Construction of a PANoptosis-Related Gene Signature for Diabetic Nephropathy Open Access

Diabetic nephropathy (DN) is a microvascular complication of diabetes that affects the kidneys, representing a leading cause of chronic kidney disease and end-stage renal disease globally [1]. DN affects approximately 25–40% of individuals with type 1 diabetes after 20–25 years and 20–40% of those with type 2 diabetes over time [2]. Early signs, like microalbuminuria, are seen in around 20–40% of all diabetics [3], while approximately 2% progress to the more concerning macroalbuminuria annually [4]. With a global diabetic population exceeding 500 million, over a hundred million individuals are at risk for or have already developed DN [5]. Early detection of DN is challenging because symptoms often appear until significant damage has occurred [6, 7]. The lack of specific biomarkers further complicates early identification and intervention, making timely treatment and management difficult [8].

Inflammation plays a role in the pathogenesis of DN. The immune microenvironment in DN is characterized by chronic inflammation driven by hyperglycemia, including the accumulation of proinflammatory macrophages, activation of inflammatory pathways, and increased production of cytokines, all contributing to renal damage and fibrosis in DN [9]. Furthermore, this cascade also primes the environment for PANoptosis, a unique form of programmed cell death integrating pyroptosis, apoptosis, and necroptosis. Pyroptosis is a highly inflammatory, gasdermin-mediated lytic process initiated by inflammasome activation; apoptosis is a caspase-dependent, non-lytic mechanism that enables controlled cell removal; and necroptosis is a regulated necrotic form of death driven by RIPK1, RIPK3, and MLKL. PANoptosis represents an integrated cell death program that combines features of all three pathways through the assembly of a multiprotein PANoptosome complex, which enables their coordinated activation under conditions of immune or metabolic stress [10]. This process is triggered by the activation of inflammasome molecules such as NLR family pyrin domain containing 3 (NLRP3) and absent in melanoma 2, which are involved in the innate immune response to microbial infection and changes in cellular homeostasis [11‒13]. Studies have shown that high glucose levels can activate the NLRP3 inflammasome, contributing to kidney damage in metabolic-associated kidney diseases [14‒16]. On the other hand, pyroptosis, apoptosis, and necroptosis can contribute to the loss of kidney cells, the development of fibrosis, and the stimulation of inflammatory responses in the kidney [17]. Given these insights, building a risk assessment model based on PANoptosis-related genes could facilitate earlier detection and intervention, potentially enabling more targeted therapeutic strategies in DN.

In this study, we conducted a comprehensive analysis to identify differentially expressed genes (DEGs) associated with DN. Focusing on PANoptosis-related DEGs, we developed a risk assessment model comprising pyruvate dehydrogenase kinase 4 (PDK4), tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein theta (YWHAH), and protein kinase X-linked (PRKX). This model exhibited high accuracy in distinguishing between DN and control samples and suggested clinical relevance. Our study also explored immune differences, risk stratification, and potential therapeutic targets, offering insights for better understanding and managing DN.

Transcriptomic expression profiles were obtained from the Gene Expression Omnibus database at https://www.ncbi.nlm.nih.gov/geo/, including GSE96804, GSE142025, and GSE30122. The GSE96804 dataset comprises transcriptomic profiles from renal glomeruli of 41 type 2 DN patients and 20 controls. The DN group included both early-stage (n = 20, class I–IIa) and late-stage (n = 21, class III–IV) cases based on biopsy-confirmed glomerular pathology and estimated glomerular filtration rate criteria [18]. GSE96804 was used in all analyses except for external data validation. GSE142025 contains RNA-seq data from human renal biopsies (28 DN and 9 control samples) [19] and was used to validate differential expression and model performance. GSE30122 includes microarray-based expression profiles of 19 DN patients and 50 controls and served as an additional external validation dataset [20, 21]. Probe names were mapped to gene symbols. When multiple entries corresponded to the same gene, the median expression value was used to represent the gene’s expression level. The datasets were normalized using normalizeBetweenArrays. Sample information and data sources are summarized in Table 1.

