Immune Cell Genes Track Ferroptosis Signals in Kidney Cancer

Clear cell renal cell carcinoma, the most common form of kidney cancer, carries a distinctive metabolic and immune landscape, and a new multi-dataset study reports that five genes long read as bulk-tissue signals in these tumors are best understood as markers of the immune microenvironment rather than as ferroptosis machinery. The genes, CCR5, CSF1R, CD86, CD8A, and LCP2, were consistently upregulated in tumor tissue across three public cohorts, showed strong capacity to distinguish tumor from normal kidney, and localized by single-cell analysis mainly to macrophages and T or natural killer cells. The work, published in Frontiers in Genetics, deliberately separates a ferroptosis-associated transcriptional state from direct regulation of the cell-death pathway itself.

Ferroptosis is an iron-dependent form of regulated cell death driven by phospholipid peroxidation when antioxidant defenses fail. Its canonical network spans iron handling, polyunsaturated fatty acid metabolism, glutathione synthesis, and the GPX4/SLC7A11 antioxidant axis. Clear cell renal cell carcinoma is a particularly relevant setting because its hallmark VHL-HIF metabolic phenotype reshapes lipid storage, redox balance, mitochondrial metabolism, and oxygen sensing, conditions under which lipid-peroxide homeostasis may matter both biologically and therapeutically. The study’s central question was whether immune genes that co-vary with ferroptosis-related transcriptional remodeling actually participate in that remodeling, or simply report the immune cells that accompany it.

The analysis drew on three public datasets: GSE53757, with 72 clear cell renal cell carcinoma tissues and 72 matched normal kidney samples; GSE159115 single-cell RNA-sequencing data from a male patient’s tumor; and TCGA-KIRC, comprising 542 primary tumors and 72 normal kidney tissues. Differential expression testing in GSE53757 identified 8,650 differentially expressed genes among 22,189 assessed, including 3,722 upregulated and 4,928 downregulated transcripts. Notably, of 35 FerrDb v2 ferroptosis genes, 34 were detected and 13 were significantly upregulated, among them SLC7A11, HMOX1, ALOX5, TP53, CDKN1A, MDM2, BAX, BCL2, PMAIP1, FANCD2, BRCA1, BRCA2, and EGFR. A hypergeometric test put this enrichment at p = 2.3 × 10−3, supporting genuine ferroptosis-related transcriptional remodeling in the tumors.

The five hub genes were themselves strongly upregulated in the discovery cohort: CCR5 with a log2 fold change of 2.54, LCP2 at 2.35, CD86 at 2.30, CD8A at 1.95, and CSF1R at 1.44, all with adjusted p values far below conventional thresholds. Hierarchical co-expression clustering of the 5,200 most variable genes produced 20 modules, and the M03 module showed a strong positive association with the cancer phenotype, with a Spearman correlation of 0.80. All five genes sat within M03, whose significant differentially expressed genes were enriched for immune processes including regulation of interleukin-1β production, defense response to symbiont, and T-cell activation, plus KEGG terms covering chemokine signaling, phagosome biology, and natural-killer-cell-mediated cytotoxicity. The dominant functional identity of the module was immune, not ferroptotic.

A methodological nuance matters here. The data structure did not satisfy the scale-free fit expected for standard weighted gene co-expression network analysis, with a maximum scale-free topology fit of R2 = 0.13, so the authors labeled the procedure Ward hierarchical co-expression clustering rather than WGCNA. A STRING v12 protein-protein interaction network built from the hub module contained 1,431 edges, and CCR5, with degree 51, and LCP2, with degree 47, ranked among the top ten nodes, reinforcing their central position in an immune-associated interaction network.

To resolve which cell types produced the bulk-tissue signal, the team applied MCP-counter immune scoring and single-cell localization. Paired MCP-counter scores for cytotoxic lymphocytes, monocytic lineage cells, endothelial cells, NK cells, T cells, myeloid dendritic cells, and fibroblasts were all higher in tumors, while B-lineage and neutrophil scores were lower. After leave-one-out scoring designed to reduce circularity, CD8A correlated with the cytotoxic-lymphocyte score and CSF1R with the monocytic-lineage score, and MCP-counter and ssGSEA estimates were highly concordant across all nine cell populations. In the single-cell dataset, quality control retained 1,683 of 1,704 cells, dominated by 995 tumor cells, 517 macrophages, and 59 CD8 T cells. CCR5, CSF1R, and CD86 were enriched in macrophage-lineage populations, CD8A showed the highest positivity in T/NK and CD8 T cells, and LCP2 was prominent in macrophage-lineage compartments as well.

