-stromal and extracellular-matrix feedback, and metabolic and epigenetic regulation. We next distinguish mechanical neural injury from chemical and neuroimmune sensitization in the generation of cancer pain. On this basis, we evaluate pathological and imaging-based detection, risk stratification, and therapeutic opportunities while separating direct evidence in PDAC-associated PNI from preclinical findings and indirect translational implications. Finally, we discuss how spatial omics may refine characterization of the neural-invasion niche and identify priorities for validation. This framework highlights the article’s specific contribution at the intersection of spatial biology, neural-immune-stromal-metabolic crosstalk, pain sensitization, and evidence-bounded clinical translation.

METHODOLOGY

This is a focused narrative review. This article searched PubMed, China National Knowledge Infrastructure, and Wanfang databases, covering literature from the establishment of the databases up to March 2026, with the final search updated in March 2026.

The PubMed core search was (“pancreatic ductal adenocarcinoma” OR PDAC OR “pancreatic cancer”) AND (“perineural invasion” OR PNI OR “neural invasion” OR “nerve invasion”). The core search was supplemented by topic-specific searches in which it was combined using AND with thematic term blocks related to molecular mechanisms, the TME, pain, diagnostic assessment, artificial intelligence, and spatial omics. Terms within each thematic block were combined using OR. A total of 787 papers were retrieved. Based on the core search strategy, we further conducted topic-specific searches, combining the core strategy with keywords like molecular mechanisms, TME, pain, diagnostic assessment, artificial intelligence, and spatial omics using AND; related search terms within the same topic group were connected with OR. The six topic-specific searches retrieved 1537 records in aggregate. Because substantial overlap existed across thematic searches, the six search sets were combined using OR in PubMed, yielding 766 unique records.

China National Knowledge Infrastructure and Wanfang Data used the corresponding Chinese disease and PNI terms along with the same six thematic concepts. They used subject fields when possible, and expanded to title, keyword, and abstract fields if subject-only searches weren’t enough, especially for spatial-omics queries.

The studies included need to involve PDAC-related PNI, tumor-nerve interactions, neural remodeling, cancer pain, or related immune, stromal, metabolic, and spatial processes. Original clinical and experimental studies directly involving PDAC-PNI are prioritized. Studies without PNI-specific endpoints are only included if they provide relevant mechanistic, spatial, or translational context, and they are clearly treated as indirect evidence. Recent studies are emphasized, while groundbreaking older studies and relevant reviews are kept when necessary. Studies that are not clearly related to PDAC, nerve invasion, or the biological context discussed here, as well as duplicate reports and publications with insufficient information, are excluded. As this was a narrative review, a PRISMA flow diagram, formal dual-independent screening, and standardized risk-of-bias assessment tools were not applied.

The evidence is interpreted in four levels: First, direct clinical evidence of PNI related to PDAC, meaning studies that directly assess PNI in human PDAC cohorts or tissues; second, direct PDAC evidence without PNI-specific endpoints; third, preclinical PDAC evidence supporting the target principle, coming from cells, organoids, in vitro, or animal models used to back up mechanism or target rationale; and finally, indirect evidence from non-PNI settings, including reviews, non-PNI-specific findings, or evidence from outside PDAC used to provide context or form hypotheses. Conclusions are expressed based on the maturity and directness of supporting evidence, with non-PNI-specific or preclinical findings clearly distinguished from clinically validated PDAC-PNI evidence. Note that these levels describe the directness and maturity of the evidence, not a formal risk-of-bias rating, nor do they indicate treatment efficacy.

PATHOLOGICAL DEFINITION, ASSESSMENT, AND CLINICAL CORRELATES OF NEURAL INVASION IN PDAC
Pathological definition and diagnostic criteria

PNI is an important pathological feature of PDAC and is typically defined as tumor-cell invasion into the perineural sheath or perineural space, or encirclement of more than one-third of a nerve’s circumference[1]. Increasingly, PNI is viewed not simply as a microscopic finding but as an active invasive process involving complex interactions among tumor cells, neural structures, and the TME[8]. This process promotes local tumor spread and progression, is associated with cancer-related pain, and correlates with poor outcomes in patients with PDAC[8].

Histopathological diagnosis of PNI relies primarily on hematoxylin and eosin (HE) staining of resection specimens. However, routine HE staining may be insufficient to detect minute foci of PNI; immunohistochemistry is therefore often used as an adjunct. S100 can highlight Schwann cells, whereas PGP9.5 labels nerve fibers, facilitating evaluation of the spatial relationship between tumor cells and nerves[2].

PNI assessment generally remains binary, categorizing cases as present or absent, and uniform quantitative criteria are lacking[12]. Proposed metrics include the number of involved nerves, nerve diameter, and depth of invasion; however, their prognostic value varies among studies. This variability may reflect differences in pathological assessment criteria, sample size, tumor-stage distribution, and sampling depth. Studies relying solely on HE staining may also differ from those incorporating PGP9.5 or S100 immunostaining, which can increase the detection of small invasive foci. One study proposed a severity scoring system based on nerve diameter and invasion pattern: Score 0, no PNI; score 1, invasion of nerves < 3 mm in diameter; and score 2, invasion of nerves ≥ 3 mm in diameter, extensive perineural infiltration, and/or nerve-bundle necrosis. This score was associated with invasive features of PDAC and patient prognosis[12].

Relationship between neural invasion, clinicopathological parameters, and prognosis

PNI is an important adverse prognostic factor in PDAC and is associated with tumor progression, local recurrence, and neuropathic pain[6]. Clinical studies have also linked PNI to aggressive clinicopathological features, including larger tumor size, poorer histological differentiation, lymph-node metastasis, lymphatic or vascular invasion, and more advanced tumor-node-metastasis stage. In a multicenter retrospective study of 778 patients with PDAC who underwent pancreatectomy, the overall PNI rate was 87%; even among patients with early-stage disease and tumors ≤ 20 mm, the rate remained 78.7%[3]. PNI was independently associated with both DFS [hazard ratio (HR) = 1.71, 95% confidence interval (CI): 1.14-2.58] and OS (HR = 1.51, 95%CI: 1.05-2.18). Median OS was shorter in PNI-positive than in PNI-negative patients (21.0 months vs 55.0 months), as was median DFS (14.0 months vs 47.0 months).

Another retrospective analysis of 571 patients with PDAC who underwent radical surgery found PNI in approximately 93% of patients; only 7% were PNI-negative. Median OS was 82 months in PNI-negative patients and 21 months in the PNI-positive group, corresponding to a significantly increased risk of death (HR = 2.39, 95%CI: 1.34-4.26)[4]. This survival difference remained significant among patients who underwent R0 resection, suggesting that PNI may be an important prognostic factor for long-term outcomes after curative-intent surgery[4].

Greater PNI severity is also associated with poorer prognosis. Invasion of larger nerve trunks or extensive perineural infiltration is associated with a significantly higher risk of local recurrence and poorer survival[12]. Neural invasion at retroperitoneal margins, including the superior mesenteric artery margin, is an important predictor of recurrence after R0 resection and may provide prognostic information beyond tumor-cell margin positivity alone[3].

These prognostic estimates mainly come from retrospective surgical cohorts and may be influenced by the selection of resectable patients, differences in pathological sampling and scoring, treatment heterogeneity, and residual confounding factors[3,4,12]. Therefore, these studies support the prognostic association of PNI with adverse outcomes but do not establish PNI as a modifiable causal determinant of survival or justify treatment intensification based on PNI status alone.

Overall, PNI is both an important pathological feature of PDAC and a clinically relevant marker for prognostic assessment and risk stratification.