Differential gene expression analysis between DN and control groups was conducted using the R package limma. GSE96804 and GSE142025 datasets were employed to identify DEGs in DN. The criteria for DEG identification were set as p value <0.05 and |logFC| > 0.5. The common DEGs in both datasets were determined and defined as DEGs_com. PANoptosis-related genes were obtained from a previous study [22]. The intersection of DEGs_com and PANoptosis-related genes was defined as PANoptosis-related DEGs.

DEGs_com underwent Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional enrichment analyses using the R package “clusterProfiler.” p values were adjusted using the Benjamini-Hochberg method.

The expression profiles of PANoptosis-related DEGs were obtained from each patient in the GSE96804 dataset. Negative matrix factorization (NMF) was employed for the clustering analysis of DN using the R package “NMF,” along with the Brunet method as the specific algorithmic approach. The clustering process was repeated 50 times with the number of clusters ranging from 2 to 10. The optimal number of clusters was determined as 2 based on cophenetic correlation, dispersion, and silhouette scores. The two clusters were visualized using the R package “Rtsne” and principal component analysis (PCA).

The infiltration of 22 immune cell types in the GSE96804 dataset was evaluated using the “CIBERSORT” algorithm, excluding immune cells with zero abundance in more than half of the samples. Stacked bar charts and box plots were generated to visualize the differences in immune cell populations across different groups in GSE96804.

Candidate genes relevant to DN were identified based on PANoptosis-related DEGs in the GSE96804 dataset. Gene ranking was carried out using the Random Forest algorithm, with genes having a Gini coefficient greater than 1 considered significant features. Then, support vector machine-recursive feature elimination (SVM-RFE) was performed using the radial basis kernel function “SVM” from the R package “caret.” Specifically, within the “rfeControl” settings, “functions = caretFuncs” were specified to facilitate feature selection. Within the “rfe” function, method = “svmRadial” was used to determine the most informative features. The feature genes were defined by intersecting the genes obtained through both approaches.

The performance of risk scores across different datasets was assessed using the R package “pROC.” Patients were stratified into high- and low-risk groups based on the median risk score using the “clusterProfiler” package. Gene Set Enrichment Analysis (GSEA) was conducted to investigate the enrichment of GO functions and KEGG pathways.

A nomogram was constructed based on the feature genes using the R package “rms.” The clinical decision-making performance of the nomogram was assessed through calibration curves utilizing the “calibrate” function from the R package “rms.”

To investigate the interaction relationships between feature genes and functionally related genes, a protein-protein interaction (PPI) network was constructed using “GeneMANIA” (http://www.genemania.org).

The interaction between drugs and model genes was explored using the drug-gene interaction database (DGIdb, www.dgidb.org). The interaction network was visualized using Cytoscape software.

Quantitative PCR (qPCR) validation was conducted to confirm the expression patterns of feature genes identified through transcriptomic analysis. Blood samples were collected in 2-mL anticoagulant tubes from healthy volunteers (control group, n = 10) and patients with diabetic kidney disease (DKD) (experimental group, n = 10) recruited from The Second Norman Bethune Hospital of Jilin University (Changchun, Jilin, China). DKD was diagnosed based on American Diabetes Association standards, defined by at least two urine samples showing albumin excretion ≥30 mg/24 h or an albumin-to-creatinine ratio ≥30 mg/g [23]. Participants aged <30 or >75 years, or those with urinary infections, cardiovascular diseases, pregnancy, severe liver failure, or chronic inflammatory conditions, were excluded. Following collection, samples were centrifuged at 3,000 rpm for 5–10 min, and the supernatant was stored at −80°C. The study adhered to the Declaration of Helsinki and was approved by the Institutional Ethics Committee, with informed consent obtained from all participants.

