Abstract

   Studies have indicated a complex association between chronic
   obstructive pulmonary disease (COPD) and lung adenocarcinoma (LUAD).
   However, the underlying mechanisms of their coexistence are still not
   fully understood. Thus, this study evaluated the possible mechanisms
   and biomarkers of COPD and LUAD by analyzing public RNA sequencing
   databases via bioinformatics analysis. This study obtained the LUAD
   datasets (TCGA-LUAD, [36]GSE118370, and [37]GSE30219) and the COPD
   dataset ([38]GSE11784 and [39]GSE39874) from TCGA and GEO databases,
   respectively. The differentially expressed genes (DEGs) were analyzed
   using the DESeq2 and limma packages. These DEGs were then intersected
   with pyroptosis-related genes (PRGs) to produce PRDEGs, which were
   examined via GO analysis and KEGG enrichment analyses. Simultaneously,
   a prognostic model was developed using PRDEGs by the TCGA-LUAD dataset
   to generate diagnostic PRDEGs (DPRDEGs). The STING database was
   employed to develop a protein-protein interaction (PPI) network for
   DPRDEGs. Transcription factors-associated with DPRDEGs were also
   identified in the ChIPBase and hTFtarget databases. The comparative
   toxicogenomics database (CTD) was employed to detect possible drugs or
   small molecules that interacted with DPRDEGs, and results were
   illustrated using Cytoscape. Moreover, this study developed a
   prognostic model using multivariate analysis and simultaneously
   conducted a prognostic analysis. The results were further validated by
   immunohistochemistry (IHC), western blotting (WB), and qPCR of clinical
   specimens. A total of 273 DEGs were identified, and 12 PRDEGs were
   detected after intersecting with PRGs. Inflammation and infectious
   diseases were the primary enriched regions for these PRDEGs, as
   indicated by GO and KEGG enrichment analyses. The study identified six
   DPRDEGs (BNIP3, FTO, NEK7, POLR2H, S100A12, and TLR4) via prognosis
   modeling of PRDEGs. The expression of these DPRDEGs in COPD and LUAD
   was verified through IHC, WB, and qPCR examinations. Based on
   multifactorial prognosis modeling, among six, FTO, POLR2H, S100A12, and
   TLR4 revealed enhanced prognostic predictive effects. This study
   demonstrated that COPD and LUAD have common pathogenic mechanisms. The
   identified DPRDEGs and predictive models offer new perspectives for
   understanding and addressing COPD and LUAD.

   Keywords: COPD, Lung adenocarcinoma, Co-occurrence, Biomarkers, Early
   diagnosis, Pyroptois, Bioinformatics analysis

   Subject terms: Non-small-cell lung cancer, Computational biology and
   bioinformatics

Introduction

   Based on the Global Initiative for Chronic Obstructive Lung Disease,
   COPD is defined by prolonged respiratory manifestations and airflow
   obstruction resulting from disruptions in the airways or alveoli^[40]1.
   It is globally recognized as the third leading cause of mortality,
   affecting approximately 12% of the population and contributing to 3
   million deaths annually^[41]2. The etiology of COPD is complex, with
   smoking being the primary environmental risk factor, while genetic
   factors and abnormal lung development could also contribute to its
   occurrence. Currently, COPD remains an incurable chronic disease
   without specific treatments, highlighting the critical importance of
   preventive interventions over pharmacological therapies. Smoking
   cessation is regarded as the most critical preventive measure, while
   vaccination is an alternative preventive approach. All patients
   aged ≥ 65 years are recommended to be administered either the 13-valent
   pneumococcal conjugate vaccine (PCV13) or the 23-valent pneumococcal
   polysaccharide vaccine (PPSV23). Younger COPD patients with major
   comorbidities, such as chronic heart disease, are also advised to
   initiate treatment with the PPSV23. Available pharmacological
   treatments at present involve inhaled long-acting bronchodilators,
   corticosteroids, etc., based on quitting smoking^[42]3.

   Lung cancer (LC) is the most lethal tumor disease, with COPD being the
   most common comorbidity. Approximately 40% to 70% of lung cancer
   patients exhibit varying degrees of airflow obstruction, and these
   patients typically have poor clinical treatment outcomes^[43]4,[44]5.
   There is increasing evidence of a complex interaction between COPD and
   lung cancer. The widespread and sustained inflammatory response in the
   lung tissue of COPD patients increases the risk of lung cancer
   development^[45]6. Concurrent COPD could promote primary tumor
   metastasis in the lungs^[46]7,[47]8.

   Pyroptosis is a novel form of programmed cell death mediated by
   Gasdermin (GSDM) proteins, which is characterized by the release of
   inflammatory factors^[48]9. It plays dual roles in homeostasis:
   moderate activation clears pathogens via inflammation, while excessive
   activation exacerbates tissue damage. It has an important role in both
   COPD and lung cancer^[49]10. Cigarette smoke-induced endoplasmic
   reticulum stress activates pyroptosis via the ROS/NLRP3/caspase-1
   signaling pathway, contributing to the progression of COPD.
   Particularly, ROS activates the NLRP3 inflammasome, which in turn
   triggers caspase-1 stimulation and the release of inflammatory
   mediators like IL-1β. This process induces pyroptosis and contributes
   to COPD progression^[50]11. Pyroptosis can induce an inflammatory
   microenvironment that may increase the proliferation of tumor
   cells^[51]12. It could function as a “bridge” between lung cancer and
   COPD; however, its mechanisms have yet to be fully understood.

   This study aims to identify pyroptosis-related genes shared between
   LUAD and COPD by downloading datasets from the Cancer Genome Atlas
   (TCGA) and the Gene Expression Omnibus (GEO) databases for LUAD and
   COPD, respectively. The datasets were analyzed to identify DEGs, which
   were then intersected with PRGs to obtain PRDEGs. As a result, BNIP3,
   FTO, NEK7, POLR2H, S100A12, and TLR4 were identified as six diagnostic
   PRDEGs (DPRDEGs) via prognosis modeling of PRDEGs. The expression of
   these six DPRDEGs in COPD and LUAD was verified using IHC, WB, and qPCR
   techniques.

