Identification of shared diagnostic biomarkers and molecular pathways between chronic kidney disease and renal cell carcinoma using transcriptomics and machine learning

Introduction

Chronic kidney disease (CKD), affecting 11% to 13% of the global population, is a significant public health concern characterized by persistent kidney abnormalities lasting for over 3 months (1). CKD notably elevates the risk of kidney cancer, particularly renal cell carcinoma (RCC), and is linked to poorer cancer outcomes (2), including advanced tumor stages, larger tumor sizes, and reduced survival rates (3). This association is likely driven by chronic inflammation, carcinogen accumulation, oxidative stress, impaired DNA repair, and gut microbiota dysbiosis (4).

RCC is a biologically and clinically heterogeneous disease comprising distinct histological subtypes with divergent molecular characteristics and clinical behaviors (5). Among these, clear cell RCC (ccRCC) is the most prevalent subtype, accounting for approximately 70–80% of RCC cases, and represents the dominant histology encountered in routine urologic practice (6). Accordingly, studies focusing on ccRCC provide a clinically meaningful foundation for mechanistic investigation and translational exploration.

In 2020, over 430,000 new cases of kidney cancer were reported globally, with RCC accounting for the majority (7). RCC, the deadliest genitourinary tumor (8), often presents with subtle early symptoms with limited treatment options (9). Surgical interventions, such as nephrectomy, frequently lead to CKD (10), while therapies like immune checkpoint inhibitors can cause acute kidney injury, exacerbating CKD development (11). Treatment-induced CKD increases cardiovascular morbidity and mortality, negatively affecting long-term outcomes (12). Effective CKD management in RCC patients is essential for improving survival (13), yet the pathogenesis and biomarkers of RCC in CKD remain poorly understood (14).

CKD and RCC are interconnected conditions with shared risk factors, including hypertension, diabetes, obesity, and smoking, which influence each other’s progression and outcomes (15). CKD may promote RCC development via mechanisms like cystic disease and oxidative stress, while RCC can induce CKD through tumor burden, nephrectomy, and renal-impairing treatments (16). Despite studies highlighting RCC’s interplay with comorbidities such as obesity, CKD, and diabetes (17), the shared genetic pathways and therapeutic implications remain unclear, necessitating further research to clarify these interactions (17). CKD is a multifactorial process in which glomerular injury initiates and contributes to disease progression, while tubulointerstitial damage plays a pivotal role in accelerating renal functional decline. Given that most RCCs, particularly ccRCC, arise from renal tubular epithelial cells, we focused on tubulointerstitial CKD transcriptomic datasets to better align with the relevant cellular origin and to identify shared molecular alterations linking CKD progression with RCC development. Importantly, from a translational standpoint, it is also unknown whether molecular features shared by CKD and RCC could be leveraged to support clinically relevant scenarios in urologic practice, such as early risk stratification of CKD patients, perioperative decision-making, or post-diagnosis prognostic assessment.

Against this background, the present study was designed to systematically investigate the relationship between CKD and RCC across genetic and transcriptomic levels, with a primary focus on ccRCC, given its predominance and clinical relevance. We first evaluated genetic correlation (r_g) and causality using genome-wide association study (GWAS) data, and subsequently focused on transcriptomic alterations shared between CKD and ccRCC. By integrating differential expression analysis, network-based approaches, and machine learning (ML) algorithms, we aimed to identify key hub genes associated with immune infiltration and disease status. Rather than proposing immediate clinical implementation, we sought to establish a biologically informed framework in which shared transcriptomic biomarkers could, upon future validation, contribute to risk stratification, treatment decision support, and prognostic evaluation in ccRCC patients with CKD. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2490/rc).

Methods

Public data acquisition

The GWAS summary data for CKD and RCC were obtained from the CKDGen consortium (https://ckdgen.imbi.uni-freiburg.de/datasets/Wuttke_2019) (18) and the FinnGen consortium (https://www.finngen.fi/en/access_results) (19). Transcriptome data were sourced from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/), with GSE104954 (20) and GSE104948 (20) providing CKD gene expression profiles, and GSE53757 (21) and GSE53000 (22) providing RCC profiles. GSE15641 dataset was used for differential expression analysis of the three hub genes across different RCC subtypes. This dataset includes 23 normal kidney samples and samples from five RCC subtypes: 32 ccRCC, 11 papillary RCC (pRCC), 6 chromophobe RCC (chrRCC), 12 oncocytoma (OC), and 8 transitional cell carcinoma (TCC). RNA-sequencing (RNA-Seq) expression data for kidney renal clear cell carcinoma (KIRC) patients were downloaded from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/), yielding a gene expression matrix of 614 samples and 59,427 genes. The datasets used in this study are summarized in Table 1. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Mendelian randomization (MR) and linkage disequilibrium (LD) score regression analysis reveal causal effects of genetic variation

This study utilized MR analysis to explore the causal relationship between CKD and RCC, incorporating GWAS data and leveraging genetic variations as instrumental variables (23). LD was assessed using LD score regression (LDSC) to estimate heritability and genetic effects, while the rho-Heritability Estimation from Summary Statistics (ρ-HESS) approach evaluated regional genetic contributions (24,25). Genetic overlap analysis further identified shared genetic associations between CKD and RCC, with all single nucleotide variants (SNVs) adjusted for multiple testing to ensure robustness (26).

Identification of differentially expressed genes (DEGs) between CKD patients and controls

The limma package in R was used to identify DEGs in the CKD dataset (27), |log2 fold change| ≥1, and adjusted P<0.05. DEGs were visualized via volcano plots.

Weighted gene coexpression network analysis (WGCNA) and gene module selection on the RCC dataset

WGCNA identified functional modules by constructing scale-free gene co-expression networks (28), filtering low-variance genes, removing outliers, and clustering topological overlap matrices, with hub genes defined by module membership and clinical correlations assessed via Spearman analysis.