MOLECULAR AND CELLULAR MECHANISMS DRIVING NEURAL INVASION IN PDAC

For this article, PNI in PDAC is conceptualized as a spatially organized neural-immune-stromal-metabolic niche in which tumor cells interact with neurons, Schwann cells, immune cells, CAFs, and the extracellular matrix through molecular and metabolic signals[2,8-11]. Rather than assigning equivalent weight to evidence from different mechanistic domains, this framework integrates them while retaining their distinct evidentiary contexts (Table 1). The discussion progresses from neurotrophic chemotaxis and tumor-nerve convergence[8,13] to Schwann-cell remodeling[9,14,15], immune-stromal feedback and extracellular-matrix remodeling[8,9,11,16], and metabolic, epigenetic, and non-coding RNA regulation[8,17,18]. Neural signaling and its implications for pain sensitization and therapeutic resistance are then considered[5,7,10,19]. By linking mechanisms previously discussed in isolation, this sequence preserves distinctions among direct evidence in PDAC-associated PNI, preclinical findings, and indirect translational implications (Figure 1).

Figure 1
Figure 1 Neural-invasion niche and molecular mechanisms in pancreatic ductal adenocarcinoma. Some graphic elements in the images come from the BioRender standard materials/icons library and were used in accordance with BioRender’s license. This figure illustrates the neural-immune-stromal-metabolic niche associated with perineural invasion (PNI) in pancreatic ductal adenocarcinoma (PDAC). The central anatomical interface consists of PDAC cells approaching and interacting with a peripheral nerve, together with Schwann cells, TAMs, cancer-associated fibroblasts, and extracellular matrix. Neurotrophic tumor–nerve signaling, including nerve growth factor/tropomyosin receptor kinase A - and glial cell line-derived neurotrophic factor/rearranged during transfection-related pathways, may promote tumor-nerve approximation and invasive behavior. Tumor-nerve adhesion further stabilizes contact at the interface. Schwann-cell remodeling may provide a permissive neural microenvironment for tumor-cell migration. Immune–glial signaling, including the TAMs-Schwann-cell basic fibroblast growth factor/interleukin-33 paracrine circuit, may reinforce local neural and immune remodeling. Extracellular matrix remodeling may provide a migration-permissive structural environment, whereas CAF-associated lactate metabolism and related RNA/epigenetic programs may further modify PNI-associated tumor behavior. Neural signaling represents additional communication between the nerve compartment and the surrounding TME. Arrowheads indicate the direction of the schematic relationship toward the relevant cellular or anatomical compartment. Solid connectors indicate relationships more directly supported by PDAC/PNI evidence; dashed connectors indicate predominantly preclinical, contextual, or integrative relationships. The spatial-evidence panel distinguishes direct human PNI spatial evidence[28], human neural-invasion-associated cellular states[29], and predominantly experimental neural injury/repair evidence[30]. These evidence categories indicate differences in the source and maturity of evidence rather than therapeutic efficacy. PDAC: Pancreatic ductal adenocarcinoma; CAFs: Cancer-associated fibroblasts; bFGF: Basic fibroblast growth factor; TAMs: Tumor-associated macrophages; PNI: Perineural invasion; ECM: Extracellular matrix; IL-33: Interleukin-33.
Table 1 Mechanistic layers and evidence boundaries of perineural invasion in pancreatic ductal adenocarcinoma.
Mechanistic layer
Representative axis/process
Cellular context and biological role
Evidence level and translational implication
Key references
Neurotrophic tumor-nerve convergenceNGF-TrkA signalingContext: Tumor-nerve interfaceEvidence: Review-level synthesis[8]
Role: Associated with chemotaxis, adhesion, and invasive behavior within reciprocal tumor–nerve signalingTranslation: Mechanistic target for further study
Neurotrophic tumor-nerve convergenceGDNF-RET signalingContext: Experimental pancreatic cancer modelsEvidence: Experimental pancreatic cancer evidence[13]
Role: Reported to engage CDC42/RHOA cytoskeletal regulation and SRC/GRB2-associated invadopodia formationTranslation: Preclinical therapeutic rationale
Schwann-cell remodelingInjury-responsive Schwann-cell redistribution and neural remodelingContext: Schwann cells at tumor-associated peripheral nervesEvidence: Review-level synthesis[14]
Role: May provide a permissive neural-remodeling context for tumor-cell migration without establishing a single recruitment directionTranslation: Mechanistic process requiring direct validation
Immune-glial feedbackTAM-Schwann bFGF/IL-33 paracrine feedbackContext: Tumor-associated macrophages and Schwann cellsEvidence: Experimental pancreatic cancer evidence[9]
Role: Mechanistic evidence suggests reciprocal activation, macrophage recruitment, and M2-like polarization in the perineural nicheTranslation: Candidate target-discovery pathway; clinical relevance requires validation
Tumor-nerve adhesionL1CAM-associated adhesion and metalloproteinase signalingContext: Tumor-nerve interfaceEvidence: Experimental pancreatic cancer evidence[15]
Role: Associated with adhesive interactions and metalloproteinase-linked invasive behaviorTranslation: Mechanistic target for further study
Extracellular-matrix remodelingLAMA3 and extracellular-matrix organizationContext: Perineural tumor–stromal matrixEvidence: Human pancreatic cancer association[8,16]
Role: LAMA3 expression and matrix reorganization are associated with perineural invasion and adverse pathological features; causality is not establishedTranslation: Association requiring validation
Metabolic regulationCAF-associated high-lactate microenvironmentContext: Cancer-associated fibroblast-rich tumor microenvironmentEvidence: Experimental pancreatic cancer evidence[11]
Role: High lactate is linked to epigenetic regulation of perineural-invasion-related programs, including L1CAM and SLIT1Translation: Mechanistic pathway for further study
Non-coding RNA regulationCAF-derived circKLHL24-Sec31A/miR-615-5p-CXCL12 regulationContext: Cancer-associated fibroblasts and pancreatic cancer cellsEvidence: Experimental pancreatic cancer evidence[17]
Role: Reported to regulate Sec31A and miR-615-5p/CXCL12 signaling in perineural-invasion-related phenotypesTranslation: Preclinical target-discovery pathway
RNA modificationLactate-NSUN2-m5C-CDCP1/STC1 regulationContext: High-lactate microenvironment and pancreatic cancer cellsEvidence: Experimental pancreatic cancer evidence[18]
Role: Lactate-associated NSUN2 modification is linked to m5C-dependent stabilization of CDCP1 and STC1 transcripts and perineural-invasion-related phenotypesTranslation: Candidate metabolic-epigenetic pathway requiring validation
Neural immune regulationAcetylcholine-M3R/HDAC1-CCL5 signalingContext: Nerve-associated cholinergic signaling, pancreatic cancer cells, and local immune contextEvidence: Human pancreatic cancer mechanistic evidence[10]
Role: Reported to alter CCL5-dependent CD8+ T-cell recruitment and function in mechanistic modelsTranslation: Preclinical therapeutic rationale; no established clinical efficacy
Neural signaling and drug-response phenotypeNeurotensin-NTSR1-Akt signalingContext: Neuron-tumor signaling in experimental pancreatic cancerEvidence: Experimental pancreatic cancer evidence[19]
Role: Associated with invasiveness and a gemcitabine-resistance phenotype in experimental modelsTranslation: Preclinical therapeutic rationale
Pain-related neural signalingNGF-related sensory sensitization and nociceptor signalingContext: Tumor–sensory nerve interfaceEvidence: Review-level synthesis[5,7,22]
Role: Supports a symptom-relevant sensitization pathway but does not establish a single cause of pain or a direct relation between invasion severity and pain intensityTranslation: Symptom-relevant mechanism requiring direct validation

This organization represents an integrative framework. And the molecular biological mechanisms mentioned above aren’t independent processes. Neurotrophic signals help tumors move closer to nerves, adhesion makes these local contacts more stable, Schwann cell remodeling might create a permissive nerve interface, and subsequent immune-matrix/extracellular matrix feedback along with metabolic/RNA regulation could further change the local microenvironment and tumor adaptability.