   Their prognostic effects were compared through multifactorial
   prognostic modeling. Therefore, this study aimed to clarify the
   comorbid mechanisms between COPD and LUAD from the perspective of
   pyroptosis, thereby offering novel insights into their association and
   establishing the theoretical basis for their clinical diagnosis and
   treatment.

Materials and methods

Data acquisition

   The TCGA portal ([52]https://portal.gdc.cancer.gov/) was the main
   source to obtain the LUAD dataset (TCGA-LUAD). This dataset consists of
   59 similar adjacent normal (Group: Normal) and 539 LUAD (Group: LUAD)
   samples. The clinical data associated with these samples originated
   from the UCSC Xena database ([53]http://genome.ucsc.edu).

   Further, LUAD patient datasets ([54]GSE118370 and [55]GSE30219) were
   obtained from the GEO database ([56]http://www.ncbi.nlm.nih.gov/geo).
   The [57]GSE118370 dataset comprised 12 samples (6 tumor tissues from
   LUAD and 6 normal lung tissues). The dataset [58]GSE30219 consisted of
   307 samples (293 tumor tissues from LUAD and 14 normal lung tissues).

   Moreover, datasets of COPD patients ([59]GSE11784 and [60]GSE38974)
   were obtained from the GEO database. The [61]GSE11784 dataset contains
   171 samples (72 samples from healthy smokers and 22 samples from
   smokers with COPD). The [62]GSE38974 dataset comprised 59 samples (23
   lung tumor samples from COPD patients and 9 non-tumor lung tissue
   samples). Details on each dataset have been uploaded to (Supplementary
   Table 1).

Data processing

   The LUAD datasets ([63]GSE118370 and [64]GSE30219) and the COPD
   datasets ([65]GSE118370 and [66]GSE30219) were initially merged and
   processed individually. The R package sva (Version 3.50.0)^[67]13 was
   employed to eliminate batch effects, and the resulting data were
   standardized using the limma package (Version 3.58.1)^[68]14.
   Therefore, both LUAD and COPD datasets were acquired. Further, the
   limma package was used to standardize the TCGA-LUAD dataset.

Identification of differentially expressed genes

   The DESeq2 package (Version 1.42.0)^[69]15 was used to conduct
   differential analysis on the TCGA-LUAD dataset and the limma package on
   the LUAD dataset and COPD dataset to identify DEGs between distinct
   groups (LUAD/Normal, COPD/Normal). To ensure the broad capture of
   potential genes, DEGs were identified by applying inclusive thresholds,
   with |logFC|≥ 0 and p.adj ≤ 0.05^[70]16. However, PRDEGs were obtained
   by the intersection of DEGs and PRGs.

Functional and pathway enrichment analyses

   Gene ontology (GO)^[71]17 analysis is a widely used technique for
   large-scale functional enrichment studies. It involves analyzing
   biological processes (BP), molecular functions (MF), and cellular
   components (CC). The Kyoto Encyclopedia of Genes and Genomes
   (KEGG)^[72]18 is a well-known database that stores enriched genomic
   data, biological mechanisms, diseases, and drugs. The R package
   clusterProfiler (Version 4.10.0)^[73]19 was used to conduct GO and KEGG
   annotation analysis on PRDEGs. The thresholds for entry selection were
   p.adj ≤ 0.05 and FDR (q) ≤ 0.05 value, with the method of
   Benjamini-Hochberg (BH) to adjust p.adj values.

Construction of PRDEGs prognostic model

   First, a support vector machine (SVM)^[74]20 model was constructed
   using the expression matrix and grouping data from the TCGA-LUAD
   dataset. PRDEGs were selected based on achieving the highest accuracy
   and lowest error rate as determined by the SVM algorithm. Subsequently,
   a model was developed using the randomForest package (Version
   4.7-1.1)^[75]21.
   graphic file with name d33e556.gif 1

   To select PRDEGs and develop the logistic diagnostic model, an analysis
   was performed on PRDEGs using a significance threshold of p ≤ 0.05.
   Their molecular expression within the model was visualized using a
   forest plot. Accordingly, the glmnet package (Version 4.1–8)^[76]22 was
   employed to conduct a Least Absolute Shrinkage and Selection Operator
   (LASSO) regression analysis on PRDEGs, yielding the Logistic-LASSO
   regression model. Risk scores were computed according to the following
   equation:
   graphic file with name d33e571.gif 2

   To illustrate the interconnections of DPRDEGs within the PRDEGs
   prognostic model, a nomogram was generated using the rms package
   (Version 6.7-1) based on the logistic LASSO regression analysis
   results^[77]23. Calibration curves were plotted to assess the accuracy
   and discriminative ability of the PRDEGs prognostic model. Finally,
   Decision Curve Analysis (DCA) plots were generated using the ggDCA
   package (Version 1.1)^[78]24 according to the risk scores to further
   evaluate the model’s performance.

Protein-protein interaction network

   A protein-protein interaction (PPI) network associated with DPRDEGs was
   constructed using the STRING database^[79]25
   ([80]https://string-db.org/), with an interaction threshold set to
   0.150 (low confidence), and the network was visualized using
   Cytoscape^[81]26 (Version 3.9.1).

Construction of mRNA-TF network and mRNA-drugs network

   To further explore the regulatory mechanisms of DPRDEGs, common
   transcription factors (TFs) interacting with the DPRDEGs were
   identified using the CHIPBase^[82]27
   ([83]https://rna.sysu.edu.cn/chipbase/) and hTFtarget^[84]28
   ([85]http://bioinfo.life.hust.edu.cn/hTFtarget/) databases, followed by
   the construction of an mRNA-TF interaction network. Additionally, the
   comparative toxicogenomics database (CTD)^[86]29
   ([87]http://ctdbase.org/) was employed to predict potential drugs or
   small molecules interacting with DPRDEGs and to construct an mRNA-Drug
   interaction network. The mRNA-TF and mRNA-drugs interaction networks
   were then visualized using Cytoscape.