Enrichment analysis of genes related to CKD-RCC

Gene Ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation of the shared genes were conducted via the org.Hs.eg.db and clusterProfiler packages in R. The ggplot2 package was used to visualize GO pathways with P<0.05, and all identified KEGG metabolism and signaling pathways.

Identifying key hub genes for CKD-RCC via ML

Using least absolute shrinkage and selection operator (LASSO), support vector machine-recursive feature elimination (SVM-RFE), and random forest (RF) algorithms, three key hub genes were identified for CKD and RCC diagnosis through feature selection, cross-validation, and consensus analysis (29).

Prediction performance in the validation cohort

To further evaluate the diagnostic accuracy of the three identified genes, external validation was conducted using two independent datasets: GSE104948 (175 CKD and 21 control samples) and GSE53757 (72 RCC and 72 control samples). Gene expression patterns were visualized with boxplots, and diagnostic performance was assessed by calculating the area under the curve (AUC).

Immune infiltration and correlation analysis

The composition and abundance of 22 immune cell types were analyzed using CIBERSORTx (https://cibersortx.stanford.edu/) based on transcriptome data from CKD datasets GSE104948 and GSE53757. Correlation analysis revealed associations between the expression of key CKD-RCC hub genes and immune cell infiltration, providing insights into their potential immunological roles.

Implementation of gene set enrichment analysis (GSEA) for single genes

Single-gene GSEA in the TCGA-KIRC dataset identified key pathways [|normalized enrichment score (NES)| >1, P<0.05] by stratifying samples into high- and low-expression groups, with top pathways visualized using MSigDB and enrichplot.

Development and validation of predictive models

To evaluate the diagnostic utility of the three hub genes, multivariable predictive models were developed and externally validated using the expression levels of C3, CYP27B1, and nicotinamide N-methyltransferase (NNMT) as predefined predictors. For CKD, the binary outcome was defined as CKD [1] versus non-CKD [0], and for RCC as tumor [1] versus normal tissue [0].

Model development was conducted in the discovery cohorts GSE104954 (CKD) and GSE53000 (RCC) using three classification algorithms: logistic regression (LR), RF, and support vector machine (SVM). LR models used maximum likelihood estimation, RF models were constructed with 500 trees, and SVM models employed a radial basis function kernel.

External validation was performed in the independent cohorts GSE104948 (CKD) and GSE53757 (RCC), with predicted probabilities generated from parameters learned in the development datasets. Predicted probabilities in the validation datasets were generated by applying the fitted LR, RF, and SVM models—trained on the development cohorts—to the log2-transformed gene expression values without model retraining. Model discrimination was assessed in both development and validation cohorts using receiver operating characteristic (ROC) analysis.

Statistical analysis

All bioinformatics and statistical analyses were performed using R (version 4.4.0). Group differences were evaluated using the Wilcoxon signed-rank test, and correlations were assessed via Spearman analysis. Statistical significance was set at P<0.05, with significance levels indicated as *, P<0.05; **, P<0.01; ***, P<0.001; and ****, P<0.0001.

Results

End-to-end analytic workflow

Figure 1 illustrates the comprehensive analytic workflow. Genetic associations between CKD and RCC were analyzed, followed by the identification of shared genes through differential expression and coexpression network analysis. Three ML algorithms identified key hub genes (C3, CYP27B1, and NNMT), whose predictive performance was evaluated via ROC curve analysis. Immune cell infiltration and its correlation with key gene expression were assessed in validation datasets, alongside analyses of their associations with RCC clinical stages and single-gene GSEA.

Figure 1 The flow chart for the whole design. CKD, chronic kidney disease; DEGs, differentially expressed genes; KEGG, Kyoto Encyclopedia of Genes and Genomes; KIRC, kidney renal clear cell carcinoma; GO, Gene Ontology; GSEA, gene set enrichment analysis; GWAS, genome-wide association studies; LASSO, least absolute shrinkage and selection operator; LDSC, linkage disequilibrium score regression; ρ-HESS, rho-Heritability Estimation from Summary Statistics; RCC, renal cell carcinoma; ROC, receiver operating characteristic; SVM-RFE, support vector machine-recursive feature elimination; TCGA, The Cancer Genome Atlas; WGCNA, weighted gene coexpression network analysis.

Potential causal effects between CKD and RCC

With a highly significant threshold of P<5×10⁻⁸, we selected 22 SNVs associated with CKD as instrumental variables. Inverse variance weighting (IVW) analysis revealed no significant causal effect of CKD on RCC [odds ratio (OR) =0.14, P=0.99]. Similarly, reverse MR analysis revealed no significant causal effect of RCC on CKD (OR =0.02, P=0.57) (Table S1). LDSC indicated no significant r_g (r_g =−0.0234, P=0.95) (Table S2). Additionally, regional r_g analysis via heritability estimation from summary statistics (ρ-HESS) estimated the genome-wide genetic covariance as −1.18×10⁵±5.68×10⁻⁵, confirming minimal and nonsignificant genetic covariance between the two traits (Figure S1A). Genetic overlap analysis further revealed that no SNVs were significantly associated with either trait (Figure S1B). Overall, no significant causal relationships or genetic associations were observed between CKD and RCC. However, epidemiological studies have suggested an association between these phenotypes (30). To further explore their potential shared molecular basis, we performed a transcriptome correlation analysis at the gene expression level.

