Identification of CD14 as a Potential Biomarker Relating to PANoptosis in Chronic Obstructive Pulmonary Disease

Introduction

Chronic obstructive pulmonary disease (COPD) is a chronic inflammatory condition characterized by persistent airway inflammation. It typically progresses over several years, leading to substantial morbidity and mortality.^1–3 Worldwide, the prevalence of COPD among individuals aged 30–79 years ranges from 7.6% to 10.6%.⁴ COPD patients face a heightened risk of developing lung cancer and other comorbidities compared to healthy individuals.⁵ Given this context, early diagnosis and effective intervention are of paramount clinical importance for improving the overall survival rate of COPD patients. Consequently, there has been significant interest in identifying biomarkers for COPD over the past few decades. For example, through single-cell RNA sequencing (scRNA-seq), IGFBP5 serves as a signature gene for ciliated epithelial cells in COPD, playing a pivotal causal role in driving cellular senescence, inflammatory responses, and airway remodeling.⁶ Using mass spectrometry analysis of metabolomics, Richard et al searched for new predictive biomarkers associated with COPD.⁷ Recently, Zhong et al provided a theoretical basis for linking aging effectors to COPD progression.⁸ Therefore, identifying biomarkers associated with the pathogenesis of COPD has emerged as a major research focus in recent years.

Apoptosis, pyroptosis, and necroptosis are recognized as the core pathophysiological mechanisms underlying the onset and progression of COPD.^9–11 As a fundamental process for maintaining tissue homeostasis, apoptosis is triggered by persistent oxidative stress and inflammatory mediators; however, excessive apoptosis leads to the loss of epithelial and endothelial cells, thereby compromising the integrity and function of lung tissue.¹² Pyroptosis is a highly pro-inflammatory form of programmed cell death, typically activated by inflammasomes. It is characterized by persistent cell swelling that leads to plasma membrane rupture and the release of a large amount of inflammatory cytokines, which subsequently exacerbates pathological damage in COPD.^13,14 Necroptosis is a molecularly regulated form of “controlled necrosis”, usually triggered by the RIPK1/RIPK3/MLKL protein complex when apoptosis is inhibited.¹⁵ This leads to cell membrane perforation and the leakage of cellular contents, which likewise induces a strong inflammatory response.¹⁵ Previous studies have demonstrated that within the pathological context of COPD, necroptosis mediates cigarette smoke-induced pulmonary epithelial inflammation by promoting the extracellular release of damaged mitochondrial DNA, serving as a key mechanism driving the progression of COPD.¹⁶ However, apoptosis, pyroptosis and necroptosis do not function as isolated processes.¹⁷ Recent studies have demonstrated intimate signaling crosstalk and synergy among these three pathways. This integrated network, composed of multiple cell death pathways, is defined as PANoptosis, representing a more advanced form of regulated cell death in response to complex environmental stress.¹⁸

PANoptosis is a newly defined and highly integrated programmed cell death mode. It is not merely a simple juxtaposition of apoptosis, pyroptosis, and necroptosis. Instead, it organically integrates key molecules from these three signaling pathways, such as RIPK1 and Caspase-8, through the formation of a large multi-protein complex called the PANoptosome.^19,20 This mechanism allows cells to synergistically trigger multiple death pathways upon exposure to pathogens or inflammatory stimuli, thereby generating a more robust inflammatory response.²¹ PANoptosis has been linked to various inflammatory diseases.^22–24 It has been reported that RIPK1 kinase activity drives activation of the NLRP3 inflammasome and PANoptosis, which could be targeted for the treatment of TAK1-associated myeloid proliferation and septicemia.²⁴ In rats with sepsis-related encephalopathy, inhibiting TLR9 has been indicated to lead to the suppression of the MAPK pathway, thereby suppressing PANoptosis and improving the survival rate of the affected animals.²⁵ In addition, there is growing evidence that PANoptosis is involved in the pathogenesis of pulmonary disease.^26,27 The inhibition of PANoptosis provided protection for mice from COVID-19-induced acute lung damage.²⁸ Yasmine et al demonstrated that stimulators of interferon genes agonists induced acute respiratory distress syndrome by PANoptosis.²⁹ Nisa et al have summarized different patterns of host cell death mediated by Mycobacterium tuberculosis, including PANoptosis.³⁰ Therefore, understanding the mechanisms of PANoptosis is crucial for developing effective therapeutic strategies for various human diseases. However, to the best of our knowledge, reports on COPD-specific PANoptosis are limited. While PANoptosis has been implicated in the pathogenesis of various inflammatory disorders, its clinical significance in COPD remains largely unexplored. The ability of PANoptosis to integrate multiple programmed cell death pathways and drive robust inflammatory responses aligns closely with the persistent chronic inflammation characteristic of COPD. Therefore, this study aims to investigate the connection between these two conditions and identify core PANoptosis-related genes (PANRGs) that may serve as novel diagnostic biomarkers for COPD.