RNA was extracted using TRIzol reagent (Pufei, Shanghai, China) and quantified using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA). cDNA synthesis was performed using the M-MLV Reverse Transcriptase kit (Promega, Madison, WI, USA). Gene-specific primers, synthesized by GeneChem Co., Ltd. (Shanghai, China) were as follows: YWHAH (forward: CCG​CTA​CTT​AGC​AGA​GGT​CG, reverse: TGG​CAT​CAT​CGA​AGG​CTT​GT), PRKX (forward: TCA​TCT​GTG​CCA​TCG​AGT​A, reverse: AGG​GTC​CAA​GTC​CTG​TCT​A), PDK4 (forward: TCA​CAT​CGT​GTA​TGT​TCC​T, reverse: TAT​AAC​TAA​AGA​GGC​GGT​C), and ACTB as the internal control (forward: GCG​TGA​CAT​TAA​GGA​GAA​GC, reverse: CCA​CGT​CAC​ACT​TCA​TGA​TGG). qPCR was performed using SYBR Green Master Mix (Takara, Dalian, Liaoning, China) on a Roche LightCycler 480 II instrument (Roche Diagnostics, Basel, Switzerland). Relative expression levels were calculated using the 2^−ΔΔCt method, normalized to ACTB, and compared between groups.

All statistical analyses were carried out using R (v4.3.0). The R packages “FactoMineR” and “factoextra” were employed for performing PCA and visualizing the results. Heatmaps were generated using the “pheatmap” package. Venn diagrams were produced using the “ggvenn” package. Receiver operating characteristic (ROC) curves were plotted using the “pROC” package. The “ggpairs” package in R was employed for conducting and visualizing the analysis of correlation among feature genes. The “RCircos” package was used for visualizing the genomic positions of genes on chromosomes. The graphical representations of the results were generated using “ggplot2” or “plot” unless specific descriptions were provided. Correlation analysis was performed using the Pearson method. Differences in continuous and categorical variables between two groups were assessed using the Wilcoxon test and the chi-squared test, respectively. A p value less than 0.05 was considered statistically significant.

To identify DEGs associated with DN, we performed differential gene expression analyses on the GSE96804 and GSE142025 datasets. By intersecting the DEGs identified in both datasets (Fig. 1a, b), we found 573 shared DEGs_com, comprising 242 upregulated (Fig. 1c) and 331 downregulated genes (Fig. 1d). GO and KEGG analyses revealed that these genes were involved in GO terms such as “organic anion transport,” “collagen-containing extracellular matrix,” and “sulfur compound binding,” as well as the “TCA cycle” and “fatty acid degradation” pathways (Fig. 1e, f). All enriched GO terms and pathways are summarized in online supplementary Tables S1 and S2 (for all online suppl. material, see https://doi.org/10.1159/000546764).

To identify PANoptosis genes related to DN, we intersected PANoptosis-related genes with DEGs_com. A Venn diagram showed an overlap of six PANoptosis-related DEGs (Fig. 2a), including YWHAH, UNC5B, VIM, PRKX, PDK4, and MAPT. The positions of these six genes on the chromosomes are displayed in Figure 2b. Pearson correlation analysis revealed varying degrees of pairwise correlations among the six genes (Fig. 2c). For example, YWHAH and UNC5B had a positive correlation of 0.61, whereas PRKX and MAPT had a negative correlation of −0.72. These data suggest potential co-regulation or functional interactions among these genes.

To identify clusters based on the expression profiles of the six PANoptosis genes, we used NMF for consensus clustering in the 41 patients from the GSE96804 dataset. The best rank value was determined by identifying the point before the most pronounced shift in the cophenetic value as the cluster number (K) changes. The results showed that the most significant deviation in the cophenetic value occurred when the rank was between 2 and 3. Therefore, the optimal rank value was chosen as 2 (Fig. 2d). As shown in Figure 2e, the six PANoptosis genes categorized DN patients into two groups: cluster1 and cluster2. PCA and t-SNE visualizations of the distribution of the two clusters are shown in Figure 3f and g, respectively. The expression patterns of the six genes across different clusters are displayed in Figure 3h. These data suggest that the six identified PANoptosis-related DEGs enable the categorization of DN patients into two distinct clusters, emphasizing potential co-regulation or functional interactions among these genes in DN.

