Microplastics and nanoplastics-related genes signature predicts prognosis in pancreatic ductal adenocarcinoma and functional validation of interleukin 1 alpha

Abstract

Background

Microplastics and nanoplastics (MNPs), as emerging environmental pollutants, have garnered significant attention from the global scientific community due to their potential threats to human health, particularly their association with the occurrence and development of cancer. The goal of our study is to create a predictive marker for pancreatic ductal adenocarcinoma (PAAD) based on MNPs-related genes, with the purposes of predicting survival outcomes and assessing the tumor immune microenvironment.

Methods

Using multi-cohort data from The Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), and International Cancer Genome Consortium (ICGC), we assessed the association between MNPs and PAAD prognosis through the Xiantao Academic (https://www.xiantao.love/). The development of a prognostic signature was followed by an assessment of its significance through the Kaplan-Meier method, time-dependent receiver operating characteristic (ROC), and decision curve analysis (DCA). The validity of the risk model was confirmed through the ICGC and GSE71729 cohorts. The model was then assessed for levels of tumor immune infiltration. To explore MNPs-related genes expression characteristics within immune cells in PAAD, we performed single-cell RNA sequencing and spatial transcriptomics analysis through the Sparkle Platform (https://grswsci.top/). Finally, in vitro experiments were conducted to investigate the biological function of interleukin 1 alpha (IL1A).

Results

A four-gene signature comprising XDH, IL1A, KIF20A, and ASPM, based on MNPs, was developed to stratify PAAD patients into two distinct risk groups. The high-risk group showed a significantly poorer prognosis. A similar trend was verified in the external cohorts ICGC and GSE71729. The signature risk score affected immune cell infiltration in the PAAD microenvironment. The infiltration of B cells, CD8⁺ T cells, cytotoxic cells, immature dendritic cells (iDCs), mast cells, plasmacytoid dendritic cell (pDC), T cells, Tem cells, T follicular helper (TFH) cells, and T helper 17 (Th17) cells had a positive correlation with the low-risk group. In contrast, high-risk patients tended to have increased number of T helper (Th2) cells and higher expression of SIGLEC15, CD274, IGSF8. Knockdown of IL1A in PAAD cells inhibited their tumor proliferation ability in vitro.

Conclusions

Using MNPs-related genes, we built a prognostic model for PAAD, revealing that patients with high-risk scores are likely to have a worse prognosis. This model is designed to develop personalized treatment strategies tailored to the specific needs of each patient, thereby improving clinical outcomes for PAAD patients. Furthermore, IL1A could be a promising therapeutic candidate for PAAD.

Highlight box.

Key findings

• Multi-omics integration analysis provides a powerful framework for microplastics and nanoplastics (MNPs)-related genes in pancreatic ductal adenocarcinoma (PAAD).

What is known and what is new?

• MNPs induce oxidative stress, chronic inflammation, and tissue fibrosis. Sustained exposure to MNPs is associated with an elevated risk of developing chronic diseases and malignancies.

• The MNPs-related genes score may have the potential to improve prognostic assessment for PAAD and could offer preliminary insights for individualized treatment approaches, though further clinical validation is required.

What is the implication, and what should change now?

• The IL1A gene emerged as a candidate gene potentially involved in the response of PAAD to MNPs, warranting further validation.

Introduction

Pancreatic ductal adenocarcinoma (PAAD), a highly fatal malignancy of the digestive system, has a 5-year survival rate that has remained stagnant at below 10% over the past five decades (1,2). This is largely because of the insidious onset of this malignancy, its strong invasive potential, and the absence of effective therapeutic strategies (1,2). Although surgical resection, chemotherapy, immunotherapy and targeted therapy may improve the outcome of selected patients, rapid recurrence due to the development of resistance to therapy is still common in clinical practice while reliable biomarkers to predict drug resistance are still lacking (3). Therefore, a comprehensive investigation of the molecular pathological mechanisms of PAAD and the identification of new prognostic biomarkers would hold substantial clinical value for optimizing personalized treatment strategies.

In recent years, increasing attention has been directed toward the association between environmental pollutant exposure and tumorigenesis (4). As a class of emerging persistent environmental contaminants, Microplastics and nanoplastics (MNPs) can enter the human body through multiple routes, including food, drinking water, and inhalation (5). MNPs have been detected in various human tissues and organs such as blood, liver, and placenta (6,7). Previous studies suggested that MNPs cause oxidative stress, chronic inflammation, and tissue fibrosis (8-10). Prolonged exposure to MNPs increases the risk of the development of chronic diseases and malignancies (11). However, the contribution of MNPs to PAAD progression remains poorly characterized and has not been systematically investigated.

Multi-omics integration provides a powerful framework for addressing this question. At the macro level, bibliometric analysis was employed to delineate research trends in MNPs and cancer, providing an academic foundation for subsequent gene selection. Transcriptomic data, single-cell sequencing, spatial transcriptomics, and in vitro functional assays were then integrated to systematically characterize the expression profiles and prognostic significance of MNPs-related genes in PAAD. By incorporating environmental exposure factors with multi-omics datasets, this study aims to provide a novel “environment-tumor” interaction perspective for elucidating PAAD pathogenesis. This approach not only advances the understanding of environmental carcinogenesis but also offers new insights into precision diagnosis and treatment strategies for patients with PAAD. We present this article in accordance with the TRIPOD and MDAR reporting checklists (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1405/rc).