Neurotrophic chemotaxis and tumor-nerve convergence

Neurotrophic signaling is a central component of tumor-nerve convergence in PDAC-associated PNI. Among the best-characterized pathways, nerve growth factor/tropomyosin receptor kinase A (NGF/TrkA) signaling has been associated with directed tumor-cell migration toward neural structures and with increased adhesive and invasive behavior[8]. Because the cellular source of NGF and the distribution of TrkA may vary across experimental contexts, this axis is more appropriately interpreted as reciprocal tumor–nerve signaling than as a fixed unidirectional circuit. Glial cell line-derived neurotrophic factor (GDNF)/RET receptor tyrosine kinase (RET) represents a complementary pathway. In experimental PDAC models, RET signaling has been linked to cell-polarity and cytoskeletal remodeling through CDC42 and Ras homolog family member A, as well as to invadopodia formation and matrix degradation involving SRC proto-oncogene, non-receptor tyrosine kinase and growth factor receptor-bound protein 2[13]. Brain-derived neurotrophic factor and neurotrophin-3 may contribute to the broader neurotrophic network[8], whereas Nodal has been reported to increase neurotrophic-factor expression and promote tumor-nerve convergence[20]. Viewed collectively, the evidence supports a model in which multiple neurotrophic cues facilitate spatial approximation and reciprocal adaptation between tumor and neural compartments, although their relative importance and clinical relevance remain to be established.

Schwann cell-mediated neural remodeling

Schwann cells, the principal glial cells of the peripheral nervous system, have been implicated as active participants in PDAC-associated PNI[14]. Within the neural-invasion niche, their redistribution around invaded nerves may link local neural remodeling with tumor-cell migration. The available evidence is most consistent with a permissive, rather than exclusively initiating, role: Schwann cells may reorganize the perineural microenvironment and facilitate tumor movement along neural structures, although the sequence and cellular direction of these events vary among experimental models[14]. Immune-derived signals may further shape Schwann-cell behavior, connecting glial remodeling with the immune and matrix components of the niche. Schwann-cell remodeling thus provides a mechanistic bridge to immune–stromal feedback and extracellular-matrix remodeling.

Immune-stromal feedback and extracellular-matrix remodeling

Immune feedback may amplify Schwann-cell remodeling within the neural-invasion niche. A reported bFGF/IL-33 paracrine circuit links TAMs and Schwann cells and has been associated with Schwann-cell activation, macrophage recruitment, and polarization toward an M2-like phenotype[9]. This reciprocal interaction may contribute to maintaining conditions permissive to PNI; however, the evidence is predominantly mechanistic and does not constitute prospective clinical validation.

Direct tumor-nerve contact constitutes a complementary route for local invasion. Neural cell adhesion molecule and L1 cell adhesion molecule have both been implicated in strengthening adhesion between pancreatic cancer cells and neural structures[8]. In particular, L1 cell adhesion molecule expression has been linked to metalloproteinase upregulation in pancreatic cancer cells and to enhanced invasive behavior[15]. Adhesion-dependent contact and localized matrix degradation may therefore act in concert to stabilize the tumor-nerve interface and facilitate extension along neural structures.

These cellular interactions occur within a remodeled perineural extracellular matrix. Increased deposition and rearrangement of laminin and collagen may create a migration-permissive matrix architecture around invaded nerves[8]. High expression of laminin subunit alpha 3 has also been associated with PNI and poor prognosis in PDAC[16]. This association does not establish laminin subunit alpha 3 as a causal driver, but it is consistent with a contribution of matrix composition to the neural-invasion niche. Immune-glial feedback, adhesion-dependent tumor-nerve contact, and extracellular-matrix remodeling consequently provide the structural context in which metabolic, epigenetic, and non-coding RNA programs may further influence invasion.

Metabolic, epigenetic, and non-coding RNA regulation

Metabolic reprogramming adds a stromal regulatory layer to the neural-invasion niche. A high-lactate microenvironment associated with CAFs has been linked to epigenetic regulation of PNI-associated genes, including L1 cell adhesion molecule and slit guidance ligand 111]. These observations connect extracellular metabolic conditions with tumor-cell programs that may favor PNI, but they do not establish metabolic intervention as a clinically validated strategy for PDAC-associated PNI.

Non-coding RNAs add a mechanistically diverse regulatory layer. Long non-coding RNAs and circular RNAs have been implicated in regulating PNI-related gene expression through competing endogenous RNA and related mechanisms[8]. As a more specific example, CAF-derived circKLHL24 has been reported to stabilize Sec31A and sponge miR-615-5p, thereby increasing C-X-C motif chemokine ligand 12 expression and promoting neural invasion in pancreatic cancer[17]. This example indicates how stromal RNA signals may influence tumor behavior, but it should not be extrapolated to all non-coding RNAs implicated in PNI.

Lactate may also regulate RNA modification. Lactate-dependent lactylation of the RNA methyltransferase NOP2/Sun RNA methyltransferase 2 has been linked to enhanced m5C modification and stabilization of CUB domain containing protein 1 and stanniocalcin 1 mRNAs, with consequent promotion of PNI-related phenotypes[18]. Rather than forming a single uniform pathway, CAF-associated lactate metabolism, non-coding RNA regulation, and NSUN2-mediated RNA modification represent complementary regulatory layers. These processes may link stromal remodeling with neural signaling and tumor adaptation.

Neural signaling and therapeutic resistance

Cholinergic signaling is one route through which neural activity may reshape local immunity. In experimental PDAC models, acetylcholine-dependent M3R signaling has been linked to HDAC1-mediated suppression of C-C motif chemokine ligand 5 release from pancreatic cancer cells, reduced CD8+ T-cell recruitment, and impaired interferon-γ production[10]. These changes may collectively support an immunosuppressive microenvironment favorable to tumor progression and PNI. This interpretation is confined to the reported effects on recruitment and effector function and does not imply an unsupported change in CD8+ T-cell lineage.

Neurotensin is a distinct neural mediator with potential implications for invasion and chemoresistance. Neuron-derived neurotensin has been reported to engage neurotensin receptor 1 and activate protein kinase B signaling, thereby increasing pancreatic cancer-cell invasiveness and resistance to gemcitabine[19]. This evidence supports a mechanistic link between neural input and tumor adaptation but does not demonstrate clinical benefit from targeting this pathway in patients with PDAC-associated PNI. At the intracellular level, Ras/mitogen-activated protein kinase, phosphoinositide 3-kinase/protein kinase B, and Janus kinase/signal transducer and activator of transcription should be interpreted as convergent tumor-associated pathways rather than as PNI-specific mechanisms[8]. A more specifically defined cascade involves GDNF-associated, cyclin-dependent kinase 1-mediated phosphorylation of the intracellular domain of mucin 21, cell surface associated, enhanced interaction with Rac family small GTPase 2, and activation of c-Jun N-terminal kinase/zinc finger E-box-binding homeobox 1 signaling, with reported effects on epithelial-mesenchymal transition, migration, metastasis, and neural invasion[21]. This sequence indicates how extracellular neural signals may converge on tumor-cell invasive programs. Cholinergic immune regulation, neurotensin-linked chemoresistance, and intracellular signaling convergence thereby complete the mechanistic framework and provide a transition to the distinct mechanisms of pain generation and sensitization.

MECHANISMS OF PAIN GENERATION AND SENSITIZATION

Pain in PDAC is multifactorial, and PNI does not account for every pain phenotype. For analytical clarity, PNI-associated pain can be organized into two overlapping pathways. The mechanical or anatomical pathway involves perineural infiltration, nerve encasement or compression, structural nerve injury, and altered local neural signaling. The chemical or neuroimmune pathway involves sensitization mediated by neurotrophic, nociceptor-derived, inflammatory, immune, and glial signals. Both pathways may coexist, although their relative contributions are likely to vary among patients and across disease stages. Distinguishing anatomical injury from molecular sensitization provides a clearer framework for interpreting the heterogeneous pain associated with PDAC (Figure 2).