In this study, our primary objective is to identify robust PANoptosis-related hub genes in COPD using integrated bioinformatic approaches, aiming to establish highly reliable diagnostic biomarkers. Subsequently, we explore the correlation between these core genes and the infiltration of the lung immune microenvironment, alongside predicting potential targeted therapeutic agents, to provide a scientific reference for the precision medicine of COPD. The overall analytic workflow is illustrated in Figure 1.

Figure 1 Flowchart for research.

Materials and Methods

Subjects

The human mRNA expression profile datasets GSE8545 (18 COPD samples and 18 control samples), GSE20257 (23 COPD samples and 59 control samples), GSE11784 (22 COPD samples and 72 control samples), and GSE1650 (18 COPD samples and 12 control samples) were downloaded from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). The COPD cohorts specifically included samples from current smokers diagnosed according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) I or II criteria.^31–33 In this study, the control group consists of non-COPD smokers obtained from the public datasets. GEO is a robust, public functional genomics data repository maintained by the National Center for Biotechnology Information, which supports the archiving and distribution of high-throughput gene expression datasets generated by gene chips, next-generation sequencing, and other functional genomic techniques. The datasets GSE8545, GSE20257, and GSE11784 were obtained using the Affymetrix Human Genome U133 Plus 2.0 Array platform. Following data download, probe information was converted into Gene Symbols based on the platform’s annotation files to facilitate subsequent analysis. Besides, a scRNA-seq data set GSE173896 was also downloaded from the GEO database, 5 COPD samples and 2 control samples were selected from GSE173896.

Weighted Gene Co-Expression Network Analysis (WGCNA)

WGCNA was performed based on the expression values of the top 25% genes—screened by median absolute difference—using the “WGCNA” function package (version1.72–1)³⁴ of R software. WGCNA can characterize patterns of genetic association between different samples for the identification of highly cooperative gene modules. Candidate markers can be identified based on the relationship between gene modules and phenotypes. Finally, we assessed the correlation between each module and the pathogenesis of COPD, and selected the significantly associated modules as the central gene modules derived from WGCNA. p<0.05 was considered with significant correlation.

Differential Gene Expression Analysis (DEGs)

DEGs were screened using “limma” package (version 3.52.4)³⁵ of R software with a threshold of |log₂FC|> 0.5 and p.adjust <0.05.

Functional Enrichment Analysis

The enrichment analysis of gene ontology (GO, including biological process (BP), molecular function (MF) and cellular component (CC)) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway were performed using the “clusterProfiler” R package (version 4.7.1.2).³⁶ The significantly enriched GO term and KEGG pathway were screened using p.adjust <0.05 as a criterion.

Screening of the PANoptosis-Related Hub Genes

The hub genes related to PANoptosis in COPD were further screened using the intersection of support vector machine recursive feature elimination (SVM-RFE) and least absolute shrinkage and selection operator (LASSO). The SVM-RFE algorithm, implemented using the “e1071” R package (https://CRAN.R-project.org/package=e1071), was utilized to identify genes with high discriminatory power. Moreover, the LASSO analysis was performed using the “glmnet” package (version 4.1–8) of the R software,³⁷ which represented a regression analysis algorithm applying regularization for variable selection.