Methods

Data collection

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. RNA-Seq transcriptomic data (TPM format) from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) projects, uniformly processed using the Toil pipeline, were obtained from UCSC Xena (https://xenabrowser.net/datapages/) (12). Dataset GSE62165 was retrieved from the Gene Expression Omnibus (GEO) database. For external validation, cohorts with compatible expression profiles and survival information were selected, including GSE71729 from GEO and datasets from International Cancer Genome Consortium (ICGC). MNPs-related genes were extracted from the Comparative Toxicogenomics Database (CTD) (https://ctdbase.org/) and further details are available in table online: https://cdn.amegroups.cn/static/public/tcr-2026-1405-1.xlsx. Xiantao Academic (https://www.xiantao.love/ was used for statistical analysis and visualization. Missing or indeterminate clinical information for individual participants was treated as missing data. The overall study workflow is presented in Figure 1.

Figure 1.

Flowchart of the study. CTD, Comparative Toxicogenomics Database; DEGs, differentially expressed genes; ICGC, International Cancer Genome Consortium; MNPs, microplastics and nanoplastics; PAAD, pancreatic ductal adenocarcinoma; PMNPs, MNPs-related genes in PAAD; TCGA, The Cancer Genome Atlas.

Bibliometrics analysis of MNPs in cancer

This study systematically curated literature on MNPs in cancer from January 1, 2017, to December 31, 2025, using the Web of Science Core Collection (WoSCC). The search strategy was defined as (TS=(“microplastic*” OR “nanoplastic*”)) AND (TS=(“tumor*” OR “cancer*” OR “neoplasm*” OR “malignancy*”)). Only original articles and reviews in English were included, and duplicates were removed manually. Relevant topic terms were further mapped to MeSH terms in the NCBI database. Countries/regions collaboration networks and keyword co-occurrence patterns were analyzed using VOSviewer 1.6.20 (13). Publication trends and the top 10 most frequent keywords were visualized using Microsoft Office Excel 2019.

Identification of MNPs-related genes in PAAD (PMNPs)

In CTD, 165 genes related to MNPs were found using the “microplastics” or “nanoplastics” search terms. The DESeq2 package was employed to determine differentially expressed genes (DEGs) in PAAD compared to normal tissues, using criteria of adjusted P<0.05 and |log₂ fold change| >2 (14). Nineteen overlapping genes were identified through Venn diagram analysis and defined as “PMNPs”.

Establishment and validation of the PMNPs signature

A univariate Cox analysis of overall survival based on PMNPs was performed using the TCGA cohort, and four PMNPs were subsequently identified via ten‑fold cross‑validation combined with least absolute shrinkage and selection operator (LASSO) regression, which were then used to build a prognostic risk signature. The formula used for calculating the risk score is: risk score = Σ(Expi × coefi), with Expi indicating the levels of gene expression and coefi signifying the associated coefficients (15). Patients in the TCGA training cohort (n=177; alive =84, dead =93) were classified into high- and low-risk groups based on the median risk score. To assess outcomes between these groups, Kaplan-Meier survival analysis and log-rank tests were conducted (16). Using the calculated risk scores, we developed a nomogram based on PMNPs for disease risk prediction. We further assessed the independent prognostic value of the risk score and clinical features via univariate and multivariate regression analyses. Furthermore, the model’s predictive capability was evaluated using time-dependent receiver operating characteristic (ROC) curve and decision curve analysis (DCA) (17). The prognostic signature was further validated in the GSE71729 (n=123; alive =40, dead =83) and ICGC cohorts (n=267; alive =106, dead =161). Patients with missing survival data were excluded from this study.

The analysis of immune cell infiltration

Immune scores at the individual level were calculated using single-sample gene set enrichment analysis (ssGSEA) to quantify immune infiltration (18). In addition, the CIBERSORT algorithm was used to deconvolute bulk transcriptomic data from low- and high-risk groups, estimating the relative proportions of tumor-infiltrating immune cell subsets within the tumor microenvironment (19). Finally, potential associations between the PMNPs signature and immune checkpoint-related genes were further explored.

Single-cell analysis

Through the Sparkle platform at https://grswsci.top/, the dataset PAAD_CRA001160 (n=35) and GSE111672 (n=33) were analyzed for multi-omics correlations. This analysis was completed using the TISCH2 database at http://tisch.comp-genomics.org/ (20). Through the AUCell algorithm (21), PMNPs activity score of each sample was computed from the area under the curve (AUC) values of gene ranking distribution. All the genes were ranked according to expression levels in each cell and the PMNPs enrichment was evaluated within the top 5% of the ranked gene list to calculate the respective AUC scores.

Spatial transcriptome analysis

To characterize the spatial transcriptomic landscape of PAAD, analyses of GSE203612-GSM6177618, GSE211895-GSM6505133, GSE211895-GSM6505134 and GSE211895-GSM6505135 were conducted using both the Sparkle database and the SpatialTME platform (https://www.spatialtme.yelab.site/) (22). Cellular deconvolution of the tumor microenvironment was performed using the “Cottrazm” R package obtained from the SpatialTME repository (23,24). In addition, integration of 10× Visium datasets within the Sparkle framework enabled construction of pan-cancer spatial transcriptomic maps. Gene expression patterns and dominant cellular distributions across spatial microregions were visualized using the SpatialFeaturePlot and SpatialDimPlot functions from the Seurat package (23,24). Malignancy scores assigned to each spatial region were interpreted as follows: a score of 1 indicated a malignant region and 0 indicated normal tissue. Wilcoxon rank-sum tests were applied to compare PMNPs expression differences between the two groups. Spearman correlation analysis was further performed to evaluate associations between cell composition and gene expression across spatial spots, and results were visualized using the linkET R package (25).