Figure 2
Figure 2 Pain generation and sensitization associated with perineural invasion in pancreatic ductal adenocarcinoma. This figure summarizes two overlapping pathways involved in pain associated with perineural invasion in pancreatic ductal adenocarcinoma. On the left is the mechanical and anatomical pathway. After tumor cells infiltrate the perineural space, they may further encase or compress nerves, causing structural nerve damage and altering local neural signaling. These anatomical changes may constitute the structural basis for the characteristics of persistent cancer pain and neuropathic pain. The middle section depicts the chemical and neuroimmune pathway. Nerve growth factor receptor signaling may enhance the responsiveness of sensory neurons and promote sensitization of nociceptors; signals from nociceptors, along with modulation by immune and glial cells, may further increase local neural sensitivity. The right-hand section illustrates clinical pain outcomes that may arise from the integration of mechanical/anatomical injury and chemical/neuroimmune sensitization, including persistent cancer pain, neuropathic features, and interpatient heterogeneity. Solid lines show the mechanical/anatomical sequence from perineural infiltration to nerve sheath compression, structural nerve damage, and changes in nerve signaling. Dashed lines represent the convergence or regulation between altered nerve signals, nerve growth factor-related sensitization, nociceptor signaling, immune and glial modulation, and pain outcomes. The statement “PNI burden ≠ pain intensity” in the figure emphasizes that the anatomical burden of perineural invasion does not directly determine pain intensity. Pain in different patients may result from a combination of structural nerve damage and chemical/neuroimmune sensitization to varying degrees. PNI: Perineural invasion; NGF: Nerve growth factor.
Mechanical injury caused by neural infiltration and compression

Perineural infiltration in PDAC may extend into the perineurium or perineural space and may encase or compress nerve fibers. These anatomical changes can disrupt nerve-bundle organization and alter local neural signaling, providing a plausible structural basis for persistent deep pain and neuropathic features[2,8]. Tumor-nerve contact, increased perineural tension, and compression may therefore contribute to pain generation; however, no single anatomical feature fully explains the diversity of pain experienced by patients with PDAC. Mechanical injury should consequently be considered one component of PNI-associated pain rather than a complete explanation for cancer pain.

Pathological assessment can characterize the anatomical burden of PNI by evaluating the extent of neural invasion, the number of involved nerves, and the caliber of affected nerves. A severity-score system based on quantitative PNI features has been proposed[12]. These measures may provide a reproducible basis for future studies relating anatomical burden to patient-reported symptoms, neuropathic features, imaging findings, or analgesic requirements. However, broader invasion or involvement of larger nerves cannot be taken as evidence of greater pain intensity because existing severity scores were developed to quantify PNI rather than pain itself.

Chemical sensitization through neuroimmune signaling

Chemical sensitization provides a complementary route through which PNI may contribute to pain. NGF signaling through TrkA and p75NTR has been linked to neuronal survival, axonal growth, and increased sensory-neuron responsiveness[5,7]. In the context of pain, this signaling may lower the nociceptive threshold and amplify responses to local stimuli. This sensory effect should be distinguished from the role of NGF in tumor-nerve convergence discussed earlier, where the relevant processes are chemotactic attraction and tumor-cell migration.

Nociceptor-derived signaling may further amplify communication among sensory nerves, stromal cells, inflammatory components, and tumor cells. Nociceptor activity has been linked to both PDAC progression and cancer pain[22]. Such signaling may reinforce local inflammatory communication and increase neural responsiveness, creating a potential feedback relationship among tissue inflammation, sensory activation, and tumor–nerve interactions. The available evidence does not justify assigning a dominant role to any single neuropeptide; the process is therefore best described at the broader level of nociceptor-derived signaling.

Immune and glial pathways may further modulate the sensitizing environment. The bFGF/IL-33 circuit between TAMs and Schwann cells has been associated with PNI-promoting cellular feedback[9], whereas cholinergic signaling can reshape local immune conditions in experimental PDAC models[10]. By altering inflammatory, immune, or glial conditions around invaded nerves, these pathways may indirectly influence neural sensitization. Direct pain evidence for either pathway remains limited; their potential contribution to sensitization should therefore be distinguished from their mechanistic roles in PNI biology and immune regulation.

Integration of pain mechanisms and clinical implications

Mechanical injury and chemical or neuroimmune sensitization may coexist, but their relative contributions are likely to vary among patients. Anatomical PNI burden, even when quantified using pathological severity features[12], does not directly determine pain intensity or fully explain variation in neuropathic symptoms. Future studies integrating symptom profiles with imaging, pathological measurements, and molecular features may help distinguish predominantly structural nerve injury from prominent sensitization or mixed mechanisms. This framework is descriptive rather than prescriptive: It identifies variables that may inform subsequent diagnostic assessment, risk stratification, and individualized pain evaluation without specifying treatment.

DIAGNOSTIC ASSESSMENT AND RISK STRATIFICATION

Postoperative histopathology remains the diagnostic reference for assessing PDAC-associated PNI. Quantitative pathological descriptors can extend binary reporting, contrast-enhanced computed tomography (CT), radiomics, artificial intelligence (AI)-assisted pathology, and intraoperative imaging provide complementary information at different stages of maturity. These approaches span established tissue-based assessment, retrospective prediction models, and experimental imaging. Arranging them along a continuum from detection through burden characterization to risk estimation clarifies their distinct roles without conflating diagnostic performance with demonstrated clinical benefit.

Pathological assessment and quantitative grading

Postoperative histopathology remains the diagnostic reference for PNI in PDAC. On hematoxylin-eosin-stained sections, PNI is identified when tumor cells invade the perineural sheath or space or encircle more than one-third of a nerve circumference[1]. Minute foci may be difficult to recognize in routine sections. S100 and PGP9.5 immunostaining may serve as adjunctive markers of Schwann cells and nerve fibers, respectively, and may clarify tumor–nerve spatial relationships[2]. These stains are complementary rather than obligatory because detection also depends on tissue sampling, section depth, and the distribution of small invaded nerves.

Binary reporting of PNI as present or absent does not capture the anatomical burden or heterogeneity of PNI. Proposed quantitative descriptors include the number and caliber of involved nerves and the extent or pattern of invasion. A severity-score system incorporating nerve caliber and invasion extent has been associated with pathological aggressiveness and prognosis[12]. However, differences in sampling, stage distribution, and assessment criteria limit comparisons across studies. Standardized definitions, sampling protocols, and external validation are required before quantitative grading can inform clinical decisions.

Radiomics and preoperative risk prediction

Contrast-enhanced CT is central to preoperative staging and assessment of resectability in PDAC, but it does not directly resolve microscopic PNI. Treatment-related changes after neoadjuvant chemoradiotherapy may further complicate interpretation, and CT has been reported to overestimate extra-pancreatic PNI in this setting[23]. Conventional CT findings should therefore be interpreted as anatomical context rather than as definitive evidence of microscopic neural invasion.

CT-based radiomics remains a research-stage approach to preoperative PNI prediction that combines quantitative tumor-region features with clinical variables. One reported model achieved an area under the receiver operating characteristic curve of 0.84, with 81.8% sensitivity and 80.2% specificity in its validation cohort[24]. These values represent study-specific discrimination and do not establish routine clinical utility or patient benefit. External validation, calibration, and assessment of transportability across scanners and institutions remain necessary before radiomics can support clinical risk-stratification decisions.

AI-assisted pathology and intraoperative imaging

AI-assisted pathology is a research-stage approach that may support automated recognition of neural structures, tumor cells, and their spatial relationships in digitized tissue sections. A reported PDAC PNI-detection model achieved an area under the receiver operating characteristic curve of up to 0.98[25]. Although this performance suggests potential gains in screening efficiency and reproducibility, it remains dependent on the dataset, annotation strategy, and validation setting. AI should therefore be regarded as decision support rather than as a replacement for pathological interpretation.

Intraoperative frozen-section assessment can provide real-time evaluation of selected margins or suspicious neural tissue. Its utility is limited by tissue coverage, sampling error, specimen quality, and dependence on expert interpretation. Frozen sections cannot exclude unsampled microscopic PNI and should not be represented as a comprehensive method for real-time risk stratification.

Experimental optical imaging offers a potential additional route to intraoperative visualization. A growth-associated protein 43-targeted near-infrared fluorescent probe has been reported for real-time mapping of PNI lesions in pancreatic cancer in vivo[26]. This approach remains experimental and should not be described as clinically available or as improving surgical outcomes. Its specificity, safety, model context, and performance in patients require further evaluation before clinical translation.