Construction of multifactorial DPRDEGs prognostic model

   Statistically significant DPRDEGs (p ≤ 0.05) were initially selected
   using univariate Cox regression analysis. Subsequently, a
   multifactorial DPRDEGs prognostic model was established by performing
   multivariate Cox regression analysis on the selected genes. The rms
   package was employed to construct a nomogram for evaluating the
   survival prediction capability of the Cox prognostic model.

Validation and assessment of multifactorial DPRDEGs prognostic model

   Model accuracy was assessed by comparing predicted survival rates with
   actual outcomes through calibration curves. Additionally, the clinical
   predictive value of the model was examined via DCA using the ggDCA
   package.

Human samples

   Lung tissue specimens were prospectively obtained from four distinct
   cohorts: (1) patients with co-occurring LUAD and COPD (n = 5), (2) LUAD
   patients without documented COPD (n = 5), paired with histologically
   normal adjacent tissues located at least 5 cm from tumor margins
   (n = 5), and (3) COPD patients without malignancy (n = 5). All
   specimens were either surgically resected or biopsy-derived from the
   Department of Respiratory Medicine, Second Affiliated Hospital of
   Shenyang Medical College.

   The diagnosis of COPD was confirmed according to the 2024 Global
   Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines
   (post-bronchodilator FEV1/FVC < 0.70), excluding individuals with a
   history of malignancy or chronic respiratory comorbidities. The LUAD
   cohort required a pathologically confirmed diagnosis of primary lung
   adenocarcinoma, as defined by the 2021 WHO Classification of Tumours
   (5th edition), with normal lung function (FEV1/FVC ≥ 0.70) and tumor
   stages I-III, as per the AJCC 9th edition TNM staging. Adjacent normal
   tissues were validated by two independent pathologists to ensure the
   absence of tumor infiltration or atypical hyperplasia. The COPD-LUAD
   cohort met both diagnostic criteria.

   Exclusion criteria included: concurrent pulmonary conditions (e.g.,
   asthma, interstitial lung disease), autoimmune disorders, malignancies
   diagnosed within the past 5 years, use of corticosteroids or
   immunosuppressants within 3 months prior to enrollment, neoadjuvant
   therapy in LUAD, suboptimal specimens (e.g., improper fixation or
   autolysis), pregnancy or lactation, and severe organ dysfunction.

   No statistically significant differences in age or gender were observed
   between the patient groups. Informed consent was obtained from all
   participants, and the study protocol was approved by the Ethics
   Committee of the Second Affiliated Hospital of Shenyang Medical
   College.

Immunohistochemistry

   All tissue samples were placed in 4% paraformaldehyde (PFA) for
   fixation at 4 ℃ for 24 h. They were further placed in paraffin wax and
   sectioned (5 μm thick) for IHC analysis. The primary antibodies used
   for staining the sections are presented in (Supplementary Table 2). All
   these antibodies were used as per the provided protocols.

RNA isolation and quantitative real-time PCR

   The total content of RNA from human samples was isolated and
   transcribed into cDNA via PrimeScriptTM RT Master Mix (TAKARA, Tokyo,
   Japan) under the precise protocol of TRIzol reagent (Invitrogen, CA,
   USA). PCR procedures were carried out using the TB Green Premix Ex Taq
   II (TAKARA, Tokyo, Japan) on the Accurate 96 Real-time fluorescence
   quantitative PCR System (DLAB, Beijing, China). The primers used for
   PCR amplification of specific genes are presented in (Supplementary
   Table 3). The relative transcription pattern was estimated via the
   2-ΔΔCT method and β-actin was used to normalize each expression.

Western blotting

   Human samples were processed to extract the total protein content.
   After separation on 10% SDS-PAGE, 30 μg of proteins were transferred to
   PVDF membranes (Millipore, Bedford, MA, USA). Based on the provided
   instruction, the membranes were kept for two h blocking for 24 h at
   4 °C with the following primary antibodies, which are listed in
   (Supplementary Table 4). Subsequently, each membrane was kept at 25 °C
   for 1 h with peroxidase-linked anti-mouse and anti-rabbit secondary
   antibodies (1:5000, ZSGB-BIO, Beijing, China). The protein bands were
   depicted via the respective ECL reagent (Tanon, Shanghai, China),
   detected under the Tanon5200 gel imaging system (Tanon, Shanghai,
   China), and quantified using Image J software. The relative protein
   expression levels were estimated by comparing them to the quantity of
   the control protein β-actin (1:2000, ZSGB-BIO, Beijing, China).

Statistical analysis

   Bioinformatics analyses were performed using R software (version
   4.1.2). For normally distributed continuous variables, comparisons
   between two groups were conducted using an independent Student’s
   t-test; non-normally distributed variables were assessed with the
   Mann–Whitney U test (Wilcoxon rank-sum test). Survival analysis was
   carried out using the survival package in R, with Kaplan–Meier curves
   for visualization and log-rank tests for statistical significance.
   Spearman’s correlation analysis was used to evaluate relationships
   between variables, unless otherwise specified.

   Immunohistochemistry, Western Blotting, and qPCR analyses were
   performed using GraphPad Prism 8.0 (GraphPad Software, San Diego, CA,
   USA) with two-tailed tests. Data were analyzed using SPSS 17.0, and
   results are presented as mean ± standard deviation; statistical
   significance was set at P < 0.05.

Results

Identification and analysis of PRDEGs

   Figure [88]1 illustrates the study workflow. Differential gene
   expression analysis was performed to assess expression variations among
   groups in the TCGA-LUAD, LUAD, and COPD datasets. The TCGA-LUAD dataset
   was processed using the DESeq2 package, whereas the LUAD and COPD
   datasets were analyzed with the limma package. This analysis identified
   DEGs in each dataset, and volcano plots were generated to display the
   results (Fig. [89]2A). Common DEGs were determined by intersecting the
   TCGA-LUAD, LUAD, and COPD datasets, and these were represented in a
   Venn diagram (Fig. [90]2B). Subsequently, intersecting the common DEGs
   with PRGs from the three datasets yielded 12 PRDEGs, which are depicted
   in a Venn diagram (Fig. [91]2C). These 12 PRDEGs are BNIP3, CD14, FADD,
   FTO, GTF2H4, IL1B, NEK7, NLRP3, POLR2H, S100A12, TLR2, and TLR4.
   Finally, the differential expression of these 12 PRDEGs was compared
   among various groups within the respective datasets, and a heatmap was
   generated using the pheatmap package (Version 1.0.12) (Fig. [92]2D).