Cell culture, polystyrene treatment, and transfection

The CFPAC-1, MIA-PaCa-2, CAPAN-1 and BXPC-3 pancreatic cancer cell lines were procured from Savaire Biological Technology Co., The HPDE6-C7 normal human pancreatic ductal epithelial cell line was sourced from the BeNa Culture Collection located in Beijing, China. The BXPC-3, MIA-PaCa-2 and HPDE6-C7 cells were cultured in DMEM (Gibco, New York, USA) containing 10% fetal bovine serum and 1% penicillin-streptomycin (Servicebio, Wuhan, China). CAPAN-1 cells were maintained in IMDM containing 20% FBS and 1% penicillin-streptomycin (Servicebio, Wuhan, China), while CFPAC-1 cells were cultured in the same medium supplemented with 10% FBS and 1% antibiotics. A humidified atmosphere containing 5% CO₂ at 37 ℃ was used to maintain the cell lines. CAPAN-1 cells were seeded at 5×10⁵ cells per well in 6-well plates and transfected with IL1A-targeted shRNA. sh-IL1A-1, sh-IL1A-2 and non-targeting control shRNA (sh-NC) were acquired from SuZhou GenePharma Co., Ltd. To account for non-specific shRNA effects, cells were transfected with either sh-IL1A-1, sh-IL1A-2, or sh-NC; the latter possesses no sequence homology to IL1A mRNA and thus served as the statistical baseline. Transfections were conducted according to the manufacturer’s guidelines after cells had attached.

Polystyrene microspheres were obtained from Aladdin (Cat. No. P107085-500g, Shanghai, China). Prior to cell exposure, polystyrene microspheres were ultrasonicated and diluted in serum-free medium to final concentrations of 0–20 µg/mL, which were chosen based on previously published protocols (26,27). After 48 hours of polystyrene treatment, total cellular proteins were extracted, and IL1A protein expression levels were detected by Western blotting.

Western blot

RIPA lysis buffer with 1% PMSF was used to extract total protein, and protein levels were measured with a BCA kit (Servicebio, Wuhan, China). Proteins in equal quantities were resolved on 10% SDS-PAGE gels and then transferred to 0.45 µm PVDF membranes (Millipore, Massachusetts, USA). Non-fat milk was used to block the membranes at room temperature for 3 hours, followed by three TBST washes. Subsequently, membranes were incubated overnight at 4 ℃ with primary antibodies, including IL1A (cat. no. 84411-5-RR, 1:20,000, Proteintech, Wuhan, China) and β-tubulin (cat. no. 80713-1-RR, 1:50,000, Proteintech, Wuhan, China). After TBST washing, membranes were treated for 1 hour at room temperature with HRP-conjugated AffiniPure goat anti-rabbit IgG secondary antibody (cat. no. SA00001-2, 1:5,000, Proteintech, Wuhan, China). Signal detection was performed using an enhanced chemiluminescence kit (cat. no. BL523A, Biosharp, Anhui, China).

Cell Counting Kit-8 (CCK-8) assay

The viability of the cells was assessed using CCK-8 assay. CAPAN-1 and BXPC-3 cells were plated at 5×10³ cells/well in 96-well plates in the logarithmic growth phase, and incubated for 5 hours before being transfected with shRNA. After 24 and 48 hours, medium was changed; then, the cells were incubated in complete medium containing 10% CCK-8 reagent (C0037, Beyotime, Shanghai, China) for 2 hours. The absorbance was measured at 450 nm with a Synergy NEO₂ microplate reader (Agilent, USA). To calculate viability, the formula (experimental group-blank group)/(control group-blank group) ×100% was used. Each experimental set comprised six replicate wells and was conducted across three independent trials.

Cell colony formation assay

In the colony formation assay, CAPAN-1 and BXPC-3 cells during their logarithmic growth phase were placed into 6-well plates with a concentration of 1,000 cells per well (2 mL). Three days later, the medium was changed and the experiment terminated on day 10. The cells were rinsed twice with phosphate-buffered saline (PBS), then fixed in 4% paraformaldehyde for 30 minutes, and stained with crystal violet for an additional 30 minutes. The surplus stain was eliminated by rinsing with ultrapure water, and the plates were left to air-dry. Images captured on Thermo Fisher live cell imaging system were analyzed using ImageJ software. All the experiments were independently performed in triplicate. The experiments were performed using independent biological replicates.

Transwell invasion assays

After trypsinization, CAPAN-1 and BXPC-3 cells were resuspended in a medium without serum and counted. The Transwell inserts’ upper chamber was filled with 200 µL of a cell suspension prepared at a density of 1×10⁵ cells/mL. For the purpose of the invasion assay, Matrixgel (catalog number 354234, Corning, New York, USA) was applied as a pre-coating on the inserts. Before seeding the cells, a solution of 1 mg/mL Matrigel in serum-free medium was added to the upper chamber. A medium with 20% FBS, totaling 700 µL, was used to supplement the lower chamber. After incubating for 48 hours, the inserts were rinsed with PBS, fixed using 4% paraformaldehyde for 10 minutes, and then stained with 1% crystal violet. A sterile swab was used to remove all the cells left on the upper surface. An inverted microscope was used to visualize the cells on the lower membrane. Five fields were randomly selected to count the cells that had migrated or invaded. Independent biological experiments were repeated three times.