Identification of DEGs related to CKD-RCC

We identified 1,223 DEGs related to RCC in the GSE53000 dataset (6 normal vs. 56 RCC samples), including 547 upregulated and 676 downregulated genes, as illustrated in the volcano plot (Figure 2A). Since RCC originates primarily from renal tubular epithelial cells, we selected the renal tubular mesenchymal transcriptome dataset GSE104954 (21 normal samples vs. 174 CKD samples) as the primary dataset for CKD analysis. This dataset is morphologically most relevant to CKD and RCC pathogenesis. Additional CKD datasets were used for validation. The GEO datasets used in this study contain gene expression profiles and case-control status but do not include demographic or detailed clinical variables. All gene expression data and outcome labels were fully observed with no missing data. A total of 54 CKD-related DEGs were identified, with 29 upregulated and 25 downregulated genes in CKD patients (Figure 2B). The Venn diagram (Figure 2C) highlights 13 common DEGs shared between RCC and CKD.

Figure 2 Identification of DEGs related to CKD-RCC and GO and KEGG pathway enrichment analysis. (A) Differential gene expression in RCC (GSE53000). (B) Differential gene expression in CKD (GSE104954). (C) The 13 common genes between RCC- and CKD-related DEGs. (D) GO enrichment analysis: shared genes were significantly enriched in 23 BPs, CCs, and MFs. (E) KEGG pathway analysis. BP, biological process; CC, cellular component; CKD, chronic kidney disease; DEGs, differentially expressed genes; FC, fold change; FDR, false discovery rate; KEGG, Kyoto Encyclopedia of Genes and Genomes; GO, Gene Ontology; MF, molecular function; RCC, renal cell carcinoma.

GO and KEGG pathway enrichment analysis

GO and KEGG enrichment analyses of the 13 shared genes associated with CKD revealed 23 significantly enriched GO pathways and 36 KEGG signaling pathways. GO analysis highlighted involvement in biological processes such as “renal filtration cell differentiation”, “podocyte differentiation”, and “kidney development”. Cellular component enrichment included “endoplasmic reticulum lumen” and “vesicle lumen”, while molecular functions were linked to “endopeptidase inhibitor activity” and “enzyme inhibitor activity” (Figure 2D). KEGG pathway enrichment analysis highlighted several critical pathways, including “steroid biosynthesis”, “renin-angiotensin system”, and “nicotinate and nicotinamide metabolism”. “Pertussis”, “alcoholic liver disease”, and “tuberculosis”. These pathways are strongly linked to key metabolic processes (Figure 2E).

Identification of genes related to CKD-RCC

WGCNA was performed on 62 samples from the RCC dataset GSE53000, using a soft threshold of β =16. The topological map (Figure 3A) illustrates the gene connections within the module. Correlation analysis of module eigengenes with tumor and normal samples (Figure 3B) revealed that the blue module, which contains 737 genes, was strongly associated with RCC tumor phenotypes (correlation coefficient r=−0.92, P<0.01). Network topology analysis demonstrated a significant positive correlation (r=0.87, P<0.01) between intramodular connectivity metrics and phenotypic association indices within the co-expression framework, suggesting that genes associated with RCC play crucial roles in key modules (Figure 3C). Finally, 13 DEGs intersected with the 737 genes in the blue module, yielding 6 genes associated with both CKD and RCC, as shown in the Venn diagram (Figure 3D).

Figure 3 WGCNA analysis of RCC dataset and identification of CKD-RCC key hub genes. (A) Gene cluster dendrogram for RCC dataset (GSE53000) by dynamic tree cut algorithm. (B) Heatmap of the association between modules and clinical traits in RCC. (C) The correlation between MM and GS. (D) The Venn diagram demonstrates the intersection of common genes obtained by WGCNA and DEGs. (E-G) Three different algorithms (LASSO, SVM-RFE, and random forest) were applied to select key hub genes with distinctive characteristics. (H) The Venn diagram demonstrates the key hub genes obtained by three different algorithms. CKD, chronic kidney disease; DEGs, differentially expressed genes; GS, gene significance; LASSO, least absolute shrinkage and selection operator; MM, module membership; RCC, renal cell carcinoma; RMSE, root mean square error; SVM-RFE, support vector machine-recursive feature elimination; WGCNA, weighted gene coexpression network analysis.

Identification of CKD-RCC key hub genes

A comprehensive ML framework integrating LASSO regression, SVM-RFE, and RF algorithms identified key CKD-RCC hub genes with strong intergroup discrimination. Among six candidate genes, LASSO regression identified three, SVM-RFE identified four, and RF ranked six by importance. The top six genes were ranked by importance, and we selected five genes (importance >1) as the results. Combining results, NNMT, CYP27B1, and C3 were identified as the final CKD-RCC hub genes (Figure 3E-3H).

Diagnostic value and validation of key hub genes

We analyzed the expression levels of key hub genes associated with CKD-RCC in two discovery datasets: GSE104948 (21 normal samples vs. 175 CKD samples) and GSE53757 (72 normal samples vs. 72 RCC samples). As shown in Figure 4A, C3 and NNMT were significantly upregulated, whereas CYP27B1 was significantly downregulated in CKD patients (P<0.001). Similarly, Figure 4B shows that these three genes exhibited the same expression trends in RCC patients (P<0.001).

Figure 4 Diagnostic value and validation of key hub genes and immune infiltration. (A,B) Differential expression of C3, CYP27B1, and NNMT in the discovery datasets for CKD (GSE104948) and RCC (GSE53757). (C,D) ROC curve of C3, CYP27B1, and NNMT in the discovery datasets for CKD (GSE104948) and RCC (GSE53757). (E) The box plots indicated that the CKD and RCC exhibited significantly different types of immune cells. (F) Correlation between three key hub genes expression and immune cells in the CKD and RCC. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. AUC, area under the curve; CKD, chronic kidney disease; NK, natural killer; RCC, renal cell carcinoma; ROC, receiver operating characteristic; TME, tumor microenvironment.