Protein–Protein Interaction (PPI) Analysis

The STRING database (version 11.0; https://string-db.org/), a comprehensive resource, was employed to explore PPI and their functional associations.³⁸ STRING is a comprehensive online resource designed to integrate both known and predicted physical interactions and functional associations between proteins, deriving information from experimental evidence, computational prediction methods, and public text mining. We utilized Cytoscape (version 3.7.2; https://cytoscape.org/) to visualize the PPI networks.³⁹

To gain insights into the intricate interactions between proteins and assess potential biological pathways, we employed the online database GeneMANIA (http://www.genemania.org) to look for genes sharing functions.

scRNA-Seq Data Analysis

The processing of scRNA-seq data was carried out using the “Seurath” package (version 5.0.1) in R software.⁴⁰ Initially, the raw scRNA-seq data were filtered to eliminate low-quality single cells. This was followed by normalization of the remaining data using the “NormalizedData” function. Using “RunPCA” function, the principal component analysis (PCA) was conducted on the normalized data. Next, the “FindClusters” function in the “Seurat” package was utilized for unsupervised clustering of the major cell subtypes. The clusters were visualized using “umap” for manual annotation based on cell markers.

Analysis of the Immune Cell Infiltration

To calculate the relative abundances of the 22 immune cell types in every sample, we employed the CIBERSORT software.⁴¹ This software employed a deconvolution algorithm and utilized a predefined set of 547 barcode genes to assess the composition of infiltrating immune cells based on the gene expression matrix. Notably, the sum of all estimated immune cell type proportions in each sample was normalized to 1.

Targeted Drug Prediction for COPD

The Comparative Toxicogenomics Database (CTD, http://ctdbase.org/) offered comprehensive information on the intricate interactions between various chemical exposures (both environment and drugs), genes, proteins, phenotypes and diseases. This database was used to analyze the relationship between the key gene and COPD. Additionally, the Drug-Gene Interaction Database (DGIdb) (version 4.2.0-sha1 afd9f30b) (https://dgidb.genome.wust-l.edu) was employed for predicting drugs that targeted the gene.

Statistical Analysis

The differences in gene expression levels and immune cell infiltration among various groups were determined using the Mann–Whitney U-test. The Pearson correlation analysis was conducted using the “cor” function in the R software. To establish the diagnostic value of the marker gene, receiver operating characteristic (ROC) curves were constructed using the “pROC” R package (version 1.18.4).⁴² Statistical significance was set at p <0.05. All statistical analyses were performed using R software (version 4.3.1).

Results

Identification of DEGs in COPD

To identify potential biomarkers of COPD, differential expression analysis was performed between COPD and control samples using the GSE8545 dataset. After applying stringent criteria of an adjusted p value <0.05 and a |log₂FC | > 0.5, we identified a total of 11,648 DEGs in the COPD group compared to the control group, including 11,510 up-regulated genes and 138 down-regulated genes (Figure 2A and B).

Figure 2 DEG analysis between the COPD and control samples and functional enrichment analysis in the GSE8545 cohort. (A) Volcano plot of DEGs. Red dots indicate up-regulated genes, blue dots indicate down-regulated genes, and gray dots represent genes with no significant difference. (B) Heatmap of the DEGs. (C) The top 10 KEGG pathways enrichment of DEGs. (D) The top 10 GO enrichment analysis of DEGs.

We conducted GO and KEGG pathway enrichment analyses to gain insights into the potential function of these DEGs. The results revealed that 140 KEGG pathways, including Salmonella infection, Human T−cell leukemia virus 1 infection, Autophagy, and Apoptosis, were significantly enriched (Figure 2C), along with 2,154 GO terms. In the BP assessment, DEGs were mostly engaged in ribonucleoprotein complex binding and other functions. DEGs have been localized to the mitochondrial matrix, focal adhesion and other structures in CC. DEGs associated with MF include regulation of cadherin binding, GTPase binging and other functions. The top 10 most significantly enriched GO terms were shown in Figure 2D. Further details of the enrichment results are provided in Table S1.

Construction of WGCNA and Identification of COPD-Related Modules

DEG analysis was conducive to mine potential biomarkers of COPD from the perspective of differences, and the co-expression patterns of genes were also important. To identify the potential gene modules associated with COPD, WGCNA was conducted using the GSE8545 dataset. The soft threshold in this study was 5 (Figure 3A), and a sum of 9 distinct gene modules were revealed (Figure 3B and C). The trait data for WGCNA included sample type (COPD and control), age, and gender. The correlation between each gene module and traits was calculated (Figure 3D). Notably, totally four gene modules were significantly correlated with COPD (p-value <0.05), including pink (Figure 3E), brown (Figure 3F), turquoise (Figure 3G), and green (Figure 3H) modules, which contained a total of 3,583 genes.