Flow cytometric analysis of apoptosis

Cells cultured in 6-well plates were harvested and gently rinsed with PBS, followed by staining with the Annexin V-APC/7-AAD Apoptosis Detection Kit (cat. no. AP105, MultiSciences, Hanzhou, China) according to the manufacturer’s protocol. After staining, the cells were incubated for 15 min at room temperature protected from light and then subjected to flow cytometric analysis. For gating, the major cell population was first defined on the FSC-A versus SSC-A plot to exclude cellular debris and non-cellular particles. Single cells were subsequently selected using the FSC-A versus FSC-H plot to remove doublets and cell aggregates. For fluorescence compensation, apoptosis-induced positive cells were mixed with viable cells at an equal ratio and then prepared as unstained, Annexin V-APC single-stained, and 7-AAD single-stained controls. Data were analyzed using FlowJo 10 software. The apoptotic cell population was calculated as the combined percentage of Annexin V-positive/7-AAD-negative early apoptotic cells and Annexin V-positive/7-AAD-positive late apoptotic cells. Independent biological experiments were repeated three times.

Statistical analysis

Quantitative results are derived from three independent experiments and reported as mean ± standard error of the mean (SEM). GraphPad Prism 9 was used for statistical analysis of experimental data, while bioinformatics data were processed via the Xiantao Academic and Sparkle platforms. Continuous variables were compared using t-tests or Wilcoxon rank-sum tests, whereas Chi-squared tests were employed for categorical variables. Using Spearman’s rank correlation test, the analyses of correlation were conducted, with statistical significance set at a P value of less than 0.05.

Results

The overall trend of publications on MNPs in cancer

A total of 477 publications were retrieved from the WoSCC, of which 416 original articles and reviews were included for subsequent analysis and visualization. The annual number of publications remained below 10 between 2017 and 2019. After 2020, a marked increase in publication output was observed, indicating growing scientific interest in MNPs-related cancer research (Figure 2A). Trend modeling of annual publication output demonstrated a strong correlation between year and publication volume (R²=0.9787; Figure 2A). Collaborative network analysis of countries/regions revealed that China contributed the highest number of publications (Figure 2B). The temporal distribution of the top 10 keywords is shown in Figure 2C, and keyword co-occurrence density mapping is presented in Figure 2D.

Figure 2.

Bibliometrics analysis of MNPs in cancer. (A) The annual publication trend of MNPs in cancer. (B) An overlay map of publications in different countries/regions. (C) The trend of annual occurrences of the top 10 keywords. (D) An overlay map of the co-occurrence analysis of keywords. MNPs, microplastics and nanoplastics.

Identification of PMNPs in PAAD

A total of 2871 DEGs were identified between PAAD and normal tissues from the TCGA dataset, and 844 DEGs were obtained from GSE62165. In addition, 165 MNPs-related genes were retrieved from the CTD database. The relationship between MNPs-related genes and PAAD-related genes was investigated using a Venn intersection analysis, which identified 19 overlapping genes, defined as PMNPs (Figure 3A). The expression profile of these 19 PMNPs is shown in the heatmap, and most genes exhibited upregulated expression in PAAD samples (Figure 3B). Chromosomal localization of the 19 PMNPs is presented in Figure 3C, and the interaction network illustrates their expression patterns (Figure 3D).

Figure 3.

Differentially expressed PMNPs in PAAD. (A) The Venn diagram of screening PMNPs. (B) Heatmap of 19 differential expression of PMNPs between tumor and normal tissues. (C) The genome location of PMNPs on 23 human chromosomes. (D) Correlation analysis for PMNPs. MNPs, microplastics and nanoplastics; PAAD, pancreatic ductal adenocarcinoma; PMNPs, MNPs-related genes in PAAD.

Development and verification of a PMNPs-based risk signature

To evaluate the prognostic significance of PMNPs in PAAD, a prognostic signature was constructed using TCGA-PAAD as the training cohort, with ICGC and GSE71729 datasets serving as validation cohorts. In the training set, a univariate Cox regression analysis was used to analyse the correlation of overall survival with the nineteen PMNPs in PAAD (Figure 4A) and the LASSO regression method was used to optimize the model and assess regression coefficients (Figure 4B,4C). A four-gene PMNPs signature was established, comprising XDH, IL1A, KIF20A, and ASPM. The risk score was calculated as follows: risk score = (0.2868 × KIF20A expression) + (0.1281 × ASPM expression) + (0.0718 × XDH expression) + (0.0567 × IL1A expression). Using the median risk score, patients in the TCGA cohort were separated into high- and low-risk groups. Elevated expression of XDH, IL1A, KIF20A, and ASPM was associated with the high-risk group and poor prognosis (Figure 4D). According to Kaplan-Meier analysis, overall survival was significantly extended in the low-risk group compared to the high-risk group (Figure 4E). The time-dependent ROC analysis validated the model’s prognostic accuracy, showing AUCs of 0.714 (0.614–0.814) for 1 year and 0.773 (0.670–0.875) for 3 years, respectively (Figure 4F). To assess incremental performance, we added the four-gene score to a clinical baseline model [including tumor-node-metastasis (TNM) stage] in the TCGA-PAAD cohort, which showed that the four-gene score remained an independent prognostic factor [hazard ratio (HR) =2.079, 95% confidence interval (CI): 1.364–3.167, P<0.001] (Figure 5A). Subsequently, a nomogram was developed based on four-gene score and TNM stage with a C-index of 0.671 (0.640–0.702) (Figure 5B). The calibration plots showed close agreement between predicted and observed OS rates at 1 and 3 years (Figure 5C). The DCA also suggested this model might be useful in clinical settings to help improve PAAD patients’ overall survival (Figure 5D,5E).