To explore the expression patterns of the three target genes across different RCC histological subtypes, we further analyzed the GSE15641 dataset, comparing cRCC, pRCC, chrRCC, OC, and TCC with normal renal tissue. As shown in Figure S2, cRCC and pRCC exhibited consistent expression trends for C3, CYP27B1, and NNMT. In contrast, C3 expression showed no significant difference from normal tissue in chrRCC, OC, and TCC, while CYP27B1 and NNMT displayed opposite expression trends in OC and chrRCC compared with cRCC and pRCC.

Integrating ROC curve analysis quantified the diagnostic precision of key hub genes across clinical phenotypes. In the CKD dataset (GSE104948), CYP27B1 (AUC =0.96), NNMT (AUC =0.80), and C3 (AUC =0.82) showed strong performance. Similarly, in the RCC dataset (GSE53757), CYP27B1 (AUC =0.94), NNMT (AUC =0.94), and C3 (AUC =0.96) demonstrated sustained prognostic validity (Figure 4C,4D), confirming their roles as key hub genes for CKD and RCC.

Immune infiltration and correlation analysis of CKD-RCC key hub genes

Considering the strong immune response associated with CKD and RCC, CIBERSORTx analysis of GSE104948 and GSE53757 revealed decreased memory B cells and resting CD4⁺ T cells in both diseases. CKD samples showed increased resting natural killer (NK) cells and decreased activated NK cells, while RCC samples exhibited the opposite pattern, with decreased resting NK cells and increased activated NK cells (Figure 4E).

Spearman correlation analysis was performed to assess the relationships between key hub genes and immune cell abundance. The correlation heatmap (Figure 4F) illustrates the associations between key hub genes and the relative abundances of immune cell types. In CKD samples, memory B cells were positively correlated with NNMT and C3 (r=0.30, P=0.04; r=0.39, P<0.01), and C3 was positively correlated with resting CD4⁺ T cells (r=0.33, P=0.02). In contrast, in RCC samples, memory B cells were negatively correlated with C3 (r=−0.47, P=0.01), and C3 was negatively correlated with resting CD4⁺ T cells (r=−0.40, P=0.04). In the CKD samples, NNMT and C3 were significantly negatively correlated with monocytes (r=−0.47, P<0.001; r=−0.62, P<0.001). In addition, activated CD4⁺ T cells were significantly positively correlated with CYP27B1, NNMT, and C3 (r=0.60, P<0.001; r=0.52, P<0.01; r=0.39, P=0.050). These findings suggest that immune function plays a crucial role in the development of CKD and RCC.

Clinical value of key hub genes in RCC patients

The expression levels of CKD-RCC hub genes were analyzed in the TCGA-KIRC dataset. C3 and NNMT were significantly upregulated, while CYP27B1 was downregulated in RCC patients (P<0.0001, Figure 5A). Transcriptomic analysis revealed a reciprocal regulation pattern in progressive RCC, with C3 upregulated and CYP27B1 downregulated (Figure 5B). Survival analysis based on TCGA-KIRC data showed significantly reduced survival in progressive RCC patients compared to localized RCC (P<0.0001) (Figure 5C).

Figure 5 Clinical value of key hub genes in RCC patients and single-gene GSEA. (A) Differential expression of C3, CYP27B1, and NNMT in the datasets for TCGA-KIRC. (B) Correlation between three key hub genes expression and clinical stages. (C) Survival analysis of localized versus progressive RCC patients. (D,E) ROC analysis demonstrated robust diagnostic performance of the three-gene signature across all models. (F) Single-gene GSEA analysis for C3 and CYP27B1 in TCGA-KIRC. ****, P<0.0001. AUC, area under the curve; CKD, chronic kidney disease; KIRC, kidney renal clear cell carcinoma; GSEA, gene set enrichment analysis; RCC, renal cell carcinoma; ROC, receiver operating characteristic; TCGA, The Cancer Genome Atlas.

Comparison of predictive model performance

The diagnostic performance of the three-gene signature was evaluated across all predictive models. Predictions were obtained by inputting log2-transformed expression levels of C3, CYP27B1, and NNMT into the trained LR, RF, or SVM models. In the CKD external validation cohort (GSE104948), the three models showed strong discrimination, with LR achieving the highest AUC (0.9689), followed by RF (0.9557) and SVM (0.9202) (Figure 5D). In the RCC validation cohort (GSE53757), RF demonstrated the best performance (AUC =0.9568), exceeding LR (0.9337) and SVM (0.8042) (Figure 5E).

Internal validation using 5-fold cross-validation in the training datasets showed stable model performance, with high AUC, accuracy, sensitivity, and specificity across both CKD and RCC models (Tables S3,S4).

Overall, LR was selected as the preferred diagnostic model for CKD due to its high discrimination and simplicity, whereas RF was identified as the optimal model for RCC based on its superior AUC and calibration characteristics.

Single-gene GSEA of key hub genes

According to GeneCards (https://www.genecards.org), both C3 and CYP27B1 are highly expressed in the kidneys (Figure S3). Single-gene GSEA in the TCGA-KIRC dataset revealed that C3 is enriched in immune-related pathways, including vesicle transport, adaptive immunity regulation, and immune response activation (Figure 5F). CYP27B1 is associated with cellular transport, metabolic regulation, extracellular structure, vesicle systems, and lymphocyte immunity enhancement.

Discussion

Key findings

Gene expression analysis revealed key mechanisms underlying CKD comorbidity with RCC, identifying 13 DEGs, including three key hub genes (C3, CYP27B1, and NNMT) validated for significant expression. Enrichment analysis linked these DEGs to critical pathways such as renal epithelial cell differentiation, steroid biosynthesis, the renin-angiotensin-aldosterone system (RAAS), and nicotinate/nicotinamide metabolism, highlighting immune-metabolic dysregulation as a shared biological feature between CKD and RCC. Dedifferentiation or abnormal proliferation of renal tubular epithelial cells contributes to fibrotic progression in CKD (31) and is also a key event in RCC tumorigenesis (32), providing a biological basis for shared transcriptomic alterations. RAAS activation exacerbates CKD through sodium retention and inflammation (33) and may also promote RCC progression through angiogenic factors such as vascular endothelial growth factor (VEGF) (34). In parallel, dysregulation of nicotinic acid/nicotinamide metabolism may impair NAD⁺ homeostasis, leading to oxidative stress and epigenetic remodeling that jointly drive renal injury and tumor-promoting microenvironment formation (35).