Fig. 1.

   [93]Fig. 1
   [94]Open in a new tab

   Work flowchart.

Fig. 2.

   [95]Fig. 2
   [96]Open in a new tab

   Identification and analysis of PRDEGs (A) Volcano plots of DEGs between
   different groups (LUAD/Normal, COPD/Normal) in the TCGA-LUAD, LUAD, and
   COPD datasets. Orange dots represent upregulated genes, blue dots
   indicate downregulated genes, and gray dots denote non-significantly
   different genes. (B) Venn diagram of DEGs in the TCGA-LUAD, LUAD, and
   COPD datasets, highlighting the intersections among the three datasets.
   (C) Venn diagram showing the overlap between Common DEGs across the
   three datasets and PRGs, identifying a total of 12 PRDEGs. (D) Heatmap
   of PRDEGs in different groups (LUAD/Normal, COPD/Normal) within the
   TCGA-LUAD, LUAD, and COPD datasets, illustrating the expression
   differences of the 12 PRDEGs across groups. DEGs differentially
   expressed genes, LUAD lung adenocarcinoma, COPD chronic obstructive
   pulmonary disease, PRGs pyroptosis-related genes, PRDEGs pyroptosis
   related differentially expressed genes.

Functional and pathway enrichment analyses

   To examine the association between LUAD and COPD, GO functional
   enrichment and KEGG pathway enrichment analyses were performed using
   the 12 PRDEGs. The results revealed that the 12 PRDEGs are primarily
   enriched in biological processes (BP), such as positive regulation of
   cytokine production, modulation of the inflammatory response,
   phagocytosis, positive regulation of T cell-mediated immunity, and
   lymphocyte proliferation. In contrast, these genes were enriched in
   cellular components (CC) including membrane rafts, RNA polymerase II
   holoenzyme, secretory granule membranes, the transcription factor TFIIH
   core complex, and phagocytic cups. Additionally, they showed enrichment
   in molecular functions (MF), such as pattern recognition receptor
   activity, lipopolysaccharide binding, NAD(P) + nucleosidase activity,
   interleukin-1 receptor binding, and death receptor binding. These
   PRDEGs were also abundant in KEGG pathways, including Salmonella
   infection, Coronavirus disease—COVID-19, Tuberculosis, Pertussis, and
   the IL-17 signaling pathway (Fig. [97]3A).

Fig. 3.

   [98]Fig. 3
   [99]Open in a new tab

   Functional and pathway enrichment analyses of PRDEGs (A) Bar chart
   illustrating the results of GO functional enrichment analysis and KEGG
   pathway enrichment analysis for PRDEGs, including biological process
   (BP, blue), cellular component (CC, orange), molecular function (MF,
   green), and KEGG pathways (red). (B–D) Network diagrams of GO
   functional enrichment analysis for PRDEGs, depicting BP (B), CC (C),
   and MF (D). Cyan nodes represent specific genes, orange circles
   indicate associated pathways, and the connecting lines illustrate
   gene-pathway relationships. (E) Circular network diagram of KEGG
   pathway enrichment analysis for PRDEGs, demonstrating the relationships
   between pathways and corresponding genes. Cyan nodes represent
   pathways, while orange nodes denote specific genes. The filtering
   criteria for GO and KEGG enrichment terms: P.adj < 0.05 and FDR
   (q.value) < 0.05. PRDEGs pyroptosis related differentially expressed
   genes, BP biological process, MF molecular function, CC cellular
   component.

   Moreover, the enriched GO functional groups for BP, CC, and MF
   (Fig. [100]3A–D) were illustrated via network diagrams. The findings of
   the KEGG pathway enrichment analysis were represented in a circular
   network diagram (Fig. [101]3E).

Differential expression analysis of PRDEGs

   The expression differences of the 12 PRDEGs in various groups of the
   TCGA-LUAD dataset were analyzed using Wilcoxon rank sum tests
   (Fig. [102]4A). The expression levels of BNIP3, FADD, GTF2H4, and
   POLR2H were upregulated in the LUAD group than in the normal group.
   Conversely, the levels of CD14, FTO, IL1B, NEK7, NLRP3, S100A12, TLR2,
   and TLR4 were lower in the LUAD group than in the normal group, with
   statistical significance (p ≤ 0.001). The expression variations of the
   12 PRDEGs in the LUAD and COPD datasets were examined via the same
   method (Fig. [103]4B,C). In the LUAD dataset, the expression patterns
   of BNIP3, FADD, GTF2H4, and POLR2H were enhanced in the LUAD group than
   in the normal group. Conversely, the levels of CD14, FTO, IL1B, NEK7,
   NLRP3, S100A12, TLR2, and TLR4 were reduced in the LUAD group than in
   the normal group (p ≤ 0.01). In the COPD dataset, the levels of CD14,
   FADD, GTF2H4, IL1B, NLRP3, POLR2H, S100A12, TLR2, and TLR4 were higher
   in the COPD group in contrast to the normal group, while the levels of
   BNIP3, FTO, and NEK7 were lower in the COPD group relative to the
   normal group (p ≤ 0.001). Furthermore, to identify the predictive value
   of the expression variances of the 12 PRDEGs in different groups of the
   TCGA-LUAD dataset, receiver operating characteristic (ROC) curves were
   plotted (Fig. [104]4D). Among them, CD14 (AUC = 0.660) showed
   relatively low accuracy in diagnosis. However, the expression of BNIP3
   (AUC = 0.840), FADD (AUC = 0.760), FTO (AUC = 0.804), GTF2H4
   (AUC = 0.738), IL1B (AUC = 0.793), NLRP3 (AUC = 0.713), S100A12
   (AUC = 0.852), TLR2 (AUC = 0.760) demonstrated moderate accuracy in
   diagnosis, while NEK7 (AUC = 0.914), POLR2H (AUC = 0.950), and TLR4
   (AUC = 0.936) reveled high accuracy in diagnosis.

