This study investigated the role of tumor-associated macrophages (TAMs) in kidney cancer and used their gene expression signatures to build a survival prediction model. Macrophages are immune cells that normally help fight infection and clear debris, but in tumors they can be reprogrammed to actually help the cancer grow and spread.
The researchers identified eight distinct TAM signatures from single-cell RNA sequencing data, analyzed their relationships with kidney cancer biology, and ultimately developed a 27-gene risk model capable of predicting which patients are likely to have worse outcomes. The model was validated in independent patient cohorts.
Published in Cells in 2025, this research contributes to the growing understanding of how the tumor's immune environment shapes cancer behavior, and highlights specific macrophage subpopulations that may be targetable in future therapies.
Macrophages are one of the most abundant immune cell types found within solid tumors. They are highly adaptable cells that can shift their behavior in response to signals in their environment. In normal tissues, they serve protective roles. But within tumors, they are often reprogrammed by cancer cells into forms that promote tumor growth, suppress other immune cells, and help cancer spread to new sites.
This reprogramming is not a single binary switch. Research using single-cell RNA sequencing, which reads the gene activity of individual cells rather than averaging across millions of cells, has revealed that tumor macrophages exist in many different states, each with a distinct gene expression profile and functional role in the tumor ecosystem.
In kidney cancer specifically, macrophage infiltration has been linked to both aggressive disease and potentially to the response to immunotherapy. Understanding which macrophage subtypes are present and how many are in a tumor could help predict which patients will do poorly and which might respond to treatments that target the immune environment.
The study identified eight macrophage signatures from published single-cell RNA sequencing data: TAM1 through TAM8. Most of these signatures represent genuine tissue-resident or tumor-infiltrating macrophage states. However, two signatures, TAM5 and TAM6, showed gene expression patterns more similar to neutrophils (another type of immune cell) than to true macrophages, making them less suitable as macrophage biomarkers.
The remaining six signatures, TAM1, TAM2, TAM3, TAM4, TAM7, and TAM8, were found to be reliable indicators of actual macrophage infiltration in tumor tissue. The researchers used these six for the downstream analysis, ensuring that the prognostic model is based on genuine macrophage biology rather than potentially confounded measurements.
A Principal Component Analysis (PCA) of the six valid TAM signatures found that the first principal component (PC1) explained 90.8% of the variance in TAM infiltration across patients. This means that a single combined measure captures nearly all the relevant information from the six individual signatures, simplifying the biological picture considerably.
The first principal component (PC1) of TAM signatures did not just correlate with macrophage numbers. It also correlated strongly with the expression of both M1 macrophage markers (associated with inflammatory, anti-tumor activity) and M2 macrophage markers (associated with immunosuppressive, pro-tumor activity), as well as with immune checkpoint molecules such as PD-1, PD-L1, and CTLA4.
This finding suggests that tumors with high macrophage infiltration are not uniformly pro-tumor or anti-tumor, but rather contain a complex mixture of states. The simultaneous presence of M1 and M2 markers reflects the dynamic, heterogeneous nature of the tumor immune environment, where macrophages of different activation states coexist and interact.
The correlation with immune checkpoint expression is particularly important clinically. Checkpoint inhibitors are already used to treat advanced kidney cancer. The fact that high TAM infiltration tracks with checkpoint expression suggests that patients with high macrophage scores might be most likely to respond to, or most in need of, checkpoint inhibitor therapy.
To build the survival prediction model, the researchers used LASSO Cox regression with 10-fold cross-validation. This technique fits a survival prediction model while automatically eliminating less informative variables and using cross-validation to prevent overfitting (where a model performs well on training data but fails on new patients).
The final model contained 27 genes, selected from the broader pool of TAM-associated genes. Patients were divided into high-risk and low-risk groups based on their model score. The overall hazard ratio between high-risk and low-risk groups was 4.64, meaning that high-risk patients had more than four times the hazard (instantaneous risk) of dying compared to low-risk patients. This difference was highly statistically significant (p less than 0.001).
The model was tested in the main TCGA-KIRC cohort and validated independently in the E-MTAB-1980 cohort, a European dataset collected from a different population in a different country. Consistent performance across these two independent cohorts strengthens confidence that the model reflects real biology rather than dataset-specific artifacts.
The 27-gene model showed strong discriminatory ability, with AUC values of 0.82 at 1 year, 0.79 at 3 years, and 0.82 at 8 years. These consistently high values indicate that the model performs well not just in the short term but over an extended follow-up period, which is important for counseling patients about their long-term outlook.
Strikingly, high-risk patients in this model had higher immune scores and more infiltrating immune cells overall, including more M1 macrophages, CD8+ T cells (cytotoxic immune cells that normally kill cancer), and regulatory T cells (Tregs). This seems paradoxical: why would patients with more immune cells have worse outcomes?
The answer lies in a phenomenon called immune exhaustion and suppression. When immune cells are chronically activated in the tumor environment but unable to clear the cancer, they become exhausted and lose their killing ability. The simultaneous presence of many Tregs actively suppresses the CD8+ T cells. The result is an immune environment that looks active on paper but is functionally compromised, failing to control the tumor despite abundant immune cell infiltration.
The identification of an immune-exhausted, macrophage-rich tumor microenvironment in high-risk kidney cancer patients has direct treatment implications. These patients might benefit most from therapies that reinvigorate exhausted T cells or deplete immunosuppressive Tregs, approaches that are being actively explored in clinical trials for kidney and other cancers.
The high correlation of the risk score with immune checkpoint expression suggests that checkpoint inhibitor drugs such as nivolumab (targeting PD-1) and ipilimumab (targeting CTLA4) could be especially relevant for high-risk patients identified by this model. The model might eventually help select which patients should receive checkpoint inhibitors versus other treatment types.
The validation of the 27-gene model in the European E-MTAB-1980 cohort is encouraging for generalizability, but larger prospective studies across ethnically diverse populations are needed. As RNA sequencing becomes more routinely available in clinical oncology, TAM-based signatures like this one could become part of standard molecular profiling for kidney cancer patients, guiding individualized treatment decisions.