Figure 4.

Construction of the prognostic model associated with PMNPs. (A) Univariate Cox regression analysis of the 19 PMNPs in PAAD from TCGA. (B) Cross validation for tuning parameter screening in the LASSO regression model using 19 PMNPs. (C) The coefficient profiles of the LASSO regression model. (D) Visualization of risk score distribution and survival status within the TCGA cohort. (E) Kaplan-Meier curves of patients with low and high PMNPs score in TCGA. (F) Time-dependent ROC curves of the signature for overall survival rates at 1 and 3 years in the TCGA cohort. AUC, area under the curve; CI, confidence interval; FPR, false positive rate; HR, hazard ratio; LASSO, least absolute shrinkage and selection operator; MNPs, microplastics and nanoplastics; PAAD, pancreatic ductal adenocarcinoma; PMNPs, MNPs-related genes in PAAD; ROC, receiver operating characteristic; TCGA, The Cancer Genome Atlas; TPR, true positive rate.

Figure 5.

Assessment of the PMNPs risk model. (A) Univariate and multivariate Cox analyses of clinical factors and risk score with OS. (B) Nomogram estimating the survival rates at 1 and 3 years for patients with PAAD. (C) Calibration chart to confirm the accuracy of the nomogram. (D,E) The DCA curve revealed the clinical net benefit of the constructed model in the TCGA cohort. CI, confidence interval; DCA, decision curve analysis; HR, hazard ratio; M, metastasis; MNPs, microplastics and nanoplastics; N, node; OS, overall survival; PAAD, pancreatic ductal adenocarcinoma; PMNPs, MNPs-related genes in PAAD; T, tumor; TCGA, The Cancer Genome Atlas.

In the ICGC and GSE71729 cohorts, the prognostic effectiveness of the PMNPs signature was additionally verified. Risk factor grouping was calculated using the same method as in the TCGA-PAAD training cohort. Consistent with the TCGA cohort, patients with low-risk showed better survival outcomes, and the expression patterns of PMNPs were similar in the two validation datasets. Time-dependent ROC analysis yielded AUC values for overall survival as follows: in GSE71729, 1-year AUC =0.695 (0.596–0.794) and 3-year AUC =0.828 (0.715–0.941); in the ICGC cohort, 1-year AUC =0.716 (0.6452–0.786) and 3-year AUC =0.676 (0.5677–0.7839) (Figure 6A-6J).

Figure 6.

Validation of the prognostic model associated with PMNPs. Visualization of risk score distribution and survival status within the GSE71729 (A) and ICGC cohort (F). Kaplan-Meier curves of patients with low and high PMNPs score in GSE71729 (B) and ICGC cohort (G). Time-dependent ROC curves of the signature for overall survival rates at 1 and 3 years in the GSE71729 (C) and ICGC cohort (H). The DCA curve revealed the clinical net benefit of the constructed model in the GSE71729 (D,E) and ICGC cohort (I,J). AUC, area under the curve; DCA, decision curve analysis; FPR, false positive rate; HR, hazard ratio; ICGC, International Cancer Genome Consortium; MNPs, microplastics and nanoplastics; PAAD, pancreatic ductal adenocarcinoma; PMNPs, MNPs-related genes in PAAD; ROC, receiver operating characteristic; TPR, true positive rate.

Immune characteristics of PMNPs in PAAD

The therapeutic response and prognosis in cancer patients are closely linked to the tumor immune microenvironment. The TCGA-PAAD cohort was used to examine the link between PMNPs and immune cell infiltration. Figure 7 illustrates the correlations and risk score–based immune profiles. Expression levels of XDH, KIF20A, and ASPM were positively correlated with Th2 cells and negatively correlated with most other immune cell subsets. In contrast, IL1A expression showed positive correlations with activated dendritic cells (aDCs), dendritic cells (DCs), macrophages, neutrophils, Th1, Th2, and Treg cells, while exhibiting a negative correlation with T helper 17 (Th17) cells (Figure 7A). IL1A expression was also positively associated with immune checkpoint-related genes, including SIGLEC15, TIGIT, CD274, HAVCR2, CTLA4, and IGSF8 (Figure 7C). Stratification by risk score revealed distinct immune infiltration patterns. Low-risk patients exhibited higher abundance of B cells, CD8⁺ T cells, cytotoxic cells, immature dendritic cells (iDCs), mast cells, plasmacytoid dendritic cells (pDCs), total T cells, Tem cells, T follicular helper (TFH) cells, Th17 cells, and Treg cells. In contrast, high-risk patients showed increased infiltration of Th2 cells, accompanied by elevated expression of SIGLEC15, CD274, and IGSF8 (Figure 7B,7D).