Fig. 4.

   [105]Fig. 4
   [106]Open in a new tab

   Differential expression analysis of PRDEGs (A) Comparison of PRDEG
   expression levels between different groups (LUAD/Normal) in the
   TCGA-LUAD dataset, with all genes showing statistically significant
   differences. (B) Expression comparison of PRDEGs between LUAD and
   Normal groups in the LUAD dataset, with all genes exhibiting at least
   highly significant statistical differences. (C) Comparison of PRDEG
   expression levels between COPD and Normal groups in the COPD dataset,
   with all genes displaying statistically significant differences. (D)
   ROC curves of PRDEGs in the TCGA-LUAD dataset. The AUC values indicate
   the diagnostic accuracy of each gene in distinguishing between LUAD and
   Normal groups. “NS” means P ≥ 0.05, suggesting no statistical
   significance. “*” P < 0.05, “**” P < 0.01, and “***” P < 0.001,
   suggesting statistical significance. PRDEGs pyroptosis related
   differentially expressed genes, LUAD lung adenocarcinoma, COPD chronic
   obstructive pulmonary disease, ROC receiver operating characteristic
   curve, AUC the area under the curve.

Construction of PRDEGs prognostic model

   A logistic regression analysis was performed using the expression
   pattern and clustering information of the 12 PRDEGs in the TCGA-LUAD
   dataset to assess their diagnostic significance. The results were
   presented as a forest plot (Fig. [107]5A). Subsequently, the SVM
   algorithm was applied to develop an SVM model, which exhibited the
   highest accuracy and lowest error rate when the number of genes was 12
   (Fig. [108]5B,C). Furthermore, the random forest algorithm was used to
   observe the expression pattern of PRDEGs in the TCGA-LUAD dataset,
   identifying 10 diagnostic markers (BNIP3, CD14, FADD, FTO, IL1B, NEK7,
   POLR2H, S100A12, TLR2, and TLR4) (Fig. [109]5D,E).

Fig. 5.

   [110]Fig. 5
   [111]Open in a new tab

   Construction of the prognostic model for PRDEGs. (A) Forest plot of the
   PRDEGs logistic regression model. (B) Number of genes with the lowest
   error rate identified by the SVM algorithm. (C) Number of genes with
   the highest accuracy identified by the SVM algorithm. (D) Model
   training error plot for the random forest algorithm. (E) Feature
   importance ranking of PRDEGs in the random forest model, demonstrating
   the contribution of each gene to the model. (F) Diagram of the
   diagnostic model based on the LASSO regression. (G) Variable trajectory
   plot of the LASSO regression model. (H) Venn diagram showing the
   overlap of PRDEGs in the logistic-LASSO regression model, SVM model,
   and random forest model. (I) ROC curve of the PRDEGs prognostic model
   in the TCGA-LUAD dataset. (J) Nomogram of the six DPRDEGs in the PRDEGs
   prognostic model. (K) Calibration curve for the PRDEGs prognostic
   model. (L) DCA plot of the PRDEGs prognostic model. PRDEGs pyroptosis
   related differentially expressed genes, SVM support vector machine,
   LASSO least absolute shrinkage and selection operator, DPRDEGs
   diagnosis pyroptosis related differentially expressed genes, DCA
   decision curve analysis, ROC receiver operating characteristic curve,
   AUC area under the curve.

   The diagnostic model was then constructed via LASSO regression analysis
   and visualized by LASSO regression model and LASSO variable trajectory
   plots (Fig. [112]5F,G). This diagnostic model used seven PRDEGs (BNIP3,
   FTO, NEK7, NLRP3, POLR2H, S100A12, and TLR4). Moreover, six DPRDEGs
   were obtained by combining the PRDEGs from the Logistic-LASSO
   regression, SVM, and random forest models. These DPRDEGs were
   visualized in a Venn diagram (Fig. [113]5H) and were as follows: BNIP3,
   FTO, NEK7, POLR2H, S100A12, and TLR4. Next a prognostic model was
   established as per their expression pattern in the TCGA-LUAD dataset.
   graphic file with name d33e1133.gif 3

   The prognostic model was validated by plotting ROC curves based on the
   Risk Score from the PRDEGs prognostic model and clustering information
   from the TCGA-LUAD dataset (F[114]ig. [115]5I). The TCGA-LUAD dataset
   demonstrated that the Risk Score from the PRDEGs model provided high
   diagnostic accuracy for LUAD (AUC = 0.984). Moreover, a nomogram was
   developed based on the six DPRDEGs in the prognostic model
   (Fig. [116]5J). This nomogram indicated that POLR2H levels had a
   significantly greater impact on the model’s performance compared to
   other variables. Furthermore, the predictive performance of the model
   was evaluated using calibration analysis and DCA, as illustrated in
   (Fig. [117]5K,L). These results suggest that the model is highly
   accurate in predicting the occurrence of LUAD in the TCGA-LUAD dataset.

Construction of PPI network and module analysis

   The PPI analysis was carried out to examine the six DPRDEGs in the
   PRDEGs prognostic model within the TCGA-LUAD dataset via the STRING
   database. These DPRDEGs include BNIP3, FTO, NEK7, POLR2H, S100A12, and
   TLR4. The PPI network was developed with a minimum required interaction
   score of 0.150 to lower confidence, resulting in a network that
   includes the six DPRDEGs (Fig. [118]6A). All genes within the PPI
   network were designated as hub genes for the TCGA-LUAD dataset. The
   findings of the functional similarity examinations were then
   illustrated using a boxplot (Fig. [119]6B) based on the six hub genes
   in the PPI network. Among the six hub genes, S100A12 showed the highest
   functional similarity to other network components.