To characterize the immune differences between DN patients and controls, we compared the abundance of immune cells in the GSE96804 dataset. After screening out immune cells with lower abundance, 5 out of the remaining 12 high-abundance immune cells showed significant differences between DN patients and controls. The distribution of immune cells in different samples is shown in Figure 2i. The abundance of resting dendritic cells (DCs), M1 macrophages, and M2 macrophages was significantly increased in the DN group. Conversely, the abundance of activated mast cells and neutrophils was significantly decreased in the DN group compared to the control group (Fig. 2j). These immune cell abundance differences may help understand the immune profile of DN patients and explore potential therapeutic targets.

To identify PANoptosis-related feature genes for DN, we employed two machine learning algorithms to select feature genes from the six PANoptosis-related DEGs. Using the Random Forest algorithm, we observed that the model reached its lowest error rate when the “mtry” value was set to 1 (Fig. 3a). The error rate of the model tended to stabilize when the number of trees, “nTree,” exceeded 200 (Fig. 3b). Genes with a Gini coefficient greater than 1 were considered feature genes, and all six candidate genes exhibited Gini coefficients exceeding this threshold (Fig. 3c). On the other hand, the SVM-RFE method demonstrated optimal model accuracy when the number of feature genes was three. By intersecting the feature genes identified by both algorithms, three genes were considered risk model genes, namely, PDK4, YWHAH, and PRKX (Fig. 3e).

To identify a PANoptosis-related gene signature for DN, we conducted a LASSO regression analysis. All three genes selected by the machine learning algorithms exhibited non-zero coefficients (Fig. 3f, g). We then integrated the LASSO coefficients with the expression levels of these genes to calculate the risk score for each patient. To evaluate the reliability of the risk score for DN, we conducted ROC analyses, which yielded impressive area under the curve values. Specifically, we obtained area under the curve values of 1 for the training set GSE96804, 0.988 for the external dataset GSE142025, and 0.806 for GSE30122 (Fig. 3h–j). To further evaluate the clinical application of the risk score, we developed a nomogram by combining the risk score with gender to estimate the likelihood of DN onset. In this nomogram, the red number indicates the estimated probability of DN onset (Fig. 3k). To validate the accuracy of our nomogram, we generated calibration curves that demonstrated a close alignment between the predicted and observed probabilities (Fig. 3l). Furthermore, we observed that the risk scores in all datasets consistently exhibited higher values in the DN group compared to those in the control group (Fig. 3m–o). These findings highlight the potential of the PANoptosis-related gene signature in improving the understanding and management of DN.

Based on the median risk score, we categorized patients in the GSE96804 dataset into high-risk and low-risk groups. GSEA revealed differences in enriched GO terms and KEGG pathways between the two groups. In general, in the high-risk group, there was a prominent enrichment in biological processes and pathways related to cellular transport activities and metabolic pathways (Fig. 4a, b). These findings suggest distinct molecular and functional characteristics that define the high-risk group.

To explore the potential association between risk score and the immune microenvironment, we employed the CIBERSORT algorithm to assess immune cell infiltration levels in DN patients from the GSE96804 cohort. The results demonstrated significant differences in the infiltration levels of four immune cell types between different risk groups. Specifically, high-risk DN patients exhibited a pronounced increase in M2 macrophages but a decline in neutrophils, activated NK cells, and CD4+ memory resting T cells compared to low-risk patients (Fig. 4c, d).

With the GSE96804 cohort being divided into two clusters through NMF, we analyzed the risk scores across the two clusters. Cluster 1 had elevated risk scores compared to cluster 2 (Fig. 4e). A more substantial proportion of cluster1 samples were categorized under the high-risk group than those in cluster 2 (Fig. 4f). Furthermore, the correlation analysis revealed that the strongest positive correlation existed between the risk score and M2 macrophages, while the most pronounced negative correlation was observed with neutrophils (Fig. 4g). Taken together, these data suggest a close relationship between risk stratification and variations in the immune microenvironment in DN patients.