Integrated risk stratification and current limitations

A proposed integrated PNI risk-stratification framework could combine pathological burden, CT radiomics, AI-assisted tissue assessment, and intraoperative findings, with subsequent incorporation of spatial or molecular profiling. These inputs address different tasks: Histopathology detects and characterizes PNI, radiomics estimates preoperative risk, AI may assist image interpretation, and experimental imaging may support localization. The modalities should not be treated as interchangeable measures. No unified tool integrating these modalities has been prospectively and externally validated as a clinical standard. Future studies should distinguish among detection accuracy, risk prediction, calibration, localization, decision support, and patient outcomes. This evidence-bounded framework provides a transition to therapeutic translation and spatial profiling without implying that an integrated multimodal score is currently available for routine care (Figure 3).

Figure 3
Figure 3 Diagnostic assessment, therapeutic translation, and validation road map for perineural invasion in pancreatic ductal adenocarcinoma. This figure summarizes the assessment methods of perineural invasion (PNI), the information obtained from these assessments, biological/molecular stratifications still at the research stage, therapeutic translation, and future validation requirements. The left section pertains to clinical diagnosis and post-assessment risk evaluation. Histopathology remains the clinical reference standard for PNI diagnosis, and quantitative features of PNI can further describe its detection and invasion burden. Computed tomography radiomics and artificial intelligence-assisted pathology are still in the research stage, while intraoperative imaging is mainly experimental. The information obtained from assessments includes PNI detection, characterization of PNI burden, preoperative risk evaluation, and molecular stratification, with the latter two not yet established as routine clinical tools. It is important to stress that spatial and molecular analysis is still a research-stage method for studying biological features and molecular layering, and it should not be seen as a routine tool for preoperative PNI risk assessment or treatment decisions. The third section concerns therapeutic translation, divided into disease-modification research and symptom management. Candidate molecular pathways mainly provide preclinical or conceptual guidance for therapeutic targets. Pain management, celiac plexus interventions, and local symptom control fall under symptom-oriented treatments. The fourth section outlines future validation pathways, including cross-platform validation, independent cohorts, correlation with clinical outcomes, and multimodal integration. Solid lines indicate relatively direct relationships within the current assessment framework, while dashed lines represent research-stage, proposed, or still-to-be-validated pathways, which do not imply established causal relationships or routine clinical procedures. The blue dashed line at the bottom of the figure connects potential downstream applications, including risk stratification, intervention planning, pain assessment, and outcome monitoring; however, their clinical application still requires further validation. Currently, there is no evidence showing that modifying PNI results in definite clinical benefits. PNI: Perineural invasion; CT: Computed tomography; AI: Artificial intelligence; NGF: Nerve growth factor; Trk: Tropomyosin receptor kinase; GDNF: Glial cell line-derived neurotrophic factor; RET: Rearranged during transfection.
THERAPEUTIC TRANSLATION AND EVIDENCE GAPS

Mechanistic studies have identified several candidate pathways in PDAC-associated PNI, but therapeutic evidence remains substantially less mature than the underlying biology. Current evidence spans preclinical target rationale, indirect observations outside PNI-specific settings, symptom-directed management, and clinical PDAC studies without dedicated PNI treatment endpoints. These categories are not equivalent. No PNI-specific disease-modifying therapy has been clinically validated in PDAC. Therapeutic interpretation must therefore distinguish experimental pathway modulation from established symptom control and avoid translating biological plausibility into assumed clinical benefit.

Preclinical targeting of neural-invasion signaling

NGF/Trk signaling links tumor-nerve communication with neural invasion and cancer-related pain in PDAC-associated PNI[7,8]. Experimental evidence provides a rationale for testing whether Trk inhibition can disrupt neurotrophic signaling. However, this rationale remains preclinical, and pharmacological Trk inhibition has not been shown to reduce PNI burden, relieve PNI-associated pain, or improve clinical outcomes in patients with PDAC. The NGF/Trk axis should therefore be regarded as a mechanistically grounded research target rather than an established therapeutic option. GDNF/RET signaling provides a complementary preclinical target rationale. In experimental PDAC systems, RET isoforms have been implicated in directional motility, cytoskeletal regulation, invadopodia formation, and other invasive processes[13]. These observations support further mechanistic investigation of RET signaling in tumor–nerve convergence but do not demonstrate therapeutic efficacy against PDAC-associated PNI. Therapeutic interpretation of RET-directed approaches should therefore remain bounded by the mechanistic evidence unless future studies incorporate PNI-specific pathological, imaging, pain, or clinical endpoints.

Neural signaling may also be modulated through cholinergic and β-adrenergic pathways. In experimental PDAC models, cholinergic signaling has been linked to HDAC1-dependent immune regulation, including altered C-C motif chemokine ligand 5 release, CD8+ T-cell recruitment, and interferon-γ production[10]. Pharmacological modulation of this pathway remains preclinical. β-Adrenergic signaling has also been associated with pancreatic cancer biology in experimental models, whereas human observations of nonselective β-blocker exposure remain observational[27]. These associations do not establish therapeutic causality, and propranolol should not be regarded as a clinically validated PNI-targeted treatment.

Immune-stromal and metabolic intervention concepts

The immune-glial microenvironment may represent an additional research layer within the neural-invasion niche. A bFGF/IL-33 paracrine circuit involving TAMs and Schwann cells has been associated with Schwann-cell activation, macrophage recruitment, and M2-like polarization[9]. Cholinergic signaling may further alter immune conditions through HDAC1-related regulation of tumor-cell signals and T-cell recruitment[10]. These observations provide a rationale for investigating immune and glial interactions, but neither blockade of the bFGF/IL-33 circuit nor broader immune modulation has been shown to reduce clinical PNI or improve survival in patients with PDAC.

Stromal metabolism and RNA regulation offer related avenues for target discovery and biomarker development. CAF-associated lactate accumulation has been linked to PNI-related gene regulation[11], whereas CAF-derived circKLHL24 and lactylation-dependent NSUN2-mediated RNA m5C regulation have been associated with neural-invasion phenotypes[17,18]. These mechanisms link extracellular metabolic conditions with non-coding RNA and RNA-modification programs but do not establish an available metabolic or RNA-targeted treatment. Mechanistic vulnerability should not be equated with validated therapy, and clinical efficacy against PDAC-associated PNI remains unproven.

Symptom-directed pain management and local treatment

Pain management in PDAC should prioritize symptom relief and quality of life while recognizing the multifactorial nature of cancer pain. Analgesic therapy remains central, and celiac plexus block or neurolysis may be considered for selected patients with difficult-to-control pancreatic cancer pain[2,7]. These procedures reduce neural transmission to relieve symptoms but have not been shown to reduce PNI burden or modify the neural-invasion niche. Analgesic response should therefore remain distinct from tumor control, recurrence, and survival.

Radiotherapy, including stereotactic body radiotherapy, may be used in selected local-treatment or symptom-control settings in PDAC[2]. Its role should be defined by the clinical indication rather than presented as a validated PNI-directed strategy. Current evidence does not establish that radiotherapy specifically reduces neural invasion, provides PNI-specific pain relief, or improves survival by modifying PNI biology.

Surgical extension into involved nerve plexuses remains controversial. Wider dissection may increase treatment-related morbidity, including nerve dysfunction, whereas the cited evidence has not established a reduction in recurrence or an improvement in survival[2]. PNI is associated with recurrence and survival after resection, including in early-stage disease[3], but this prognostic association does not demonstrate that intensified local treatment improves outcomes in PNI-positive patients. Treatment selection should therefore not be based on PNI status alone in the absence of prospective comparative evidence.

Evidence hierarchy and translational priorities

Therapeutic interpretation can be organized into four evidence levels: Direct clinical evidence in PDAC-associated PNI; direct PDAC evidence without a PNI-specific endpoint; preclinical PDAC evidence supporting target rationale; and indirect evidence from molecularly selected tumors outside a PNI-specific setting. No intervention discussed in this article has demonstrated clinical modification of PNI burden in PDAC. Pain-directed interventions constitute a separate symptom-control layer and should not be used as surrogates for disease modification[2,7,10,13,27].