Figure 3 Identification of COPD-associated gene modules in the GSE8545 cohort using WGCNA. (A) Determination of the soft-threshold power. (B) Dendrogram of all genes in the GSE8545 dataset based on a topological overlap matrix. The upper half presents the hierarchical clustering dendrogram of genes, while the lower half visualizes the gene modules. Each color represents a distinct module, and the gray module signifies genes that cannot be aggregated to any other modules. (C) Eigengene adjacency heatmap. The gray module was excluded. (D) Heatmap of the correlation between modules and traits. The numbers in each cell indicate the correlation and p value. (E–H) Scatter plots of the correlation of genes with the pink, brown, turquoise, and green modules.

Screening Candidate PANoptosis-Related Hub Genes in COPD

To more accurately identify candidate PANRGs in COPD, we retrieved 66 PANRGs including 26 pyroptosis-related genes, 32 apoptosis-related genes, and 8 necroptosis-related genes from a published study⁴³ (Figure 4A). Next, five overlapping genes including BOK, CD14, LY96, IL1B, and IRF2 were generated by taking the intersection of the 11,648 DEGs, 3,583 WGCNA module-related genes, and 66 PANRGs (Figure 4B). KEGG pathway enrichment was performed to further explore the function of these five overlapping genes in COPD and the results showed that the overlapping genes were significantly enriched in 12 KEGG pathways (Table 1), including Pertussis, NF-kappa B signaling pathway, Tuberculosis and other functions.

Table 1 12 Pathway of KEGG Enrichment Analysis of Five PANoptosis-Related Genes

Figure 4 Screening candidate PANoptosis-related hub genes. (A) The network of downloaded 66 PANRGs from a published study. (B) Venn diagram of the WGCNA modules, DEGs, and PANRGs. (C) The cross-validation curve for the LASSO regression analysis. (D) The results of the screening of diagnostic markers by the SVM-RFE algorithm. (E) Venn diagram of LASSO and SVM-RFE algorithms. (F) PPI network interaction map of three hub genes including BOK, CD14, and IRF2, where a node represents a gene, and the edges represent their interrelationships. (G) Network interaction map of three hub genes based on GENEMANIA. (H) Correlation heatmap among the identified candidate hub genes.

To further narrow down the pool of candidate PANRGs, LASSO regression and SVM-RFE algorithm were conducted. The LASSO regression algorithm successfully identified three genes (Figure 4C), while the SVM-RFE revealed five genes (Figure 4D). Finally, three genes including BOK, CD14, and IRF2 were generated by intersecting the genes identified through both algorithms (Figure 4E).

A PPI network was constructed for these three hub genes using the STRING database to examine inter-gene relationships. Interaction pairs were filtered based on a minimum required interaction score > 0.15. Subsequently, the PPI network was visualized using Cytoscape software (Figure 4F). Additionally, the interplay networks for the hub genes were constructed using GENEMANIA (Figure 4G and Table S2). A total of 367 interaction pairs were identified, including 80 co-expression, 12 co-localization, 20 genetic interactions, 138 physical interactions, 46 shared protein domains, 37 pathways, and 34 predicted, implying the complex crosstalk among the hub genes in COPD.

Correlation analysis revealed significant positive correlations among all three hub genes (Figure 4H). Additionally, KEGG pathway enrichment analysis was conducted on the three hub genes, revealing that they were enriched in 8 KEGG pathways (Table 2), including Apoptosis-multiple species, Pertussis, and NF-kappa B signaling pathway. These results suggested that BOK, CD14, and IRF2, as candidate PANoptosis-related hub genes, may play a key role in the occurrence and progression of COPD.

Table 2 8 Pathway of KEGG Enrichment Analysis of Three Hub Genes

Identification of CD14 as a PANoptosis-Related Key Gene in COPD

To ensure the reliability of the hub genes identified above, we validated their expression in various datasets. Initially, we analyzed the expression of the three hub genes using bulk datasets GSE8545, GSE20257, and GSE11784. We found that all three candidate genes were significantly up-regulated in COPD samples compared to control samples in the GSE8545 cohort (Figure 5A). Validation in the GSE20257 and GSE11784 datasets confirmed that CD14 was significantly up-regulated in COPD samples; however, BOK and IRF2 showed no significant differential expression (Figure 5B and C).