Fig. 6.

   [120]Fig. 6
   [121]Open in a new tab

   Development and module analysis of the PPI network (A) PPI network of
   DPRDEGs, illustrating the interactions among the six diagnostic genes.
   (B) Functional similarity analysis of DPRDEGs, presented as a boxplot
   to display the functional similarity among the genes. (C) mRNA- TF
   interaction network of DPRDEGs, depicting interactions between six
   DPRDEGs and 53 transcription factors. Light blue circular nodes
   represent mRNAs, while light green triangular nodes represent
   transcription factors. (D) mRNA-drug interaction network of DPRDEGs,
   showing the relationships between four DPRDEGs and 35 potential drugs
   or molecular compounds. Light blue elliptical nodes represent mRNAs,
   while light purple V-shaped nodes indicate drugs. PPI protein-protein
   interaction, DPRDEGs diagnosis pyroptosis related differentially
   expressed genes, TF transcription factor.

   Next, the CHIPBase database (3.0 V) and the hTFtarget database were
   then employed to identify TFs that interact with the six DPRDEGs. A
   total of 53 TF-gene interaction data were obtained by intersecting the
   association retrieved from both databases with these genes. This data
   was then displayed via Cytoscape software (Fig. [122]6C).

   Further, the CTD database was implemented to locate possible molecular
   compounds or drugs that could target the six DPRDEGs. Approximately 35
   effective drugs or molecules were detected that correspond to four
   DPRDEGs (BNIP3, FTO, NEK7, and POLR2H) via this analysis. Of these, 22
   drugs or compounds target the BNIP3 (Fig. [123]6D).

Construction and prognostic analysis of the multifactorial DPRDEGs prognostic
model in TCGA-LUAD dataset

   First, a single-factor Cox regression analysis was carried out using
   the expression profile of the six DPRDEGs (BNIP3, FTO, NEK7, POLR2H,
   S100A12, TLR4) in the TCGA-LUAD dataset to evaluate their predictive
   significance (excluding samples with missing clinical data). Therefore,
   variables that met p ≤ 0.10 in the single-factor analysis were
   incorporated into the multifactorial Cox regression analysis to develop
   the multifactorial DPRDEGs prognostic model. The findings of this
   analysis were then organized and displayed in a forest plot
   (Fig. [124]7A). The formula for estimating the RiskScore in the
   multifactorial DPRDEGs prognostic model is as follows:
   graphic file with name d33e1266.gif 4

Fig. 7.

   [125]Fig. 7
   [126]Open in a new tab

   Development and evaluation of the multi-factor DPRDEGs prognostic model
   in the TCGA-LUAD dataset (A) Forest plot of univariate Cox regression
   analysis for DPRDEGs in the TCGA-LUAD dataset, presenting the HR 95% CI
   and p-values for each gene. (B) Nomogram of the multivariate PRDEG
   prognostic model in the TCGA-LUAD dataset, illustrating the
   contribution of DPRDEGs to survival probability prediction. (C)
   Calibration curve of the multivariate DPRDEGs prognostic model,
   comparing the predicted survival probabilities at 1, 3, and 5 years
   with the actual observed survival probabilities. (D) DCA plot of the
   multivariate DPRDEGs prognostic model, demonstrating the net benefit of
   the prediction model at different thresholds, with clinical predictive
   performance ranked as 3-year > 5-year > 1-year. LUAD lung
   adenocarcinoma, DPRDEGs diagnosis pyroptosis related differentially
   expressed genes, DCA decision curve analysis.

   Next, nomogram analysis evaluated the prognostic capacity of the
   multifactorial DPRDEGs prognostic model and generated a nomogram plot
   (Fig. [127]7B). This plot showed that among the four DPRDEGs in the
   multifactorial prognostic model, S100A12 and TLR4 have increased
   efficiency in the model relative to other variables. The calibration
   analysis was performed for 1-, 3-, and 5-year prognoses based on the
   nomogram of the multifactorial DPRDEGs prognostic model, and
   calibration curves were plotted (Fig. [128]7C). The findings indicated
   that the predictive performance of the DEGs model ranked as follows:
   3-year ≥ 5-year ≥ 1-year. The DCA results are presented in
   (Fig. [129]7D), and the clinical utility of the developed
   multifactorial Cox regression model was assessed at 1-, 3-, and 5-year
   time points. The analysis further demonstrated that the clinical
   predictive efficacy of the multifactorial DPRDEGs prognostic model was
   highest for 3-year outcomes, followed by 5-year outcomes, and lowest
   for 1-year outcomes.

Detection of DPRDEGs expression levels

   To evaluate the expression profiles of the six DPRDEGs in clinical
   samples, IHC (Fig. [130]8A,B), qPCR (Fig. [131]9A), and WB
   (Fig. [132]9B) analyses were carried out. In contrast to the control
   group, the expression patterns of gene FTO, NEK7, POLR2H, and S100A12
   were considerably higher in the COPD, LUAD, and COPD with LUAD groups.
   The BNIP3 expression levels were only enhanced in the COPD group, while
   they were lower in the LUAD and COPD with LUAD groups than in the
   control group. In contrast to the control group, the levels of TLR4
   were lower in the COPD, LUAD, and COPD with LUAD groups.

Fig. 8.

   [133]Fig. 8
   [134]Open in a new tab

   Immunohistochemical examination of hub genes in clinical samples (A)
   Expression levels of 6 DPRDEGs detected by immunohistochemistry in
   normal lung tissue, COPD lung tissue, LUAD lung tissue, and the COPD
   with LUAD lung tissue. (B) Statistical analysis of immunohistochemical
   staining scores for each group. “NS” indicates that P ≥ 0.05,
   suggesting no statistical significance. “*” means P < 0.05, “**” means
   P < 0.01, and “***” means P < 0.001, suggesting statistical
   significance. DPRDEGs diagnosis pyroptosis related differentially
   expressed genes, COPD chronic obstructive pulmonary disease, LUAD lung
   adenocarcinoma.

Fig. 9.