Future studies should define PNI prospectively at enrollment or use PNI status for prespecified stratification. Standardized pathological or imaging endpoints should assess PNI burden alongside pain outcomes, local recurrence, survival, and treatment-related nerve dysfunction. Mechanism-based trials should specify whether the intended effect is symptom relief, local control, or alteration of PNI biology and should not substitute general tumor response for a PNI-specific endpoint. Prospective designs, external validation, and transparent reporting across disease stages and treatment contexts are required before experimental vulnerabilities can inform clinically useful interventions.

SPATIAL OMICS AND FUTURE DIRECTIONS

PNI is intrinsically spatial because its interpretation depends on the anatomical relationships among PDAC cells, nerves, Schwann cells, immune populations, stromal cells, and extracellular-matrix compartments. Bulk or dissociated assays can identify molecular features but cannot fully preserve these tumor-nerve-stromal relationships. In this article, spatial omics is used as an umbrella term encompassing spatial transcriptomic and multiplex protein-imaging approaches. Spatial studies are mapped onto the same four-level evidence framework used throughout the review: Direct human PNI spatial evidence[28] and human neural-invasion-associated cellular states[29] are distinguished from predominantly experimental neural injury/repair evidence[30], whereas general PDAC multiplex-imaging findings[31] are treated as contextual rather than PNI-specific. These approaches support characterization and hypothesis generation but have not established a clinically validated PNI risk score or treatment-selection tool.

Why spatial context matters in PNI

PNI develops at an anatomical interface where malignant glands or tumor-cell clusters encounter peripheral nerves and their associated Schwann, immune, and stromal compartments. The biological meaning of a transcript or cellular state may therefore depend on whether it is located within an invaded nerve, the adjacent cancer compartment, a non-invaded nerve, or tumor tissue distant from visible neural structures. Spatial transcriptomics preserves this regional information, allowing molecular measurements to be interpreted together with tissue morphology rather than after complete dissociation of the sampled tissue[28].

This design facilitates comparisons between PNI-associated and non-PNI compartments and assessment of whether molecular programs are restricted to the neural-invasion interface[28,29]. It can also position tumor, Schwann-cell, immune, and stromal states within their local anatomical context without implying that proximity alone demonstrates a directional mechanism. Spatial omics should therefore be viewed as a set of tools for characterization, cross-compartment comparison, and hypothesis generation. Causal relationships require orthogonal functional testing, and transcriptomic measurements should not be interpreted as concurrent measurements of protein abundance or activity.

Spatial transcriptomics of the neural-invasion niche

A spatial transcriptomic study of archival human PDAC tissue directly compared PNI and non-PNI regions while preserving tissue morphology and separating cancer and neural compartments for analysis[28]. By defining regions according to the presence or absence of neural invasion and evaluating the relevant compartments separately, the study provided greater anatomical resolution than bulk-tissue analysis. This compartment-aware design is important because signals detected in a nerve-containing region may originate from invaded neural tissue, adjacent cancer cells, or surrounding stromal and immune components.

The analysis identified differential programs related to axonal guidance, including roundabout guidance receptor-SLIT guidance ligand signaling, together with immune-context differences and transcriptional similarities to neural-injury responses in PNI-associated compartments[28]. Exploratory metabolic programs were also reported. These findings should be considered hypothesis-generating rather than evidence that a uniform metabolic state defines all PNI lesions. Similarly, spatial association does not establish that any single pathway initiates neural invasion or dominates its progression.

These observations are consistent with a neural-invasion niche in which anatomically adjacent tumor, neural, immune, and stromal compartments display coordinated but heterogeneous molecular features. The term “niche” is used here as an integrative framework for organizing spatial observations and guiding validation, not as a clinically validated molecular subtype. Current spatial evidence does not establish a routine biomarker, patient-outcome model, or therapeutic benefit from targeting any individual spatial program.

Neural injury, repair, and integrated cellular states

Spatial profiling and functional studies have also examined neural injury and repair in PDAC, predominantly in experimental models[30]. These investigations reported injury-associated changes in tumor-associated nerves, peripheral-neuroglial repair responses, and Schwann-cell proliferation or repair-like states. Macrophage-associated responses were also observed in the context of neural injury and cellular clearance[30]. This evidence provides mechanistic context for nerve damage and local remodeling but should not be interpreted as a uniformly validated human PNI program.

Complementary human evidence comes from an integrated single-cell/single-nucleus and spatial transcriptomic analysis of 62 samples from 25 patients with PDAC and differing neural-invasion status[29]. The study identified neural-invasion-associated tumor states and distinct Schwann-cell, fibroblast, macrophage, and immune-spatial states. Reported examples included transforming growth factor beta induced-positive Schwann cells at the leading edge of neural invasion, NLR family pyrin domain containing 3-positive macrophage and myofibroblastic cancer-associated fibroblast environments in samples with high neural invasion, and tertiary lymphoid structures near non-invaded nerves in samples with low neural invasion[29]. These observations refine the cellular description of neural-invasion regions without establishing that spatial proximity represents a directional causal interaction.

The two studies therefore occupy different but complementary evidence tiers. Reference[30] supports a predominantly experimental injury-and-repair framework involving peripheral neuroglia, whereas reference[29] provides human evidence of neural-invasion-associated cellular states and spatial relationships. Their findings are consistent with a model in which neural injury and remodeling coexist with cellular-state heterogeneity across the tumor-nerve interface. Functional perturbation, independent cohorts, and standardized spatial annotation remain necessary to determine which associations are reproducible, mechanistically consequential, and clinically informative.

Multiplex protein imaging and cellular neighborhoods

Multiplex protein imaging can complement spatial transcriptomics by evaluating protein-defined cell states while preserving tissue organization. In general PDAC tissue, imaging mass cytometry has enabled single-cell spatial analysis using multiplex antibody panels and has identified cellular neighborhoods, including features of the vascular niche[31]. This work demonstrates the capacity of imaging mass cytometry to resolve tumor, immune, stromal, and vascular organization in situ. However, reference[31] was not designed as a PNI-specific study, and its vascular and cellular-neighborhood findings should remain separate from direct evidence obtained in neural-invasion regions.

When applied to regionally annotated PNI specimens, multiplex protein imaging could test whether transcriptomically defined states correspond to protein expression, cell-cell proximity, and reproducible tissue architecture. Such studies may complement spatial transcriptomics by evaluating protein states and cellular neighborhoods that RNA measurements alone cannot fully resolve. PNI-specific protein-defined neighborhoods, however, remain to be established. General PDAC imaging findings should not be extrapolated to PNI-specific immunosuppression, pain, recurrence, survival prediction, or treatment selection without dedicated validation.

Validation priorities and clinically linked spatial studies

Future spatial studies should use region-matched PNI and non-PNI sampling with standardized pathological annotation of nerve invasion, neural and cancer compartments, and PNI burden. Where appropriate, spatial transcriptomics should be integrated with single-cell or single-nucleus profiling and multiplex protein imaging so that RNA-defined states, protein expression, and tissue neighborhoods can be compared across platforms. Independent cohorts and cross-platform validation will be required to distinguish reproducible PNI-associated features from cohort-specific or platform-dependent signals. Functional validation should then determine whether prioritized spatial associations contribute to neural invasion, injury, remodeling, or pain rather than merely reflecting proximity.

Clinical linkage should be prospective and prespecified. Spatial features may be evaluated in relation to pain phenotype, recurrence pattern, survival, quantitative PNI burden, and treatment-related endpoints, but each endpoint requires appropriate sampling, adjustment, and external validation. Associations with survival or neural-invasion status should not be interpreted as evidence that a spatial marker predicts treatment response. An integrated multimodal score remains a proposed research framework rather than a validated clinical tool, and no spatial biomarker has been established for routine PNI risk stratification.

The same evidence boundary applies to therapeutic translation. Spatial profiling may help prioritize candidate pathways or patient subsets for future investigation, but spatially guided treatment selection remains unproven, and clinical benefit from modifying the neural-invasion niche has not been demonstrated. Progress will depend on prospective outcome-linked cohorts, transparent evidence tiers, and experimental confirmation of candidate mechanisms before clinical testing. With these safeguards, spatial studies can refine the biological description of PNI and define testable validation priorities without implying an established precision-therapy pathway.