Figure 5 Identification of CD14 as a PANoptosis-related key gene in COPD. (A-C) Violin plots of hub genes’ expression in COPD and control samples from GSE8545 (A), GSE20257 (B), and GSE11784 cohorts (C). (D) Violin plots of candidate hub genes’ expression in late-stage COPD and early-stage COPD samples in GSE5060 cohort. (E) 21 cell clusters were identified based on scRNA-seq data of 5 COPD and 2 control samples from GSE173896 dataset. (F) Classification results of cell clusters based on the umap algorithm and 11 immunological cell types were classified. (G–I) The expression of CD14, IRF2, and BOK in 11 immunological cell types. Statistical significance: **P < 0.01, ***P < 0.001; ns, not significant.

Additionally, to partly evaluate the contribution of hub genes to COPD progression, we compared the expression of the three hub genes in late-stage and early-stage COPD samples using the GSE5060 dataset. The results revealed that BOK expression was significantly higher in late-stage COPD samples compared to early-stage COPD samples. Meanwhile, CD14 and IRF2 both tended to exhibit higher expression in late-stage COPD samples, although the differences did not reach statistical significance (Figure 5D). We analyzed the correlation among these three genes in COPD patients at different stages. The results showed that the correlation was more pronounced in late-stage patients (Figure S1), while it was not significant in the early stage. This suggests that the potential interplay among these genes may be associated with disease progression.

Subsequently, the expression of genes BOK, CD14, and IRF2 was also verified from the single cell resolution using the GSE173896 scRNA-seq dataset. We performed PCA using 5000 variable genes to reduce the dimensionality (Figure S2A). A total of 21 cell clusters were identified (Figure 5E), and 11 types of cells were annotated (Figure 5F), and the proportion of 11 cells in COPD and control samples were shown in Figure S2B. The results showed that the expressions of BOK, CD14, and IRF2 were relatively low in both COPD and control samples (Figure S2C) in the GSE173896 dataset. Thus, we analyzed the hub genes expressions in various cells. Among the immune cells, we noticed that CD14 and IRF2 exhibited distinct expression patterns across different immune cell types (Figure 5G and H), while BOK expression did not vary significantly (Figure 5I). Collectively, considering the consistent up-regulation across different levels, CD14 was chosen as the key target in our subsequent studies.

Distinct Immune Cell Infiltration Between High and Low CD14 Expression COPD Patients

Since the differential expression of CD14 in different immune cells was observed in the single-cell validation, we further refined the patient immune landscape to analyze the relationship between CD14 and immune cells in COPD patients. We used the CIBERSORT algorithm to calculate the proportion of 22 immune cells in the GSE8545 cohort (Figure 6A). Based on the median expression value of CD14, the COPD samples in the cohort were divided into high CD14 expression group and low CD14 expression group. The differences in immune cell infiltration between these two groups were then analyzed. The results revealed that the proportions of plasma cells, T cells CD4 memory resting, and mast cells resting were significantly increased in the low CD14 expression group in contrast to the high CD14 expression group. On the contrary, the high CD14 expression group showed a higher infiltration proportion of monocytes (Figure 6B).

Figure 6 Evaluation of immune cell infiltration analysis in COPD. (A) Stacked plot of 22 immune infiltrating cells in each sample. (B) Box plot of 22 immune cell infiltration between high and low CD14 expression groups. (C) Scatter plot of the correlation between CD14 expression and significantly different immune cells. (D) Box plot of 22 immune cells infiltration between late-stage and early-stage COPD samples. Statistical significance: *P < 0.05, ***P < 0.001.

To further investigate the relationship between CD14 and these significantly different infiltrating immune cells, the Pearson correlation analysis was performed. The results showed a significant negative correlation between CD14 and mast cells resting, T cells CD4 memory resting, plasma cells, and a significant positive correlation with monocytes (Figure 6C).

Going a step further, we calculated the proportion of immune cell infiltration between late-stage and early-stage COPD samples and observed that the proportion of T cells CD4 memory resting in late-stage COPD samples was significantly lower than in early-stage COPD samples (Figure 6D). These results suggested that the role of CD14 in COPD inflammation may be mediated through the regulation of immune cells.