Beyond mechanistic insights, these findings have potential implications for urologic clinical practice, particularly in the context of CKD-associated RCC risk. The pathogenesis of both CKD and RCC is strongly associated with chronic inflammation (36). In severe CKD, immune activation leads to tissue damage and progressive loss of function (37), further promoting cancer progression. Although the present study is exploratory and not intended to support immediate clinical implementation, the identified hub genes may serve as biologically informed candidates for future translational applications. First, in risk stratification and early screening, transcriptomic signatures involving C3, CYP27B1, and NNMT may help identify CKD patients with heightened susceptibility to RCC beyond conventional demographic or clinical risk factors. Second, in perioperative decision-making, particularly for CKD patients being considered for nephrectomy, nephron-sparing surgery, or ablative therapies, immune-metabolic dysregulation reflected by these genes may provide complementary information regarding tumor aggressiveness and renal functional vulnerability. Third, in the post-diagnosis setting, the observed associations between hub gene expression, immune cell infiltration, and RCC clinical stage suggest potential utility for prognostic stratification, which may inform postoperative surveillance intensity and adjunctive treatment considerations.

Our study further demonstrated that the three hub genes (C3, CYP27B1, and NNMT) possess strong diagnostic utility for both CKD and RCC. The predictive models developed from these genes showed high discrimination in external validation cohorts. LR achieved the best performance for CKD classification (AUC =0.9689), while RF was superior for RCC (AUC =0.9568), and both models exhibited favorable calibration. These results support the robustness and generalizability of the three-gene signature and highlight its potential clinical value as a shared diagnostic indicator for CKD-RCC comorbidity. Their functions were closely related to the immune-metabolic network. Complement C3 plays a crucial role in the progression of interstitial fibrosis and renal injury in human hypertensive nephropathy (38). Single-gene GSEA revealed that C3 was associated with immune upregulation, possibly recruiting monocytes and T cells through complement activation (e.g., C3a and C5a), thereby forming a chronic inflammatory microenvironment that promotes CKD fibrosis and RCC immune escape (39). C3 was positively correlated with RCC stage, suggesting that C3 may serve as an inflammatory biomarker for tumor progression. Notably, CKD-associated immune dysregulation may also influence responsiveness to immunotherapy in RCC patients (40). Chronic inflammation, complement activation, and altered T-cell states observed in CKD could potentially modulate the tumor immune microenvironment and affect immune checkpoint inhibitor efficacy, although this hypothesis requires dedicated clinical validation. CYP27B1, regulated by parathyroid hormone (PTH), fibroblast growth factor 23 (FGF23), and 1,25(OH)₂D₃, catalyzes the conversion of 25-hydroxyvitamin D to the active form, 1,25-(OH)₂D₃ (41). Decreased expression of CYP27B1 may lead to vitamin D deficiency, aggravating CKD mineral metabolism disorders and promoting RCC by inhibiting epithelial differentiation (42). GSEA suggested that vitamin D participates in cellular secretion, potentially related to RAAS hormones (e.g., angiotensin II) or extracellular vesicle signal transmission. One mechanism of vitamin D action is its role as a negative regulator of the RAAS (43). The vitamin D-PTH axis is crucial in CKD patients, as kidney 1,25-(OH)₂D synthesis decreases with worsening kidney function (44). NNMT is a phase II metabolic enzyme that is expressed primarily in the liver but also in the kidney (45). Energy metabolism, influenced by depletion of NAD⁺ precursors and epigenetic modifications (e.g., histone methylation), may simultaneously drive metabolic stress in CKD and the Warburg effect in RCC (46).

The immune-metabolic axis is the core network connecting CKD and RCC, and it has a double-edged sword effect in both diseases. In CKD, C3 promotes monocyte/macrophage infiltration through the complement cascade, exacerbating renal interstitial inflammation (47). In RCC, however, C3 may induce an immunosuppressive tumor microenvironment (TME) via tumor-associated macrophages (TAMs) (48), demonstrating its dual role in chronic inflammation and tumor immunosuppression. Downregulation of CYP27B1 in CKD results in active vitamin D deficiency, impairing its anti-inflammatory and antifibrotic effects (49,50). In RCC, tumor cells may inhibit CYP27B1 expression (51), signaling through proangiogenic factors such as VEGF or exosomes that promote abnormal activation and functional depletion of T cells (52). In CKD, NNMT upregulation may induce oxidative stress and mitochondrial dysfunction through NAD⁺ depletion, forcing T cells into a resting state (53). Moreover, in RCC, NNMT-mediated reprogramming of nicotinamide metabolism may provide metabolic advantages to rapidly proliferating tumor cells and immunosuppressive cells, thereby creating an immune escape microenvironment (54).

The high AUC values of C3, CYP27B1, and NNMT highlight their potential as diagnostic biomarkers for RCC and CKD, particularly in early tumor screening for CKD patients. The correlation between C3 and CYP27B1 in RCC staging suggests C3 as a potential marker for monitoring tumor progression, while elevated NNMT expression in CKD patients may indicate a risk of metabolic complications (55). These findings provide insights into RCC pathogenesis and establish a basis for future research on biomarkers and therapeutic targets to mitigate RCC risk in CKD patients.