   [135]Fig. 9
   [136]Open in a new tab

   Expression pattern of hub genes via qPCR and Western blot analyses (A)
   Expression levels of 6 DPRDEGs detected by qPCR in normal lung tissue,
   COPD lung tissue, LUAD lung tissue, and the COPD with LUAD lung tissue.
   (B) Expression levels of the same 6 DPRDEGs detected through Western
   blot in normal lung tissue, COPD lung tissue, LUAD lung tissue, and the
   COPD with LUAD lung tissue. The original blots are presented in
   (Supplementary Figure [137]S1–[138]S6). “NS” means P ≥ 0.05, suggesting
   no statistical significance. “*” means P < 0.05, “**” means P < 0.01,
   and “***” means P < 0.001, suggesting statistical significance. DPRDEGs
   diagnosis pyroptosis related differentially expressed genes, COPD
   chronic obstructive pulmonary disease, LUAD lung adenocarcinoma.

Discussion

   Globally, COPD and LC are among the leading causes of mortality,
   presenting significant challenges to public health, safety, and
   socioeconomic development. These diseases share several common risk
   factors, including genetic variations, modifications in gene expression
   patterns, and dysregulation of cellular signaling pathways^[139]30.
   Identified factors include mutations in VEGFR1, upregulation of the
   inflammatory factor TNF-α, and abnormal activation of the IL-6/STAT3
   signaling pathway^[140]31,[141]32. Interestingly, COPD has been
   identified as an independent risk factor for LC, with statistics
   indicating that the incidence of LC in patients diagnosed with COPD is
   15.3%^[142]33. The development of LC from COPD may be influenced by
   oxidative stress, DNA damage, and the inflammatory response. However,
   COPD and LC remain distinct lung diseases with different pathogenesis.
   Specifically, COPD is characterized by matrix degradation, impaired
   tissue repair, and mucosal lesions, whereas LC is primarily marked by
   increased tumor cell proliferation, invasion, and metastasis^[143]34.
   Nevertheless, elucidating the pathogenic mechanisms underlying both the
   divergences and convergences between these two diseases will advance a
   holistic understanding of their comorbid progression.

   This study explores the relationship between COPD and LUAD comorbidity
   and elucidated the mechanisms of COPD and LUAD comorbidity from the
   perspective of pyroptosis. This study detected six DPRDEGs and
   developed a multifactorial prognostic model, which demonstrated good
   performance in predicting the prognosis of LUAD.

   In recent years, an increasing number of studies have reported a
   connection between COPD and LUAD. Some perspectives suggest that the
   continuous synthesis of ROS in the lungs of COPD patients results from
   inflammation. These ROS can oxidize and damage DNA, leading to an
   increased DNA repair rate and making epithelial cells more susceptible
   to malignant transformation^[144]35,[145]36. Others propose that the
   aberrant immune microenvironment in COPD patients disrupts the
   regulation of inflammatory cytokines, promoting tumor cell
   proliferation and allowing them to evade immune
   surveillance^[146]37,[147]38. Furthermore, the mechanism underlying the
   comorbidity of COPD and LUAD has been linked to variations in microbial
   communities between healthy and diseased airways, with certain
   bacterial metabolites, such as acetaldehyde and deoxycholic acid,
   acting as carcinogens^[148]39. Although these perspectives focus on
   different aspects, they all highlight the significant role of
   inflammation in the comorbid mechanisms of COPD and LUAD.

   Pyroptosis is closely associated with cellular inflammatory responses.
   In the chronic inflammatory microenvironment of COPD, persistent
   oxidative stress and inflammatory cytokine stimulation activate the
   NLRP3 inflammasome, a key sensor of cellular damage that integrates
   innate immunity with pyroptotic signaling. The role of pyroptosis in
   LUAD pathogenesis is complex and dual-faceted. On one hand, the release
   of inflammatory cytokines mediated by pyroptosis enhances the
   pro-inflammatory tumor microenvironment, thereby promoting tumor cell
   proliferation and metastasis^[149]40,[150]41. On the other hand,
   excessive activation of pyroptosis may induce tumor cell death,
   potentially inhibiting tumor progression under certain
   conditions^[151]42. These findings underscore pyroptosis as a critical
   pathway in the progression from COPD to LUAD. This has resulted in the
   identification of six DPRDEGs: BNIP3, FTO, NEK7, POLR2H, TLR4, and
   S100A12. Among them, BNIP3 expression levels show significant
   correlation with cellular autophagy activity^[152]43. Given that
   autophagy is a key mechanism for maintaining homeostasis, its
   upregulation could offer an effective strategy to counteract excessive
   pyroptosis. This observation indicates that BNIP3 may serve as a
   potential therapeutic target to inhibit COPD-LUAD progression by
   modulating the balance between pyroptosis and autophagy within the
   shared inflammatory microenvironment. FTO acts as a demethylase for RNA
   m6A methylation modification, and abnormal m6A methylation modification
   has been observed in COPD and LUAD. Unlike other genes, FTO is a
   potential biomarker, as it plays a role in the progression of COPD and
   LUAD coexistence via epigenetic pathways^[153]44,[154]45. NEK7
   interacts with with NLRP3 to regulate inflammasome activation,
   affecting inflammation in a pyroptosis-dependent manner. However, no
   reports have linked NEK7 to COPD or LUAD^[155]46. The results
   demonstrated that the expression levels of NEK7 are higher in tissue
   samples from patients with COPD, LUAD, and COPD concomitant with LUAD,
   compared to normal lung tissue. This suggests that NEK7 may serve as a
   potential biomarker. This study is the first to uncover the potential
   significance of POLR2H in the comorbidity of COPD and LUAD. Given its
   diagnostic value in the Prognostic Model, POLR2H may serve as a novel
   biomarker for the early detection of COPD and LUAD comorbidity.
   However, current understanding of POLR2H remains limited, even though
   it encodes a critical subunit of RNA Polymerase II. Previous reports
   have linked the abnormal expression of this gene to the survival rate
   of lung cancer. Its role in lung cancer may be associated with
   functions such as DNA damage repair, gene expression regulation, and
   cell cycle control^[156]47. In contrast, S100A12 is highly expressed in
   COPD and closely associated with inflammatory responses, though fewer
   reports exist regarding its role in LUAD^[157]48. Our study reveals
   that S100A12 expression is also significantly upregulated in tissue
   samples from LUAD and COPD concomitant with LUAD patients, suggesting
   that S100A12 may facilitate tumor cell proliferation and migration in
   the tumor microenvironment. This finding offers new perspectives on the
   potential application of S100A12 as a biomarker.