To investigate the relationship among the three model genes, we utilized GeneMANIA to create a PPI network (Fig. 5a). In the GSE96804 dataset, PDK4 exhibited a negative correlation with both PRKX and YWHAH, while PRKX and YWHAH displayed a positive correlation with each other (Fig. 5b). Additionally, we observed a reverse correlation pattern between PDK4 and YWHAI/PRRX with immune cells in DN patients. Specifically, for neutrophils and activated mast cells, PDK4 displayed a strong positive correlation, whereas YWHAI and PRRX exhibited a negative correlation. In the case of M2 macrophages, PDK4 showed a robust negative correlation, whereas YWHAI and PRRX displayed a robust positive correlation (Fig. 5c). Consistently, when analyzing external datasets GSE142025 and GSE30122, we found that PDK4 was significantly decreased in DN patients, whereas PRKX and YWHAH were significantly increased in DN patients compared to the control group, except for a nonsignificant change of PDK4 in the GSE30122 dataset between the two groups (Fig. 5d, e). These data suggest a complex interplay among the three model genes and their associations with immune cell patterns in DN.

To study the therapeutic implications of the model genes, we gathered the information from the DGIdb database. PRKX was connected with various compounds, including 681640, PF-562271, dovitinib, cenisertib, GW843682X, PD-0166285, Y-27632, and palbociclib. PDK4 was connected to VER-246608. These data suggest that PRKX and PDK4 are promising therapeutic targets for DN (Fig. 5f).

qPCR analysis was conducted to validate the expression levels of the three model genes, YWHAH, PDK4, and PRKX, in plasma samples. The results showed a significant upregulation of YWHAH in DKD patients compared to the control group, consistent with findings from transcriptomic and external dataset analyses (Fig. 5g). However, the expression levels of PDK4 and PRKX were extremely low in plasma samples, falling below the detection limit of the instrument.

In this study, we developed a PANoptosis-related gene signature for DN comprising PDK4, YWHAH, and PRKX. This gene signature exhibited high accuracy in distinguishing DN from non-DN conditions, offering the potential for enhancing the understanding of the disease. Risk stratification based on the gene-derived risk score classified DN patients into high-risk and low-risk groups, revealing significant correlations between risk levels and immune cell infiltration patterns. Furthermore, our analysis identified therapeutic drug candidates targeting PRKX and PDK4, highlighting their potential as novel therapeutic targets for DN. These findings provide valuable insights into the molecular and immune characteristics of DN and propose potential avenues for treatment. While qPCR validation confirmed the upregulation of YWHAH in plasma samples from DKD patients, the detection of PDK4 and PRKX was limited by low expression levels in plasma, suggesting the need for further validation using alternative sample types or more sensitive detection methods.

The identified DEGs associated with DN are linked to organic anion transport, collagen-containing extracellular matrix, sulfur compound binding, TCA cycle, and fatty acid degradation, suggesting metabolic and structural alterations in diabetic kidneys. Indeed, the diabetic kidney undergoes metabolic and structural changes. Metabolically, the kidneys are responsible for glucose homeostasis, and mitochondrial damage and abnormalities in tubuloglomerular feedback may contribute to DN [24]. Structurally, DN is characterized by thickening of the glomerular basement membrane, expansion of the mesangial matrix, nodular glomerulosclerosis, and arteriolar hyalinosis. Additionally, changes in the tubulointerstitium result in alterations in glomerular filtration rate and urinary protein excretion [25]. Our results underscore the intricate interplay between metabolic and structural changes in the diabetic kidney.