Nevertheless, several of the identified hub genes encode proteins with extracellular or systemic relevance, supporting the biological plausibility of future non-invasive assay development. Notably, C3 is a central component of the complement cascade and is readily detectable in both blood and urine, making it a particularly attractive candidate for circulating or urinary biomarker exploration (56). Similarly, CYP27B1 and NNMT are involved in metabolic pathways with potential systemic readouts, suggesting that transcriptomic alterations may be reflected, at least in part, at the protein or metabolite level. These characteristics provide a rationale for future studies aimed at evaluating blood- or urine-based surrogates as minimally invasive alternatives to tissue-based assays.

From a translational perspective, the significance of these findings lies in their capacity to support risk-adapted clinical decision-making rather than immediate mechanistic conclusions. Following prospective validation, transcriptomic risk signatures associated with chronic inflammation, immune remodeling, or metabolic stress may complement conventional clinical parameters in specific urologic contexts. One such context is the management of small renal masses in patients with pre-existing CKD, where optimizing oncologic control while preserving renal function is critical. Current evidence supports nephron-sparing and minimally invasive strategies, including percutaneous cryoablation and other thermal ablation techniques, as effective alternatives to radical nephrectomy in carefully selected high-risk individuals, particularly those with limited renal reserve or significant comorbidities (57,58).

Limitations

There are several limitations in this study. First, the analyses were based on publicly available transcriptomic datasets, which are subject to batch effects and population heterogeneity and may limit result robustness. In addition, direct clinical validation was not performed, and the molecular mechanisms identified through integrative transcriptomic and ML analyses require further functional and longitudinal investigation to establish causality in CKD-RCC comorbidity.

Second, although the lack of adequately powered non-ccRCC datasets limited comprehensive histology-specific analyses, we conducted an exploratory assessment using the GSE15641 dataset. The expression patterns of C3, CYP27B1, and NNMT were consistent between ccRCC and pRCC, suggesting potential applicability to pRCC. In contrast, chrRCC, OC, and TCC showed distinct expression profiles, particularly for CYP27B1 and NNMT, reflecting marked histological heterogeneity. These preliminary findings should be interpreted cautiously and require validation in larger, well-annotated cohorts.

Third, publicly available datasets lack granular urologic and nephrology-related clinical variables essential for clinical decision-making, including tumor size, nephrometry score, surgical approach, baseline estimated glomerular filtration rate (eGFR), and CKD stage. As a result, this study could not evaluate the incremental predictive value of C3, CYP27B1, and NNMT beyond established clinical predictors. These genes should therefore be interpreted as biologically informative coexistence biomarkers reflecting shared immune-metabolic dysregulation, rather than as independent clinical decision tools.

The three target genes, C3, NNMT, and CYP27B1, demonstrate potential for clinical translation. C3 may serve as a circulating or urinary biomarker, NNMT may be utilized through metabolite- or enzyme activity-based surrogate markers, and CYP27B1 may be integrated with established vitamin D-related biochemical indicators in CKD management. Nevertheless, future prospective cohorts or well-annotated institutional datasets incorporating comprehensive clinical parameters will be required to validate these biomarkers and to determine their utility within integrated clinical-molecular risk stratification frameworks for urologic practice.

Conclusions

In summary, this study provides a systematic transcriptomic investigation into the shared molecular landscape linking CKD and ccRCC. By integrating r_g analysis, differential expression profiling, network-based approaches, and ML algorithms, we identified C3, CYP27B1, and NNMT as key genes associated with immune-metabolic dysregulation in the context of CKD-ccRCC coexistence. These findings contribute to a deeper understanding of the biological mechanisms underlying this clinically important comorbidity.

Importantly, the identified genes should not be interpreted as immediately deployable clinical biomarkers. Instead, they represent complementary molecular candidates that may inform future integrative clinical molecular models when combined with established urologic and nephrology-related parameters. Further validation in prospective cohorts with comprehensive clinical annotation will be essential to determine their incremental value beyond known clinical predictors and to define their potential role in precision risk stratification and management of ccRCC patients with underlying CKD.

Acknowledgments

We sincerely thank the CKDGen and FinnGen consortia for providing GWAS summary data, as well as the GEO and TCGA databases for making transcriptomic data publicly available. We also appreciate the valuable feedback from reviewers and editors, which helped improve this manuscript.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2490/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2490/prf

Funding: This study was supported by the Jiangsu Provincial Health Commission Key Research Project (No. K2023001) and the General Program of the Natural Science Foundation of Jiangsu Province (grant No. BK20251959).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-aw-2490/coif). The authors have no conflicts of interest to declare.