Mining of the Clinical Value of CD14 in COPD

To assess the diagnostic value of CD14 in COPD, the ROC curves were plotted using the GSE8545, GSE20257, GSE11784 and GSE1650 datasets. The results demonstrated Area Under the Curve (AUC) values of 0.756, 0.702, 0.703 and 0.732 respectively across these four cohorts (Figure 7A–D). Furthermore, the ROC curves for early-stage and late-stage COPD samples were constructed utilizing the GSE5060 dataset, resulting in an AUC value of 0.759 (Figure 7E). These findings indicate that CD14 exhibits favorable diagnostic performance as a biomarker for COPD.

Figure 7 Clinical value mining and drug prediction of CD14. (A–E) The ROC curves of GSE8545, GSE20257, GSE11784, GSE1650 and GSE5060 cohorts. (F) Inference score radar map of CD14 in COPD identified by CTD database. (G) CD14 target drug network predicted by DGIdb database.

In order to further validate the association between CD14 and COPD, gene-inferred scores from the CTD database were analyzed which confirmed their correlation (Figure 7F). Moreover, potential drugs targeting CD14 were predicted using the DGIdb database leading to the identification of three candidate drugs including VB-201, LOVASTATIN, and IC143 (Figure 7G).

Discussion

In this study, we compared COPD and control samples with the aim of identifying more candidate diagnostic biomarkers associated with PANoptosis in COPD. Five genes including BOK, CD14, LY96, IL1B, and IRF2 were identified through the overlap of DEGs, WGCNA modules, and PANRGs.⁴³ Subsequent KEGG enrichment analysis found that these five genes were enriched to Pertussis, NF-kappa B signaling pathway, Tuberculosis and other KEGG pathways. It is worth noting that respiratory infections were more prevalent in COPD patients and Pertussis has been shown to exacerbate COPD progression.⁴⁴ The redox-sensitive transcription factor, nuclear factor kappa B played a crucial role in regulating cytokine activity within airway pathology and has been implicated in airway inflammation associated with asthma and COPD.⁴⁵ Yakar et al reported that Tuberculosis led to increased hospitalization rates and decreased respiratory function among COPD patients, COPD diagnosis and death occurred 5 years earlier in patients with Tuberculosis.⁴⁶ Therefore, we believe that the PANRGs were involved in the inflammatory responses of COPD. We further identified three candidate genes including BOK, CD14, and IRF2 using LASSO regression and the SVM-RFE algorithm. Our stratified analysis by disease stage revealed that the positive correlations among CD14, IRF2, and BOK were specifically significant in late-stage COPD, whereas no such associations were observed in early-stage patients. This suggests that PANoptosis-related pathways may undergo potential dynamic recruitment during the late stages of COPD. Validation using three additional independent datasets consistently highlighted CD14 as a key gene associated with COPD PANoptosis, showing significantly higher expression in COPD samples.

CD14 acts as a pivotal surface antigen mediating LPS-induced inflammation by orchestrating the activation of TLR4-dependent immune pathways.^47,48 As an early regulator of oxPAPC-induced hyperactivation, CD14 predominantly mediates non-infectious inflammatory responses.⁴⁹ Multiple studies have reported elevated CD14 expression in patients experiencing various acute infectious processes. For instance, CD14+ monocytes increased significantly in Controlling latent tuberculosis infection compared with healthy controls.⁵⁰ CD14 was up-regulated in acute lung injury associated with sepsis, and inhibition of CD14 expression could reduce the injury in patients.⁵¹ In COVID-19, plasma concentrations of sCD14 increased significantly with the severity of the disease.⁵² In addition, CD14 had been reported to be involved in periodontitis⁵³ and influenza.⁵⁴ Nonetheless, recent studies have unveiled its more intricate regulatory roles in maintaining cellular and tissue immune homeostasis. Evidence suggests that membrane-bound CD14 on macrophages plays a critical role in the clearance of apoptotic cells by recognizing outer membrane phospholipids.⁵⁵ Notably, recent proteomic evidence indicates that rather than functioning as a direct receptor for phosphatidylserine, the classical apoptotic signal, CD14 recognizes externalized phosphoinositides (PIPs) as the authentic “eat-me” signals.⁵⁶ It is significant that PIPs are externalized and subsequently recognized by CD14 regardless of whether the cell undergoes apoptosis, ferroptosis, or necroptosis.⁵⁷ Given that the crosstalk and coexistence of these diverse cell death modes constitute the foundation of PANoptosis, these findings may elucidate the potential reasons for the significant elevation of CD14 within the PANoptosis-active pathological milieu of COPD. These findings suggest that CD14 levels may, to some extent, serve as a potential indicator reflecting the scale of PANoptosis in the lung tissue of COPD patients. This could potentially assist clinicians in performing a preliminary assessment of parenchymal lung damage activity without necessarily resorting to invasive biopsy.