The identification of six PANoptosis-related DEGs suggests that PANoptosis may play a crucial role in the pathogenesis of DN. Their ability to classify DN patients into distinct groups emphasizes their potential for disease stratification. In DN, upregulated UNC5B (Unc-5 netrin receptor B), a specific protein receptor involved in axon guidance and neural development, promotes angiogenesis. Reducing endogenous UNC5B is a potential strategy to mitigate early angiogenesis and kidney injury in DN [26]. VIM (vimentin) is a type III intermediate filament protein. In DN, VIM expression is upregulated in podocytes that have undergone epithelial-to-mesenchymal transition, which is associated with podocyte detachment and loss, serving as a potential therapeutic target to prevent podocyte loss and slow the progression of DN [27]. Additionally, VIM upregulation in patients with both nonproliferative retinopathy and DN, compared to nonproliferative retinopathy alone, suggests its potential as an early DN biomarker in DR patients [28]. In DN, MAPT (the microtubule-associated protein Tau) is hyperphosphorylated, leading to microtubule destabilization and podocyte injury. This destabilization of microtubules can cause podocyte foot process effacement, proteinuria, and glomerulosclerosis, which are all hallmarks of DN [29]. These gene-specific alterations suggest that PANoptosis, a highly integrated form of programmed cell death involving pyroptosis, apoptosis, and necroptosis, contributes to the progression of DN. Unlike isolated activation of individual death pathways, PANoptosis is characterized by the coordinated engagement of multiple cell death mechanisms. In the diabetic kidney, chronic inflammatory stress, metabolic dysregulation, and oxidative injury create a permissive environment for such integrated responses. The dysregulation of PANoptosis-related genes in DN supports the notion that this unified death process contributes to podocyte detachment, glomerular damage, and tubulointerstitial fibrosis. By converging immune activation, mitochondrial dysfunction, and structural cell loss, PANoptosis may serve as a central mechanism linking upstream stressors to progressive renal injury in DN.

Regarding the three genes composing the gene signature, YWHAH is a member of the 14-3-3 protein family and is involved in various processes including signal transduction, cell cycle control, and apoptosis. Studies have shown that the expression of YWHAH is elevated in the transcriptome of DN patients [20]. Furthermore, YWHAH interacts with GREM1 to contribute to the pathogenesis of DN [30, 31]. PRKX is a c-AMP-dependent serine/threonine protein kinase-associated with normal kidney development and autosomal dominant polycystic kidney disease [32]. PDK4 is identified as a key upregulated gene in DN, playing a crucial role in glomerular basement membrane and kidney development [33]. Thus, the identification of these genes can potentially lead to more targeted approaches for the treatment of DN. Given the extremely low transcript levels of PRKX and PDK4 in plasma below the detection limit of our qPCR assay, we prioritized YWHAH for experimental validation. YWHAH showed consistent upregulation across all datasets, had detectable expression in plasma, and has been mechanistically linked to DN through its interaction with GREM1 [30, 31], a known regulator of podocyte injury and renal fibrosis [34]. In contrast, PRKX and PDK4 are intracellular kinases with low extracellular expression [35], making plasma-based detection technically unsuitable. Future studies should validate these targets in renal tissue or disease-relevant cellular models to elucidate their functional roles in DN.

In this study, we found distinct immune cell profiles in DN patients compared to controls. DN patients exhibited increased levels of resting DCs, M1, and M2 macrophages, while activated mast cells and neutrophils were reduced. These immune differences were associated with risk stratification, as high-risk DN patients showed elevated M2 macrophages and reduced neutrophils compared to low-risk patients. DCs are antigen-presenting cells that link the innate and adaptive immune responses by inducing T-cell responses. High glucose leads to DC activation and maturation, and the accumulation of these cells has been observed in animal models of DN and human kidney biopsies [36]. Macrophages play a pivotal role in inflammation, with M1 macrophages driving the initial proinflammatory response, while M2 macrophages help resolve inflammation and promote tissue repair. In simple acute inflammation, macrophages transition from M1 to M2 states, but in chronic conditions like DN, both M1 and M2 macrophages can coexist, contributing to persistent inflammation and fibrosis [37].

On the other hand, in DN, mast cells may have a dual role, contributing to both pathological processes and potential kidney homeostasis. Their precise impact likely depends on the specific context within the diabetic kidney microenvironment, including interactions with other cells and inflammatory factors [38]. Neutrophils play a critical role in the pathogenesis of DN. They migrate to the kidney in response to injury or inflammation, and the secreting enzymes and oxidation products can damage the local microenvironment, leading to tissue injury [39]. Recent studies have shown that neutrophils from type I diabetes patients produce neutrophil extracellular traps, which may contribute to DN [40, 41]. Neutrophil count and the neutrophil-to-lymphocyte ratio may be important factors in evaluating diabetic patients with a higher degree of albuminuria [42]. Notably, our immune infiltration analysis revealed decreased neutrophil abundance in DN patients, especially in the high-risk group, despite prior studies implicating neutrophil activity and NET formation in DN pathogenesis. This discrepancy can be explained by the clinical composition of the GSE96804 dataset, which includes both early- and late-stage DN patients, with the late-stage group (estimated glomerular filtration rate 15–60 mL/min/1.73 m², glomerular classes III–IV) reflecting a chronic disease state. In this context, reduced neutrophils likely reflect a shift from acute immune activation to chronic inflammatory remodeling dominated by M2 macrophages and fibrotic processes. This stage-specific immunological transition may account for the diminished neutrophil signal despite their known role in DN onset, underscoring the dynamic nature of immune responses throughout disease progression.