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

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Lash LH. Cellular and Functional Biomarkers of Renal Injury and Disease. Curr Opin Toxicol 2022;31:100348. [Crossref] [PubMed]
2. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
3. Dey S, Hamilton Z, Noyes SL, et al. Chronic Kidney Disease Is More Common in Locally Advanced Renal Cell Carcinoma. Urology 2017;105:101-7. [Crossref] [PubMed]
4. Kitchlu A, Reid J, Jeyakumar N, et al. Cancer Risk and Mortality in Patients With Kidney Disease: A Population-Based Cohort Study. Am J Kidney Dis 2022;80:436-448.e1. [Crossref] [PubMed]
5. Kim CG, Koo H, Kwon WS, et al. Use of integrated multi-omics profiling in renal cell carcinoma to identify molecular subtypes and associated clinical outcomes and therapeutic vulnerabilities. J Clin Oncol 2025;43:588.
6. Jain M, Mann H, Banu T, et al. 178 Institutional Review of Renal Cell Carcinoma: Histopathological Subtypes and Immunohistochemical Profiling. American Journal of Clinical Pathology 2025;164:aqaf121.161.
7. Stevens PE, Levin A. Kidney Disease: Improving Global Outcomes Chronic Kidney Disease Guideline Development Work Group Members. Evaluation and management of chronic kidney disease: synopsis of the kidney disease: improving global outcomes 2012 clinical practice guideline. Ann Intern Med 2013;158:825-30. [Crossref] [PubMed]
8. Janebdar H, Fernandes M, Manson-Bahr D. Renal cancer. Surgery 2022;40:653-9.
9. Moch H, Cubilla AL, Humphrey PA, et al. The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs-Part A: Renal, Penile, and Testicular Tumours. Eur Urol 2016;70:93-105. [Crossref] [PubMed]
10. Schleef M, Roy P, Lemoine S, et al. Renal and major clinical outcomes and their determinants after nephrectomy in patients with pre-existing chronic kidney disease: A retrospective cohort study. PLoS One 2024;19:e0300367. [Crossref] [PubMed]
11. Nguyen M, Walia A, Narasimhan R, et al. MP47-16 decline to “severe” CKD-S is a predictive marker for all-cause and non-cancer mortality in renal cell carcinoma: analysis of the International Marker Consortium for Renal Cancer (INMARC). Journal of Urology 2022;207:e816.
12. Warrens H, Banerjee D, Herzog CA. Cardiovascular Complications of Chronic Kidney Disease: An Introduction. Eur Cardiol 2022;17:e13. [Crossref] [PubMed]
13. Pirovano M, Cosmai L. Nephrological management of renal cell carcinoma (RCC) patients: A narrative review. Journal of Onco-Nephrology 2022;6:97-106.
14. Brooks ER, Siriruchatanon M, Prabhu V, et al. Chronic kidney disease and risk of kidney or urothelial malignancy: systematic review and meta-analysis. Nephrol Dial Transplant 2024;39:1023-33. [Crossref] [PubMed]
15. Bekku K, Nagasaki N, Tsuboi I, et al. Tolerability of immune checkpoint inhibitor doublet for advanced renal cell carcinoma patients with pre-dialysis chronic kidney disease or end-stage renal disease. Int J Urol 2023;30:931-3. [Crossref] [PubMed]
16. Saly DL, Eswarappa MS, Street SE, et al. Renal Cell Cancer and Chronic Kidney Disease. Adv Chronic Kidney Dis 2021;28:460-468.e1. [Crossref] [PubMed]
17. Hossen MS, Samad A, Ahammad F, et al. System biology approaches identified novel biomarkers and their signaling pathways involved in renal cell carcinoma with different human diseases. Biosci Rep 2022;42:BSR20221108. [Crossref] [PubMed]
18. Wuttke M, Li Y, Li M, et al. A catalog of genetic loci associated with kidney function from analyses of a million individuals. Nat Genet 2019;51:957-72. [Crossref] [PubMed]
19. Kurki MI, Karjalainen J, Palta P, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 2023;613:508-18. [Crossref] [PubMed]
20. Grayson PC, Eddy S, Taroni JN, et al. Metabolic pathways and immunometabolism in rare kidney diseases. Ann Rheum Dis 2018;77:1226-33. [Crossref] [PubMed]
21. von Roemeling CA, Radisky DC, Marlow LA, et al. Neuronal pentraxin 2 supports clear cell renal cell carcinoma by activating the AMPA-selective glutamate receptor-4. Cancer Res 2014;74:4796-810. [Crossref] [PubMed]
22. Gerlinger M, Horswell S, Larkin J, et al. Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing. Nat Genet 2014;46:225-33. [Crossref] [PubMed]
23. Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet 2017;13:e1007081. [Crossref] [PubMed]
24. Bulik-Sullivan B, Finucane HK, Anttila V, et al. An atlas of genetic correlations across human diseases and traits. Nat Genet 2015;47:1236-41. [Crossref] [PubMed]
25. Shi H, Kichaev G, Pasaniuc B. Contrasting the Genetic Architecture of 30 Complex Traits from Summary Association Data. Am J Hum Genet 2016;99:139-53. [Crossref] [PubMed]
26. Andreassen OA, Djurovic S, Thompson WK, et al. Improved detection of common variants associated with schizophrenia by leveraging pleiotropy with cardiovascular-disease risk factors. Am J Hum Genet 2013;92:197-209. [Crossref] [PubMed]
27. Ritchie ME, Phipson B, Wu D, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015;43:e47. [Crossref] [PubMed]
28. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 2008;9:559. [Crossref] [PubMed]
29. Cao Y, Du Y, Jia W, et al. Identification of biomarkers for the diagnosis of chronic kidney disease (CKD) with non-alcoholic fatty liver disease (NAFLD) by bioinformatics analysis and machine learning. Front Endocrinol (Lausanne) 2023;14:1125829. [Crossref] [PubMed]