Next, we examined whether CD14 has the potential to be a diagnostic biomarker for COPD. Our findings revealed that the proportion of immune cell infiltration, including T cells CD4 memory resting, plasma cells, and mast cells resting, were significantly lower in the CD14 high expression group compared to the CD14 low expression group. Among them, CD4 T cells play an important role in the immune system, protecting tissues from reinfection and cancer, and are also involved in allergy, autoimmunity, graft rejection, and chronic inflammation.⁵⁸ Mast cells, as innate immune cells resident in tissues, play a key role in the inflammatory response and tissue homeostasis.⁵⁹ It is well known that inflammation occurs first in COPD. Based on these results, we hypothesized that CD14 regulates the COPD process by participating in the inflammatory response. To make the results more convincing, we further succeeded in demonstrating that the proportion of immune infiltrates of T cells CD4 memory resting in the late COPD cohort was also significantly lower than that in the early COPD cohort. In addition, the diagnostic value of CD14 was evaluated by ROC curve analysis, and the results showed that CD14 had a good diagnostic effect on multiple data sets. This means that CD14 may be used as a diagnostic marker for COPD in clinical practice. Compared with established COPD biomarkers such as SP-B reported by D’Ascanio et al, CD14 demonstrates unique diagnostic potential.⁶⁰ SP-B is an organ-specific protein synthesized exclusively by alveolar type II cells, and its elevated plasma levels are primarily thought to indicate structural compromise of the alveolar-capillary barrier and direct parenchymal injury.⁶⁰ However, our findings suggest that CD14 might be a more pivotal immunological driver in the progression of COPD. Unlike SP-B, which largely reflects anatomical damage and the resulting protein leakage, CD14 was identified through high-resolution scRNA-seq and appears to function as a central hub orchestrating innate immune responses and chronic inflammation. Consequently, CD14 may represent a promising biomarker that possesses deeper mechanistic support.

This study represents a translational research effort based on multi-omic integration and machine learning algorithms, aiming to systematically identify key regulatory factors associated with PANoptosis in COPD. Although we successfully identified and validated the potential of CD14 as a critical diagnostic biomarker, several limitations remain to be addressed. The current research relies primarily on the retrospective analysis of public datasets and lacks validation from large-scale prospective clinical cohorts; furthermore, our findings highlight the potential of CD14 as a diagnostic biomarker, the current study is descriptive and correlational in nature. We have not yet provided direct experimental evidence to demonstrate the precise clinical repercussions caused by CD14 dysregulation. Whether the elevated expression of CD14 is a driver of lung function decline or merely a molecular byproduct of the PANoptotic environment remains to be elucidated. Future mechanistic studies involving gene knockdown or overexpression in animal models are necessary to clarify the causal impact of CD14 on COPD progression and its associated clinical outcomes.

Conclusions

In summary, this study systematically identified and validated the signature of PANoptosis-related hub genes, notably CD14, which exhibited significantly high expression in human COPD. By integrating multiple transcriptomic datasets and utilizing advanced machine learning algorithms, we demonstrated that these biomarkers possess high diagnostic accuracy, as further confirmed by ROC curve analysis. Furthermore, our PANoptosis-based model provides a more robust and comprehensive framework for understanding the dysregulation of the lung immune microenvironment. While these findings offer novel insights into the precision diagnosis of COPD, further large-scale prospective clinical trials and mechanistic experiments (both in vivo and in vitro) are warranted to validate their therapeutic potential and to elucidate the specific regulatory axes of these hub genes.