Figure 7.

Correlation of immune cell and checkpoints infiltrates with the prognostic model in PAAD. (A) The relationship between the four PMNPs and immune cells. (B) The immune cell infiltration scores between high- and low-risk groups. (C) The relationship between the four PMNPs and immune checkpoints. (D) The expression level of immune checkpoints between high- and low-risk groups. *, P<0.05; **, P<0.01; ***, P<0.001. aDC, activated dendritic cell; Cor, correlation; DC, dendritic cell; iDC, immature dendritic cell; MNPs, microplastics and nanoplastics; NK, natural killer; PAAD, pancreatic ductal adenocarcinoma; pDC, plasmacytoid dendritic cell; PMNPs, MNPs-related genes in PAAD.

The single-cell analysis of PMNPs in PAAD

UMAP analysis of the PAAD_CRA001160 single-cell dataset demonstrated the cellular distribution following dimensionality reduction (Figure 8A). The PMNPs-high group was enriched in B cells, CD8 Tex cells, M1 macrophages, and malignant cells, whereas the PMNPs-low group was characterized by higher proportions of monocytes and pDCs. Distinct PMNPs expression patterns across different cell populations are shown in Figure 8B,8C. The Figure S1 also demonstrates results consistent with Figure 8.

Figure 8.

Correlation of the PMNPs signature with single-cell clusters. (A) Proportion of PMNPs-high and PMNPs-low expression in single-cell level. (B) UMAP plot of thirteen major cell clusters in PAAD microenvironment. (C) AUCell analysis of the PMNPs signature in PAAD microenvironment. MNPs, microplastics and nanoplastics; PAAD, pancreatic ductal adenocarcinoma; pDC, plasmacytoid dendritic cell; PMNPs, MNPs-related genes in PAAD; UMAP, Uniform Manifold Approximation and Projection.

Spatial transcriptomic analysis

Spatial transcriptomics provides a high-resolution framework for mapping cellular localization and gene expression, enabling characterization of tissue architecture and microenvironmental interactions. The distribution of malignant and normal regions in PAAD tissue is shown in Figure 9A. AUCell-derived spatial scores are mapped in Figure 9B, where each point represents a Visium spot and the color gradient reflects cell-type abundance across regions. PMNPs expression was significantly higher in malignant regions compared with normal regions (Figure 9C). Correlation analysis between PMNPs expression and cellular composition across all spots further supported this pattern (Figure 9D), consistent with transcriptomic findings. The Figures S2-S4 also demonstrates results consistent with Figure 9. These results suggest that PMNPs may be associated with immune cell infiltration patterns.

Figure 9.

The spatial transcriptome analysis. (A) The distribution of malignant, mixed and normal area in PAAD tissue. (B) Spatial distribution of AUCell scoring on PAAD tissue sections. (C) The expression level of PMNPs in different zone of PAAD tissue. (D) AUC score and the correlation curve between each gene and the proportion of deconvoluted cell types. AUC, area under the curve; Cor, correlation; DC, dendritic cell; MNPs, microplastics and nanoplastics; PAAD, pancreatic ductal adenocarcinoma; PMNPs, MNPs-related genes in PAAD.

IL1A drives in vitro proliferation and invasion of pancreatic cancer cells

To functionally characterize IL1A (Figure 10), we examined four pancreatic cancer cell lines, revealing a significant overexpression of IL1A at the protein level (Figure 10A).Based on these findings, CAPAN-1 and BXPC-3 cell lines were selected for further investigation to elucidate the functional role of IL1A in pancreatic cancer. In both cell lines, IL1A expression was modified with the use of shRNA, leading to significant changes in protein level (Figure 10B). Transwell assays showed that IL1A knockdown significantly suppressed cell invasion (Figure 10C). CCK-8 assays further revealed that IL1A silencing significantly impaired these capabilities (Figure 10D,10F) and these results were confirmed by the colony formation assays (Figure 10E). Following the knockdown of IL1A, there was an enhancement in the apoptotic capacity of two pancreatic cancer cell lines. This was accompanied by a reduction in BCL-2 expression and an upregulation of BAX expression (Figure 10G-10I). Furthermore, in the present study, after treating CFPAC-1, CAPAN-1, and BXPC-3 cells with different concentrations of polystyrene, it was found that the expression level of IL1A increased with increasing polystyrene concentrations in CFPAC-1 and CAPAN-1 cells, whereas no significant change was observed in BXPC-3 cells (Figure 10J).

Figure 10.

Role of IL1A expression in the proliferation and invasion of pancreatic cancer cells. (A,B) The transfection of CAPAN-1 and BxPC-3 cells with empty lentivirus (sh-NC) and IL1A knockdown lentivirus (sh-IL1A) was performed, and WB was used to verify the knockdown efficiency of IL1A. (C) The invasion capability of cancer cells was assessed using Transwell assays. Crystal violet staining; 200×. (D,F) The proliferation ability of cancer cells was assessed using CCK-8. (E) The proliferation ability of cancer cells was assessed using colony formation assay. Crystal violet staining; 1× (macroscopic imaging). (H) The expression level of BAX and BCL-2 following IL1A knockdown. (G,I) The degree of apoptosis in the two cancer cells was evaluated through flow cytometry. (J) Western blot for IL1A in PAAD cells treated with polystyrene microspheres. **, P<0.01; ***, P<0.001. CCK-8, Cell Counting Kit-8; PAAD, pancreatic ductal adenocarcinoma.