30. Dey S, Hamilton Z, Noyes SL, et al. Association of chronic kidney disease and locally-advanced RCC. J Clin Oncol 2016;34:e16078.
31. Taguchi K, Elias BC, Sugahara S, et al. Cyclin G1 induces maladaptive proximal tubule cell dedifferentiation and renal fibrosis through CDK5 activation. J Clin Invest 2022;132:e158096. [Crossref] [PubMed]
32. Landolt L, Eikrem Ø, Strauss P, et al. Clear Cell Renal Cell Carcinoma is linked to Epithelial-to-Mesenchymal Transition and to Fibrosis. Physiol Rep 2017;5:e13305. [Crossref] [PubMed]
33. Nagasawa Y. Positive and Negative Aspects of Sodium Intake in Dialysis and Non-Dialysis CKD Patients. Nutrients 2021;13:951. [Crossref] [PubMed]
34. Guillaume Z, Auvray M, Vano Y, et al. Renal Carcinoma and Angiogenesis: Therapeutic Target and Biomarkers of Response in Current Therapies. Cancers (Basel) 2022;14:6167. [Crossref] [PubMed]
35. Fontecha-Barriuso M, Lopez-Diaz AM, Carriazo S, et al. Nicotinamide and acute kidney injury. Clin Kidney J 2021;14:2453-62. [Crossref] [PubMed]
36. Feng Q, Yang Y, Ren K, et al. Broadening horizons: the multifaceted functions of ferroptosis in kidney diseases. Int J Biol Sci 2023;19:3726-43. [Crossref] [PubMed]
37. Plonsky-Toder M, Magen D, Pollack S. Innate Immunity and CKD: Is There a Significant Association? Cells 2023;12:2714. [Crossref] [PubMed]
38. Wu X, Lin L, Cui J, et al. Complement C3 deficiency ameliorates aging related changes in the kidney. Life Sci 2020;260:118370. [Crossref] [PubMed]
39. Meuleman MS, Petitprez F, Pickering MC, et al. Complement Terminal Pathway Activation and Intrarenal Immune Response in C3 Glomerulopathy. J Am Soc Nephrol 2024;35:1034-44. [Crossref] [PubMed]
40. Trevisani F, Angioi A, Ghidini M, et al. Immunotherapy-Associated Renal Dysfunction in Metastatic Cancer: An Emerging Challenge in Onco-Nephrology. Cancers (Basel) 2025;17:2090. [Crossref] [PubMed]
41. Joergensen H, Peersman N, Billen J, et al. #3325 Extranenal CYP27B1 activity contributes to the circulating pool of 1,25(OH)2D. Nephrology Dialysis Transplantation 2023;38:gfad063b_3325.
42. Meyer MB, Pike JW. Mechanistic homeostasis of vitamin D metabolism in the kidney through reciprocal modulation of Cyp27b1 and Cyp24a1 expression. J Steroid Biochem Mol Biol 2020;196:105500. [Crossref] [PubMed]
43. Reina-Couto M, Afonso J, Carvalho J, et al. Interrelationship between renin-angiotensin-aldosterone system and oxidative stress in chronic heart failure patients with or without renal impairment. Biomed Pharmacother 2021;133:110938. [Crossref] [PubMed]
44. Latic N, Erben RG. Interaction of Vitamin D with Peptide Hormones with Emphasis on Parathyroid Hormone, FGF23, and the Renin-Angiotensin-Aldosterone System. Nutrients 2022;14:5186. [Crossref] [PubMed]
45. Takahashi R, Kanda T, Komatsu M, et al. The significance of NAD + metabolites and nicotinamide N-methyltransferase in chronic kidney disease. Sci Rep 2022;12:6398. [Crossref] [PubMed]
46. Reustle A, Menig LS, Leuthold P, et al. Nicotinamide-N-methyltransferase is a promising metabolic drug target for primary and metastatic clear cell renal cell carcinoma. Clin Transl Med 2022;12:e883. [Crossref] [PubMed]
47. Glezerman I, Gutgarts V, Salvatore S. #1150 Neutrophil rich infiltrate in immunotherapy induced acute interstitial nephritis. Nephrology Dialysis Transplantation 2024;39:gfae069-0439-1150.
48. Zhang Y, Wang X, Gu Y, et al. Complement C3 of tumor-derived extracellular vesicles promotes metastasis of RCC via recruitment of immunosuppressive myeloid cells. Proc Natl Acad Sci U S A 2025;122:e2420005122. [Crossref] [PubMed]
49. Liu W, Zhang L, Xu HJ, et al. The Anti-Inflammatory Effects of Vitamin D in Tumorigenesis. Int J Mol Sci 2018;19:2736. [Crossref] [PubMed]
50. Fenercioglu AK. The Anti-Inflammatory Roles of Vitamin D for Improving Human Health. Curr Issues Mol Biol 2024;46:13514-25. [Crossref] [PubMed]
51. Campolina-Silva GH, Barata MC, Werneck-Gomes H, et al. Altered expression of the vitamin D metabolizing enzymes CYP27B1 and CYP24A1 under the context of prostate aging and pathologies. J Steroid Biochem Mol Biol 2021;209:105832. [Crossref] [PubMed]
52. Yang J, Yan J, Liu B. Targeting VEGF/VEGFR to Modulate Antitumor Immunity. Front Immunol 2018;9:978. [Crossref] [PubMed]
53. Zhang ZY, Wang F, Zhou G. Decrease of NAD+ Inhibits the Apoptosis of OLP T Cells via Inducing Mitochondrial Fission. J Inflamm Res 2025;18:1091-106. [Crossref] [PubMed]
54. Novak Kujundžić R, Prpić M, Đaković N, et al. Nicotinamide N-Methyltransferase in Acquisition of Stem Cell Properties and Therapy Resistance in Cancer. Int J Mol Sci 2021;22:5681. [Crossref] [PubMed]
55. Ye Q, Xu G, Huang H, et al. Nicotinamide N-Methyl Transferase as a Predictive Marker of Tubular Fibrosis in CKD. Int J Gen Med 2023;16:3331-44. [Crossref] [PubMed]
56. Shu X, Gao X, Dai Y, et al. C3 as a predictive and prognostic biomarker in adult hemophagocytic lymphohistiocytosis: a large cohort study in China. Blood Adv 2025;9:1836-46. [Crossref] [PubMed]
57. Piasentin A, Claps F, Silvestri T, et al. Assessing Trifecta Achievement after Percutaneous Cryoablation of Small Renal Masses: Results from a Multi-Institutional Collaboration. Medicina (Kaunas) 2022;58:1041. [Crossref] [PubMed]
58. Marinelli L, Mercogliano S, Selvaggio O, et al. Percutaneous Computed Tomography-Guided Cryoablation as a Treatment Option in Patients with Small Renal Masses: A 10 Year Experience in a Single Center. Radiation 2024;4:346-54.