Our data from the DGIdb database suggest that PRKX and PDK4 are potential therapeutic targets for DN, with various compounds showing connections to PRKX and PDK4. Most of these compounds are kinase inhibitors that have demonstrated efficacy in kidney disease, as well as in inflammatory and fibrotic disorders, characteristics often associated with DN. For example, PF-562271, a dual inhibitor of Pyk2 and FAK kinases, can reduce NLRP3 inflammasome activation and apoptosis-associated speck-like protein containing CARD oligomerization, potentially offering therapeutic benefits in fibrotic disorders and inflammatory conditions [43, 44]. Similarly, dovitinib, a small-molecule kinase inhibitor, has shown antifibrotic and anti-inflammatory efficacy in certain diseases [45]. In DN, Y-27632 inhibits Rho-associated kinase 1, leading to improved fatty acid metabolism in the glomeruli. This effect is achieved by optimizing fatty acid utilization and redox balance in mesangial cells through AMPK phosphorylation and PGC-1α induction [46]. Additionally, CDK4/6 inhibitor palbociclib shows potential in treating kidney diseases [47]. Notably, these compounds have shown promise in cancer treatment and are often associated with significant toxicity [48‒51]. Therefore, their safety and efficacy profiles need to be carefully evaluated for potential repurposing in DN treatment.

This study has some limitations. First, the findings are primarily based on bioinformatics analysis of public datasets, which may not fully capture the complexity of DN in diverse patient populations. Experimental validation in clinical patients is needed to confirm the clinical utility of the identified genes. Additionally, the study focuses on the association of these genes with DN but does not investigate the underlying molecular mechanisms or causality. Furthermore, while the study explores potential therapeutic targets based on model genes, the effectiveness and safety of the suggested compounds in DN treatment require clinical evaluation. Moreover, the absence of treatment data in the original dataset limits the ability to assess the effect of different therapeutic interventions on the gene expression data, which could affect the interpretation and validity of the results presented in this study. Finally, the retrospective nature of the study limits its ability to provide prospective insights into disease progression or response to treatments. Future research should address these limitations for a more comprehensive understanding of DN and its management.

Our study identified a promising PANoptosis-related gene signature for DN, with PDK4, YWHAH, and PRKX serving as key biomarkers. This model demonstrated high accuracy in distinguishing between DN and non-DN conditions, offering the potential for further research into DN management. qPCR validation confirmed the upregulation of YWHAH in plasma samples from DKD patients, supporting its relevance in DN. Moreover, although PRKX and PDK4 were identified as potential therapeutic targets for DN based on available drug candidates, it is important to recognize that changes in gene expression can be both a cause and a consequence of the disease. Therefore, further investigation is required to validate these genes as viable therapeutic targets for DN.

Ethical approval and consent were not required as this study was based on publicly available data.

The authors have no conflicts of interest to declare.

This study was funded by the Jilin Provincial Science and Technology Development Plan Project (Grant No. 20240304042SF).

L.G. and Y.C. carried out the studies, participated in collecting data, and drafted the manuscript. Y.L. and Y.C. performed the statistical analysis and participated in its design. L.G. and Y.S. participated in acquisition, analysis, or interpretation of data and drafted the manuscript. All authors read and approved the final manuscript.

All data generated or analyzed during this study are included in this article and its supplementary material files. Further inquiries can be directed to the corresponding author.