Discussion

MNPs, as emerging environmental pollutants, have been recently detected accumulating in human tissues, including PAAD. Bibliometric analysis also showed a marked upward trend in MNPs-related cancer research, with a pronounced acceleration in publication growth after 2020 (R²=0.9787), likely reflecting both the escalating burden of global plastic pollution and advances in detection technologies. China contributed the highest number of publications, highlighting regional disparities in research capacity and environmental exposure profiles and indicating the importance of international collaboration and data sharing. Keyword co-occurrence analysis showed that oxidative stress and inflammation may represent core mechanisms underlying MNPs-associated carcinogenesis, providing testable hypotheses for future experimental studies. The bibliometric trends justified the focus on microplastic or nanoplastics induced PAAD mechanisms by showing the growing research urgency in this field. In this context, the present study systematically investigated the association between MNPs, a novel environmental exposure factor, and PAAD prognosis using a multi-omics integrative strategy.

Through cross-omics integration, 19 PMNPs were identified by intersecting transcriptomic datasets from TCGA-PAAD and GSE62165 with the CTD toxicogenomics database. These genes were predominantly upregulated in pancreatic tumorigenesis, suggesting that environmental exposure-driven transcriptional dysregulation may contribute to PAAD development and may provide novel avenues for early diagnosis and risk stratification. The observed upregulation pattern implies that MNPs-induced cellular stress and/or epigenetic remodeling may represent key mechanisms driving gene activation, potentially mediated by oxidative stress-related pathways. Gene network analysis further revealed potential co-regulatory interactions, offering important clues for identifying upstream regulatory mechanisms and therapeutic targets. A four-gene prognostic signature constructed using LASSO regression (XDH, IL1A, KIF20A, and ASPM) demonstrated strong predictive performance for overall survival in the TCGA-PAAD training cohort and maintained robust prognostic value in external validation cohorts (GSE71729 and ICGC). These findings were further supported by clinical DCA, suggesting potential clinical relevance that warrants further investigation.

Among the MNPs model, IL1A may represent an important link between tumor-associated inflammation and immune remodeling. IL-1α functions as both a pro-inflammatory cytokine and a damage-associated alarmin and can mediate communication between malignant and stromal cells. Tjomsland et al. demonstrated that PAAD cell-derived IL-1α maintained an inflammatory phenotype in cancer-associated fibroblasts and promoted the production of several tumor-supporting inflammatory mediators (28). IL-1 signaling may also stimulate thymic stromal lymphopoietin production by cancer-associated fibroblasts, thereby contributing to Th2-polarizing inflammation (29). Moreover, IL1A has recently been identified as an independent prognostic biomarker and potential therapeutic target in pancreatic cancer (30). Zhao et al. found the presence of microplastics in 70% of pancreatic cancer tissue samples. In tumors infiltrated by microplastics, there was a notable reduction in CD8⁺ T cells, natural killer (NK) cells, and dendritic cells, accompanied by an accumulation of neutrophils. This observation suggests a transition towards an immunosuppressive microenvironment (31). In our study, we found that IL1A protein expression was upregulated in CFPAC‑1 and CAPAN‑1 cells upon exposure to increasing concentrations of polystyrene. Immune infiltration analysis further revealed that IL1A expression was positively correlated with the infiltration abundance of dendritic cells (including aDC and DC), neutrophils, and macrophages. Based on the above findings, we hypothesize that microplastics may reshape the TME through IL1A upregulation, thereby influencing the prognosis of PAAD patients.

The enrichment of Th2 cells in the high-risk group is particularly relevant to the established immune biology of PAAD. De Monte et al. showed that crosstalk between pancreatic cancer cells and cancer-associated fibroblasts induced a thymic stromal lymphopoietin-dependent Th2 response and that intratumoral Th2-cell accumulation was associated with reduced patient survival (32). Th2 cytokines may promote fibrosis, support alternatively activated myeloid cells, and suppress effective cytotoxic immune responses. Therefore, the observed positive correlation between IL1A expression and Th2 cell infiltration, coupled with the increased presence of Th2 cells in the high-risk group, indicates that IL1A-related inflammatory signaling pathways may play a role in fostering a Th2-biased, tumor-promoting microenvironment. However, given that the current findings are based on transcriptomic correlations, they do not provide direct evidence that IL1A induces Th2 differentiation. To evaluate this hypothesis, further functional studies are necessary.

Notably, Th17 cells exert anti-tumor effects in pancreatic cancer. Gnerlich et al. demonstrated in a murine pancreatic cancer model that IL-6 induces Th17 cell differentiation, increases IFN-γ⁺CD8⁺ T cell numbers, inhibits tumor growth, and improves survival (33). In our study, the high-risk group showed low Th17 cell infiltration but high Th2 cell infiltration, which aligns with the pro-tumoral role of Th2 cells in promoting M2 macrophage function and tumor metastasis in TME (34). The low-risk group showed higher infiltration of TFH cells, pDCs, and mast cells. Among these, TFH cells are linked to a favorable prognosis in pancreatic cancer, acting through IL-21 and CXCL13 to promote B-cell maturation and tertiary lymphoid structure formation (35).