CONCLUSION

PNI in PDAC is an active, spatially organized invasive process rather than solely a histopathological finding. The neural-immune-stromal-metabolic niche framework integrates interactions among tumor cells, nerves, Schwann cells, immune cells, CAFs, and the extracellular matrix. Within this framework, neurotrophic tumor-nerve convergence, Schwann-cell remodeling, immune-stromal and matrix feedback, metabolic and RNA regulation, and neural signaling contribute at different evidentiary levels rather than as a uniformly established causal sequence. Pain likewise reflects overlapping mechanisms, combining structural nerve infiltration and compression with chemical and neuroimmune sensitization; PNI burden alone cannot explain the full heterogeneity of pain in PDAC.

Histopathology remains the diagnostic reference for PNI, whereas quantitative grading, radiomics, AI-assisted pathology, and intraoperative imaging occupy different stages of validation and clinical maturity. PNI provides prognostic information, but no integrated diagnostic or risk-stratification tool has been prospectively established for routine care. Moreover, no PNI-specific disease-modifying therapy has been clinically validated. Current therapeutic concepts are primarily mechanistic, preclinical, or indirect, and symptom-directed pain management-including analgesic and selected nerve-directed interventions-must remain distinct from interventions intended to alter PNI biology, recurrence, or survival.

Spatial transcriptomics and multiplex imaging provide complementary approaches to characterizing the neural-invasion niche while preserving anatomical context; however, their use in PNI remains research-stage, and the niche is not a validated clinical subtype. Progress will require standardized PNI annotation, prospective and independent cohorts, functional and cross-platform validation, and PNI-specific endpoints linked separately to pain, recurrence, survival, and treatment response. A mechanistically integrated and spatially resolved approach can sharpen the questions asked of PNI, but clinically meaningful stratification or intervention will require rigorous prospective validation rather than biological plausibility alone.

ACKNOWLEDGEMENTS

The authors thank all individuals who provided assistance during the preparation of this article.

