Clear cell renal cell carcinoma (ccRCC) is the most common subtype of kidney cancer, accounting for 75% of all RCC cases. Approximately one-third of patients are already at an advanced stage at diagnosis, which is associated with high mortality and metastasis rates. ccRCC is notably resistant to chemotherapy and requires targeted or immunotherapy-based treatment approaches.
Cuproptosis is a recently discovered form of cell death in which copper ions bind to lipid-acylated components of the tricarboxylic acid cycle, inducing the aggregation of lipid-acylated proteins and downregulating iron-sulfur cluster proteins. This results in a form of proteotoxic stress that kills the cell and is distinct from classical death mechanisms such as apoptosis or ferroptosis.
Disulfidptosis is another newly identified cell death pathway triggered when abnormal expression of the transporter SLC7A11 under glucose-starved conditions depletes NADPH. Without sufficient NADPH to neutralize disulfide bonds, disulfide stress builds up, disrupts actin-cytoskeletal architecture, and causes cell death. Both cuproptosis and disulfidptosis are metabolically regulated mechanisms that may be exploitable in cancer therapy.
RNA sequencing data from 542 ccRCC tumor samples and 72 normal tissue samples were obtained from The Cancer Genome Atlas (TCGA-KIRC). The study identified 23 cuproptosis and disulfidptosis-related genes (CDRGs) from published literature, then used Pearson correlation analysis to identify 247 long non-coding RNAs (lncRNAs) significantly co-expressed with these genes.
LASSO machine learning regression was applied to the 108 lncRNAs that were significantly associated with overall survival in univariate Cox analysis. LASSO reduces overfitting by applying a regularization penalty that eliminates low-contribution variables, ultimately narrowing the candidate pool to four key cuproptosis and disulfidptosis-related lncRNAs (CDRLRs): ACVR2B-AS1, AC095055.1, AL161782.1, and MANEA-DT.
A risk score formula was constructed using the expression levels of these four lncRNAs weighted by their regression coefficients. Patients were divided into high- and low-risk groups based on the median risk score, and the model was validated in training, testing, and entire dataset splits. Calibration curves and ROC analysis confirmed model reliability with a C-index of 0.783.
The 1-year, 3-year, and 5-year AUC values for the risk score were 0.725, 0.718, and 0.762 respectively. When compared with other clinical variables, the risk score's AUC was second only to tumor stage, indicating its strong independent prognostic power relative to established clinical markers.
ACVR2B-AS1, AC095055.1, and AL161782.1 were significantly more expressed in the low-risk group and function as protective prognostic factors. ACVR2B-AS1 was independently confirmed as a protective factor with a hazard ratio of 0.48 (p less than 0.0001) in an external Kaplan-Meier Plotter database. Their upregulation is associated with better survival outcomes.
MANEA-DT was highly expressed in the high-risk group and acted as an adverse prognostic factor with a hazard ratio of 2.05 (p less than 0.0001). Multivariate Cox regression confirmed that the risk score, age, histological grade, and tumor stage were all independent prognostic factors in ccRCC patients, validating the biological and clinical relevance of the model.
RT-qPCR experiments in ccRCC cell lines (769-P and Caki-1) compared to normal renal epithelial cells (HK-2) confirmed the differential expression of these lncRNAs in cancer versus normal tissue, providing experimental validation of the computational findings and establishing that the biomarkers are detectable in vitro.
Tumor microenvironment (TME) analysis revealed that the high-risk group had significantly higher ESTIMATE and immune scores than the low-risk group. Using the CIBERSORT algorithm, significant differences in immune cell infiltration were identified, with the high-risk group showing enrichment of CD8+ T cells, follicular helper T cells, and regulatory T cells (Tregs).
The low-risk group showed upregulation of resting CD4+ memory T cells, M1 macrophages, M2 macrophages, and resting mast cells. These distinct immune compositions reflect fundamentally different tumor-immune interaction states and suggest that patients in different risk groups may respond differently to immunotherapy.
Analysis of five immune checkpoints including PD-1, PD-L1, CTLA-4, IL-6, and LAG3 showed that all were overexpressed in the high-risk group except PD-L1, which was highly expressed in the low-risk group. Using the Immunophenoscore algorithm, high-risk patients demonstrated heightened sensitivity to both single-agent and dual-agent PD-1 and CTLA-4 combination immunotherapy, pointing toward actionable treatment stratification.
Tumor mutational burden (TMB) analysis revealed that SETD2 and BAP1 hypermutations were more prevalent in the high-risk group. While TMB alone did not significantly stratify risk groups, combining TMB with the risk score produced a powerful prognostic combination: the High-TMB plus high-risk group had the lowest overall survival, while the Low-TMB plus low-risk group had the highest.
Drug sensitivity predictions using the oncoPredict algorithm indicated that low-risk patients were more sensitive to Alpelisib, Ipatasertib, Lapatinib, Selumetinib, and Pictilisib, while high-risk patients showed greater sensitivity to AZD4547, a FGFR inhibitor. These findings suggest that risk stratification could guide targeted drug selection beyond immunotherapy.
The biological differences between groups were further characterized through Gene Set Enrichment Analysis (GSEA), which found that high-risk patients had enriched complement and coagulation cascades, cytochrome P450 drug metabolism, and steroid hormone biosynthesis pathways, while low-risk patients showed enrichment in endocytosis, insulin signaling, and neurotrophin signaling pathways.
Cuproptosis and disulfidptosis represent a fundamentally new framework for understanding cancer cell metabolism and vulnerability. By identifying lncRNAs that regulate these pathways in ccRCC, this study demonstrates that metabolic cell death mechanisms are not just basic science curiosities but have measurable consequences for patient prognosis and drug response.
The four-lncRNA model is independent of established clinical parameters and provides additive prognostic value. Its ability to stratify patients across subgroups defined by age, gender, grade, and stage suggests it captures a biologically fundamental axis of tumor behavior not fully captured by anatomical or morphological assessments alone.
Future studies should explore whether these lncRNAs can be detected in blood or urine as liquid biopsy markers, which would enable non-invasive monitoring. Additionally, functional studies targeting MANEA-DT or the cuproptosis pathway directly may identify new therapeutic vulnerabilities in high-risk ccRCC patients who respond poorly to current standard-of-care regimens.