The simultaneous upregulation of SIGLEC15, CD274, and IGSF8 in high-risk group further indicates that this risk phenotype may involve several complementary mechanisms of immune escape. CD274 encodes PD-L1, which suppresses activated T cells through the PD-1/PD-L1 pathway. However, the limited efficacy of immune-checkpoint inhibitor monotherapy in most PAAD patients suggests that additional immunosuppressive pathways operate alongside PD-L1. SIGLEC15 is an emerging myeloid-associated immune checkpoint. Li et al. demonstrated that SIGLEC15 enhanced the immunosuppressive properties of tumor-associated macrophages and promoted M2-like polarization in PAAD (36). IGSF8 has also recently been identified as an innate immune checkpoint that suppresses natural killer-cell activity through inhibitory receptor interactions, whereas IGSF8 blockade restores antitumor innate immunity (37). Therefore, the concomitant upregulation of CD274, SIGLEC15, and IGSF8 in the high-risk group may suggest the concurrent presence of immunosuppressive mechanisms involving T cells, myeloid cells, and NK cells.

Single-cell RNA sequencing further revealed that high PMNPs expression was predominantly enriched in malignant cells and exhausted CD8⁺ T-cell subsets, whereas low PMNPs expression was mainly associated with monocytes and pDCs, suggesting a potential link between heterogeneous PMNPs expression and distinct tumor and immune cell states. Consistent with this observation, Raghavan et al. (38) demonstrated that the TME can drive cancer cell state plasticity through non-genetic mechanisms and thereby influence therapeutic responses, highlighting the dynamic interplay between malignant cells and their surrounding microenvironment. Moreover, Zhang et al. (39) revealed distinct immune cell compositions and functional states across different disease stages and tissue compartments of pancreatic adenocarcinoma, with specific T-cell and macrophage subsets associated with liver metastasis and patient prognosis. Elhossiny et al. (40) further demonstrated asynchronous evolution of epithelial and stromal compartments during pancreatic cancer progression, accompanied by spatial remodeling of immune and stromal populations. Notably, Park et al. (41) emphasized that the functional heterogeneity of immune cell subsets within the TME is critical for determining their roles in tumor immunity, indicating that differences in immune cell infiltration alone may not fully reflect their biological functions. In this context, the PMNPs-associated immune infiltration patterns observed in our study may reflect differences in immune cell states rather than merely changes in cellular abundance. Collectively, these findings suggest that heterogeneous PMNPs expression is associated with distinct malignant and immune cell states and may reflect the complex interplay between tumor cells and the immune microenvironment in PAAD. Further functional investigations are warranted to elucidate the underlying mechanisms and biological significance of these associations.

The remaining genes may contribute to the aggressive biological characteristics of the high-risk group. KIF20A promotes proliferation, migration, and invasion of pancreatic cancer cells, whereas inhibition of KIF20A suppresses these malignant phenotypes (42). ASPM participates in cell-cycle regulation and Wnt signaling pathways, and specific ASPM isoforms have been associated with stemness and poor prognosis in PAAD (43). XDH regulates purine catabolism and cellular redox homeostasis, although its context-specific role in PDAC requires further clarification. Thus, the prognostic signature may integrate proliferative and invasive tumor-cell programs with inflammatory and immunosuppressive features of the TME. The combination of these four genes may reflect the malignant characteristics of microplastic-related pancreatic cancer through a synergistic mechanism of “inflammation-proliferation-immune evasion.”

Despite integrating multi-omics analyses with in vitro experiments, several limitations should be acknowledged. First, reliance on toxicogenomic databases to infer MNPs-gene associations, without direct quantification of tissue MNPs burden or detailed environmental exposure histories, limits causal inference. In addition, the use of a uniform bioinformatics pipeline cannot fully eliminate batch effects and technical heterogeneity inherent to publicly available datasets derived from independent studies. Functional validation was primarily focused on IL1A, whereas the mechanistic roles of other signature genes (XDH, KIF20A, ASPM) remain unclear, and no in vivo MNPs exposure model was included to support the findings. Finally, owing to the inherent heterogeneity of publicly available external datasets (GSE71729 and ICGC cohorts), essential clinical covariates, including tumor stage, were not consistently recorded. Consequently, we were unable to validate the full clinical utility of our nomogram in an independent external setting, highlighting the need for prospective clinical studies to further assess model generalizability and translational potential.

Conclusions

In conclusion, this study establishes a novel molecular framework linking MNPs exposure to PAAD prognosis through an integrative multi-dimensional approach. Nineteen PMNPs and a four-gene prognostic signature (XDH, IL1A, KIF20A, and ASPM) were identified, demonstrating robust risk stratification performance across independent cohorts and strong associations with immunosuppressive microenvironmental features and tumor spatial heterogeneity. Functional experiments further confirmed that IL1A promotes PAAD cell proliferation and invasion while suppressing apoptosis. Polystyrene exposure upregulated IL1A expression, suggesting that polystyrene may regulate the malignant progression of PAAD through IL1A. This work introduces a framework for incorporating environmental exposure factors into pancreatic cancer prognostic modeling and provides potential directions for early warning strategies and precision therapy. Future studies should leverage MNPs exposure models to further elucidate upstream regulatory networks and evaluate therapeutic targets such as IL1A-mediated immune evasion, facilitating translation from experimental findings to clinical application.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Data Sharing Statement

Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1405/dss

tcr-15-08-585-dss.pdf ^{(82.4KB, pdf)}

DOI: 10.21037/tcr-2026-1405