References
1.  Lovecek M, Dirimtekin E, Garajová I, Gasparini G, Crippa S, Giovannetti E, Sochorova D, Reyes CM, Demir IE, Uysal-Onganer P; TRANSPAN PNI Group. Perineural Invasion in Pancreatic Ductal Adenocarcinoma: Recapitulating Its Importance and Defining Future Directions. United European Gastroenterol J. 2025;13:1678-1689.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 1]  [Cited by in RCA: 6]  [Article Influence: 6.0]  [Reference Citation Analysis (0)]
2.  Sun Y, Jiang W, Liao X, Wang D. Hallmarks of perineural invasion in pancreatic ductal adenocarcinoma: new biological dimensions. Front Oncol. 2024;14:1421067.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in RCA: 19]  [Reference Citation Analysis (0)]
3.  Crippa S, Pergolini I, Javed AA, Honselmann KC, Weiss MJ, Di Salvo F, Burkhart R, Zamboni G, Belfiori G, Ferrone CR, Rubini C, Yu J, Gasparini G, Qadan M, He J, Lillemoe KD, Castillo CF, Wolfgang CL, Falconi M. Implications of Perineural Invasion on Disease Recurrence and Survival After Pancreatectomy for Pancreatic Head Ductal Adenocarcinoma. Ann Surg. 2022;276:378-385.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 34]  [Cited by in RCA: 101]  [Article Influence: 25.3]  [Reference Citation Analysis (2)]
4.  Felsenstein M, Lindhammer F, Feist M, Hillebrandt KH, Timmermann L, Benzing C, Globke B, Zocholl D, Hu M, Fehrenbach U, Sinn BV, Pelzer U, Sauer IM, Pratschke J, Malinka T. Perineural Invasion in Pancreatic Ductal Adenocarcinoma (PDAC): A Saboteur of Curative Intended Therapies? J Clin Med. 2022;11:2367.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 15]  [Cited by in RCA: 31]  [Article Influence: 7.8]  [Reference Citation Analysis (0)]
5.  Jiang L, Cai S, Weng Z, Zhang S, Jiang SH. Peripheral, central, and chemotherapy-induced neuropathic changes in pancreatic cancer. Trends Neurosci. 2025;48:124-139.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 7]  [Cited by in RCA: 26]  [Article Influence: 26.0]  [Reference Citation Analysis (0)]
6.  Zhang J, Fu X, Liu D, Yang M, Yang J, Huo Y, Liu W, Hua R, Sun Y, Wang J. Molecular markers associated with perineural invasion in pancreatic ductal adenocarcinoma. Oncol Lett. 2020;20:5.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 4]  [Cited by in RCA: 9]  [Article Influence: 1.5]  [Reference Citation Analysis (0)]
7.  Selvaggi F, Melchiorre E, Casari I, Cinalli S, Cinalli M, Aceto GM, Cotellese R, Garajova I, Falasca M. Perineural Invasion in Pancreatic Ductal Adenocarcinoma: From Molecules towards Drugs of Clinical Relevance. Cancers (Basel). 2022;14:5793.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 47]  [Cited by in RCA: 45]  [Article Influence: 11.3]  [Reference Citation Analysis (1)]
8.  Li J, Kang R, Tang D. Cellular and molecular mechanisms of perineural invasion of pancreatic ductal adenocarcinoma. Cancer Commun (Lond). 2021;41:642-660.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 106]  [Cited by in RCA: 115]  [Article Influence: 23.0]  [Reference Citation Analysis (1)]
9.  Zhang B, Guo X, Huang L, Zhang Y, Li Z, Su D, Lin L, Zhou P, Ye H, Lu Y, Zhou Q. Tumour-associated macrophages and Schwann cells promote perineural invasion via paracrine loop in pancreatic ductal adenocarcinoma. Br J Cancer. 2024;130:542-554.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 53]  [Cited by in RCA: 63]  [Article Influence: 31.5]  [Reference Citation Analysis (1)]
10.  Yang MW, Tao LY, Jiang YS, Yang JY, Huo YM, Liu DJ, Li J, Fu XL, He R, Lin C, Liu W, Zhang JF, Hua R, Li Q, Jiang SH, Hu LP, Tian GA, Zhang XX, Niu N, Lu P, Shi J, Xiao GG, Wang LW, Xue J, Zhang ZG, Sun YW. Perineural Invasion Reprograms the Immune Microenvironment through Cholinergic Signaling in Pancreatic Ductal Adenocarcinoma. Cancer Res. 2020;80:1991-2003.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 23]  [Cited by in RCA: 154]  [Article Influence: 25.7]  [Reference Citation Analysis (8)]
11.  Li T, Hu C, Huang T, Zhou Y, Tian Q, Chen H, He R, Yuan Y, Jiang Y, Jiang H, Huang K, Cheng D, Chen R, Zheng S. Cancer-Associated Fibroblasts Foster a High-Lactate Microenvironment to Drive Perineural Invasion in Pancreatic Cancer. Cancer Res. 2025;85:2199-2217.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 65]  [Cited by in RCA: 72]  [Article Influence: 72.0]  [Reference Citation Analysis (1)]
12.  Schiavo Lena M, Gasparini G, Crippa S, Belfiori G, Aleotti F, Di Salvo F, Redegalli M, Cangi MG, Taveggia C, Falconi M, Doglioni C. Quantification of perineural invasion in pancreatic ductal adenocarcinoma: proposal of a severity score system. Virchows Arch. 2023;483:225-235.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in RCA: 8]  [Reference Citation Analysis (0)]
13.  Lian EY, Hyndman BD, Moodley S, Maritan SM, Mulligan LM. RET isoforms contribute differentially to invasive processes in pancreatic ductal adenocarcinoma. Oncogene. 2020;39:6493-6510.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 7]  [Cited by in RCA: 18]  [Article Influence: 3.0]  [Reference Citation Analysis (0)]
14.  He Y, Chen Z, Yang L, Qiao S, Su Z, Ding F, Ding F, Xin F, Xiang S, Lyu A, Li F. The supporting role of Schwann cells in perineural invasion of pancreatic ductal adenocarcinoma. Front Pharmacol. 2025;16:1540027.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in RCA: 5]  [Reference Citation Analysis (0)]
15.  Na'ara S, Amit M, Gil Z. L1CAM induces perineural invasion of pancreas cancer cells by upregulation of metalloproteinase expression. Oncogene. 2019;38:596-608.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 37]  [Cited by in RCA: 95]  [Article Influence: 11.9]  [Reference Citation Analysis (1)]
16.  Yang Y, Chen Y, Wang X, Sun Q, Qian Y. Association of LAMA3 expression with perineural invasion and chemoresistance in pancreatic ductal adenocarcinoma. Discov Oncol. 2025;16:2240.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in RCA: 1]  [Reference Citation Analysis (0)]
17.  Li T, He R, Tian Q, Huang T, Jiang H, Chen H, Yuan Y, Jiang Y, Zheng S, Hu C, Li S, Li G, Chen R. Cancer-associated fibroblast-derived circKLHL24 drives perineural invasion in pancreatic cancer via dual regulation of the sec31a-CXCL12 axis. J Exp Clin Cancer Res. 2025;44:281.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in RCA: 8]  [Reference Citation Analysis (0)]
18.  Huang T, Hu C, Chen H, Jiang H, Li T, Tian Q, He R, Yuan Y, Jiang Y, Zhou Y, Lin Q, Li Z, Xiao M, Wei X, Chen R, Zheng S. Lactylation-driven NSUN2-mediated RNA m5C modification promotes perineural invasion in pancreatic cancer. Theranostics. 2026;16:1782-1803.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in RCA: 10]  [Reference Citation Analysis (0)]
19.  Hung YH, Wang HC, Hsu SH, Wang LY, Tsai YL, Su YY, Hung WC, Chen LT. Neuron-derived neurotensin promotes pancreatic cancer invasiveness and gemcitabine resistance via the NTSR1/Akt pathway. Am J Cancer Res. 2024;14:448-466.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 14]  [Cited by in RCA: 11]  [Article Influence: 5.5]  [Reference Citation Analysis (0)]
20.  Shen S, Wang Q, Wang X, Ding J, Chen F, Xiao Y, Qin T, Qian W, Li J, Ma Q, Ma J. Nodal Enhances Perineural Invasion in Pancreatic Cancer by Promoting Tumor-Nerve Convergence. J Healthc Eng. 2022;2022:9658890.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 12]  [Cited by in RCA: 13]  [Article Influence: 3.3]  [Reference Citation Analysis (2)]
21.  Chen Y, Zhang W, Zeng Y, Yang P, Li Y, Liang X, Liu K, Lin H, Dai Y, Zhou J, Hou B, Ma Z, Lin Y, Pang W, Zeng L. GDNF-induced phosphorylation of MUC21 promotes pancreatic cancer perineural invasion and metastasis by activating RAC2 GTPase. Oncogene. 2024;43:2564-2577.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 22]  [Cited by in RCA: 27]  [Article Influence: 13.5]  [Reference Citation Analysis (0)]
22.  Wang K, Ni B, Xie Y, Li Z, Yuan L, Meng C, Zhao T, Gao S, Huang C, Wang H, Ma Y, Zhou T, Feng Y, Chang A, Yang C, Yu J, Yu W, Zang F, Zhang Y, Ji RR, Wang X, Hao J. Nociceptor neurons promote PDAC progression and cancer pain by interaction with cancer-associated fibroblasts and suppression of natural killer cells. Cell Res. 2025;35:362-380.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 110]  [Cited by in RCA: 81]  [Article Influence: 81.0]  [Reference Citation Analysis (0)]
23.  Noda Y, Pisuchpen N, Parakh A, Srinivas-Rao S, Kinowaki Y, Mino-Kenudson M, Kambadakone AR. Does CT overestimate extra-pancreatic perineural invasion in patients with pancreatic ductal adenocarcinoma following neoadjuvant chemoradiation therapy? Br J Radiol. 2024;97:607-613.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in RCA: 4]  [Reference Citation Analysis (0)]
24.  Deng Y, Yu H, Duan X, Liu L, Huang Z, Song B. A CT-based radiomics nomogram for the preoperative prediction of perineural invasion in pancreatic ductal adenocarcinoma. Front Oncol. 2025;15:1525835.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in RCA: 5]  [Reference Citation Analysis (0)]
25.  Borsekofsky S, Tsuriel S, Hagege RR, Hershkovitz D. Perineural invasion detection in pancreatic ductal adenocarcinoma using artificial intelligence. Sci Rep. 2023;13:13628.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in RCA: 17]  [Reference Citation Analysis (0)]
26.  Lu WL, Kuang H, Gu J, Hu X, Chen B, Fan Y. GAP-43 targeted indocyanine green-loaded near-infrared fluorescent probe for real-time mapping of perineural invasion lesions in pancreatic cancer in vivo. Nanomedicine. 2023;50:102671.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in RCA: 7]  [Reference Citation Analysis (0)]
27.  Renz BW, Takahashi R, Tanaka T, Macchini M, Hayakawa Y, Dantes Z, Maurer HC, Chen X, Jiang Z, Westphalen CB, Ilmer M, Valenti G, Mohanta SK, Habenicht AJR, Middelhoff M, Chu T, Nagar K, Tailor Y, Casadei R, Di Marco M, Kleespies A, Friedman RA, Remotti H, Reichert M, Worthley DL, Neumann J, Werner J, Iuga AC, Olive KP, Wang TC. β2 Adrenergic-Neurotrophin Feedforward Loop Promotes Pancreatic Cancer. Cancer Cell. 2018;33:75-90.e7.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 242]  [Cited by in RCA: 427]  [Article Influence: 53.4]  [Reference Citation Analysis (0)]
28.  Lakis V, Chan NL, Lyons R, Blackburn N, Nguyen TH, Chang C, Masel A, West NP, Boyle GM, Patch AM, Gill AJ, Nones K. Spatial Transcriptomics Reveals Novel Mechanisms Involved in Perineural Invasion in Pancreatic Ductal Adenocarcinomas. Cancers (Basel). 2025;17:852.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in RCA: 10]  [Reference Citation Analysis (0)]
29.  Chen MM, Gao Q, Ning H, Chen K, Gao Y, Yu M, Liu CQ, Zhou W, Pan J, Wei L, Dou W, Zhang D, Zhu L, Zhang Q, Chen R, Zhang Z. Integrated single-cell and spatial transcriptomics uncover distinct cellular subtypes involved in neural invasion in pancreatic cancer. Cancer Cell. 2025;43:1656-1676.e10.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 43]  [Cited by in RCA: 94]  [Article Influence: 94.0]  [Reference Citation Analysis (3)]
30.  Weitz J, Garg B, Martsinkovskiy A, Patel S, Tiriac H, Lowy AM. Pancreatic ductal adenocarcinoma induces neural injury that promotes a transcriptomic and functional repair signature by peripheral neuroglia. Oncogene. 2023;42:2536-2546.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 2]  [Cited by in RCA: 27]  [Article Influence: 9.0]  [Reference Citation Analysis (0)]
31.  Sussman JH, Kim N, Kemp SB, Traum D, Katsuda T, Kahn BM, Xu J, Kim IK, Eskandarian C, Delman D, Beatty GL, Kaestner KH, Simpson AL, Stanger BZ. Multiplexed Imaging Mass Cytometry Analysis Characterizes the Vascular Niche in Pancreatic Cancer. Cancer Res. 2024;84:2364-2376.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 23]  [Cited by in RCA: 26]  [Article Influence: 13.0]  [Reference Citation Analysis (0)]
Footnotes

Peer review: Externally peer reviewed.

Peer-review model: Single blind

Corresponding Author's Membership in Professional Societies: Chinese Medical Association, No. 51100000500004737L.

Specialty type: Oncology

Country of origin: China

Peer-review report’s classification

Scientific quality: Grade B, Grade B

Novelty: Grade B, Grade C

Creativity or innovation: Grade A, Grade B

Scientific significance: Grade A, Grade C

P-Reviewer: Yi JF, Associate Professor, MD, China S-Editor: Bai Y L-Editor: A P-Editor: Zhao YQ

Write to the Help Desk