Independent validation in TCGA-KIRC confirmed significant upregulation of all five genes in tumors versus normal tissue, with log2 fold changes between 1.71 and 2.77. Diagnostic ROC analysis yielded area-under-curve values of 0.948 for LCP2, 0.929 for CD8A, 0.922 for CCR5, 0.910 for CD86, and 0.862 for CSF1R, while a five-gene panel defined as the mean of standardized expression z scores reached 0.932. In GSE53757, single-gene AUCs ranged from 0.86 to 0.95. Hub-gene expression correlated positively with immune markers including CD8A, GZMB, PDCD1, and CD68. These are diagnostic discrimination figures for tumor versus normal tissue in retrospective cohorts, not validated clinical test performance.

The survival findings were negative, and the authors treat that as informative. In a survival dataset of 538 patients with 173 deaths, high-versus-low expression groups for the five genes showed no significant overall-survival differences, with log-rank p values between 0.41 and 0.93. In multivariable Cox analysis, age and advanced stage were significant, but the five-gene panel carried no independent prognostic value. The panel should therefore be described as a candidate diagnostic and immune-microenvironment marker set rather than a validated prognostic signature.

An in vitro qRT-PCR component with three biological replicates per group provided orthogonal experimental support. Relative expression of SLC7A11, GPX4, ACSL4, and HMOX1 was higher in the 786-O clear cell renal cell carcinoma line than in normal proximal tubule RPTEC cells, with fold changes of 2.55, 1.58, 1.70, and 2.30 respectively, and Welch t-tests on ΔCt values were significant for all genes. In a macrophage model, THP-1-derived macrophages exposed to 786-O-conditioned medium showed higher CCR5, CSF1R, and CD86 expression than M0 controls, with fold changes of 2.86, 1.71, and 1.95. These measurements are consistent with tumor-cell ferroptosis-related gene expression and with tumor-conditioned recruitment or activation of macrophage-lineage cells, though two in vitro models cannot establish mechanism.

The authors are explicit about the limits of interpretation. The principal evidence is transcriptomic and associative, so causal ferroptosis mechanisms cannot be inferred; MCP-counter scores are expression-based estimates rather than direct cell-fraction measurements; the single-cell analysis rests on a limited sample that needs larger independent cohorts; and exploratory diffusion-pseudotime ordering crossed multiple lineages and was not interpreted as a differentiation trajectory. Bulk expression of the five genes should be read primarily as a marker of immune-cell composition, since malignant epithelial cells do not overexpress every hub gene to the same degree. Whether an immune-rich tumor state is more or less susceptible to pharmacologic ferroptosis induction cannot be determined from expression alone and would require perturbation experiments with ferroptosis inducers or inhibitors, lipid-ROS measurements, labile iron assays, GPX4/SLC7A11 manipulation, and rescue experiments. Within those constraints, the study offers a cleanly scoped conclusion: CCR5, CSF1R, CD86, CD8A, and LCP2 are immune-microenvironment-associated hub biomarkers in clear cell renal cell carcinoma whose relationship to ferroptosis is supported at the pathway and co-expression levels, with candidate diagnostic and immune-stratification value that now awaits prospective testing.

Subject of Research: Immune microenvironment hub genes associated with ferroptosis-related transcriptional changes in clear cell renal cell carcinoma

Article Title: Immune microenvironment biomarkers linked to ferroptosis in clear cell renal cell carcinoma

Article References: Li, P., Liu, X., & Li, G. (2026). Immune microenvironment biomarkers linked to ferroptosis in clear cell renal cell carcinoma. Frontiers in Genetics, 17, Article 1938003. https://doi.org/10.3389/fgene.2026.1938003

Image Credits: AI Generated

DOI: 10.3389/fgene.2026.1938003

Keywords: clear cell renal cell carcinoma, ferroptosis, tumor immune microenvironment, biomarkers, single-cell RNA sequencing, TCGA-KIRC, MCP-counter, macrophages, CD8 T cells, diagnostic ROC, co-expression clustering, qRT-PCR