   The pyroptotic response mediated by TLR4 differs between COPD and LUAD.
   In COPD-related studies, inhibiting the TLR4 signaling pathway has been
   shown to alleviate smoking-induced COPD^[158]49–[159]51. In LUAD,
   however, activating the TLR4/NLRP3/caspase-1/GSDMD-dependent pyroptosis
   pathway has been reported to inhibit the progression of LUAD^[160]52.
   This discrepancy may arise from differences in the types of cells
   involved in the pyroptotic response. In COPD, the goal is to reduce
   pyroptotic damage in macrophages, thereby decreasing inflammation and
   mitigating potential fibrosis^[161]53. In LUAD, by contrast, the
   objective is to directly target tumor cells with drugs to exert
   inhibitory and cytotoxic effects^[162]54. Furthermore, analysis of the
   CTD database revealed putative interactions between BNIP3, FTO, NEK7,
   and POLR2H and existing drugs or small molecule compounds, providing
   novel insights for future clinical translational research. In summary,
   pyroptosis may constitute a common pathogenic mechanism underlying the
   comorbidity of COPD and LUAD, thereby warranting further investigation
   into its specific mechanisms.

   The limitations of this study are as follows. First, the study relies
   on publicly available data from databases, which makes the results
   exploratory in nature. Second, although DPRDEGs were rigorously
   validated using techniques such as IHC, qPCR, and WB, the sample size
   (n = 5 per group) may limit the statistical power of the results. A
   small sample size can introduce variability and restrict the
   generalizability of the findings. Additionally, the molecular
   mechanisms by which these six DPRDEGs contribute to COPD and LUAD were
   not explored in depth. To overcome these limitations, future studies
   will focus on increasing the sample size and further investigating the
   molecular pathways involved, thereby enhancing the clinical relevance
   and broader applicability of the findings.

Conclusions

   In conclusion, this study identified six DPRDEGs and developed a
   prognostic model for DPRDEGs. The expression levels of these six
   DPRDEGs were verified through IHC, qPCR, and WB. The results of this
   study will offer new and valuable information for understanding and
   treating the comorbidity of COPD and LUAD.

Supplementary Information

   [163]Supplementary Information.^ (913.4KB, pdf)

Abbreviations

   COPD
          Chronic obstructive pulmonary disease

   LUAD
          Lung adenocarcinoma

   DEGs
          Differentially expressed genes

   PRGs
          Pyroptosis-related genes

   PRDEGs
          Pyroptosis-related differentially expressed genes

   PPI
          Protein-protein interaction

   CTD
          Comparative toxicogenomics database

   IHC
          Immunohistochemistry

   WB
          Western blotting

   PCV13
          13-Valent pneumococcal conjugate vaccine

   PPSV23
          23-Valent pneumococcal polysaccharide vaccine

   LC
          Lung cancer

   GSDM
          Gasdermin

   TCGA
          The cancer genome atlas

   GEO
          Gene expression omnibus

   DPRDEGs
          Diagnostic pyroptosis-related differentially expressed genes

   GO
          Gene ontology

   BP
          Biological processes

   MF
          Molecular functions

   CC
          Cellular components

   KEGG
          Kyoto encyclopedia of genes and genomes

   SVM
          Support vector machine

   LASSO
          Least absolute shrinkage and selection operator

   DCA
          Decision curve analysis

   TFs
          Transcriptional factors

   PFA
          Paraformaldehyde

   ROC
          Receiver operating characteristic

   AUC
          Area under the curve

   IBD
          Inflammatory bowel disease

Author contributions

   C.C. and S.X. conceived this project. C.C. and Z.Z. took charge of the
   project administration. C.C. and Y.Y. performed experiments, data
   analysis and interpretation. X.X. coordinated patient enrollment and/or
   sample collection. C.C. and B.W. wrote the original draft. S.X., L.K.,
   and X.X. conducted critically review and editing of the manuscript. All
   authors have read and approved the final version of the manuscript.

Funding

   This study was supported by National Natural Science Foundation of
   China (82072008); Liaoning Provincial Department of Education
   Scientific Research Platform Construction Project (LJ232410164021);
   Liaoning Provincial Department of Education Independent Research Topic
   Project (LJ212410164020); Provincial key clinical specialty
   construction project in Liaoning Province (lnslczdzk2024); Scientific
   research project of Shenyang Health Commission (202316).

Data availability

   All sequencing raw data used in this work can be obtained from the Gene
   Expression Omnibus (GEO; [164]https://www.ncbi.nlm.nih.gov/geo/), the
   Cancer Genome Atlas (TCGA) datasets ([165]https://xenabrowser.net/),
   and the GeneCards ([166]https://www.genecards.org/). Additionally, to
   ensure the reproducibility of our findings, all scripts used for data
   preprocessing, differential expression analysis, and model construction
   have been made publicly available. Further inquiries can be directed to
   the corresponding author.

Declarations

Competing interests

   The authors declare no competing interests.

Ethics approval and consent to participate

   The study was approved by the ethic committee of the Second Affiliated
   Hospital of Shenyang Medical College. This research conformed to the
   principles of the Helsinki Declaration.Written informed content was
   obtained from every participant.

Footnotes

   Publisher’s note

   Springer Nature remains neutral with regard to jurisdictional claims in
   published maps and institutional affiliations.

Contributor Information

   Shuyue Xia, Email: syx262@126.com.

   Guixian Xiao, Email: 18002453148@163.com.

Supplementary Information

   The online version contains supplementary material available at
   10.1038/s41598-025-97727-4.

References