When cancer cells grow, they do not grow in isolation. A tumor is surrounded by a complex community of immune cells, blood vessels, and structural support tissue collectively called the tumor microenvironment (TME). Some immune cells in the TME actively fight the cancer. Others, often recruited or reprogrammed by the tumor itself, actively suppress anti-cancer immunity and help the tumor escape destruction.
Two particularly important types of immunosuppressive cells are myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs). MDSCs are a diverse group of immature immune cells that block the activity of tumor-killing T cells. Tregs are a specialized type of T cell that normally prevents the immune system from attacking the body's own tissues, but that cancer cells exploit to shut down anti-tumor responses.
In clear cell renal cell carcinoma (ccRCC), the most common kidney cancer subtype, high levels of MDSCs and Tregs in the tumor are associated with worse outcomes and reduced responses to immunotherapy drugs. However, identifying individual patients with high MDSC/Treg activity currently requires complex immune profiling. This study developed a six-gene signature measurable from standard gene expression data that serves as a surrogate for MDSC/Treg infiltration and predicts patient risk.
Gene expression data from 539 patients with ccRCC from The Cancer Genome Atlas (TCGA) was the primary dataset for this study. The researchers started by identifying which genes were specifically associated with MDSC and Treg cell populations in the tumor microenvironment, cross-referencing published immune cell gene signatures with the TCGA expression data to find genes whose activity correlated with MDSC/Treg infiltration.
Two machine learning approaches were combined to narrow down from many candidate genes to the most informative ones. LASSO (Least Absolute Shrinkage and Selection Operator) regression automatically penalizes models for including unnecessary genes, driving the coefficients of weakly contributing genes to zero and keeping only the strongest predictors. Random Forest was then applied to rank the remaining genes by their importance in predicting patient outcomes.
The intersection of genes selected by both methods produced six hub genes: WDFY4, IL16, FCGR1B, NOD2, RELT, and MKI67. These six genes were used to calculate a risk score for each patient. Patients with high scores were classified as high-risk and patients with low scores as low-risk. The model's performance was assessed using time-dependent AUC curves for survival prediction at 1, 3, and 5 years after diagnosis.
The study also examined the tumor immune microenvironment composition of high-risk versus low-risk patients, analyzed which immune checkpoint pathways were most active in each group, and screened a library of FDA-approved drugs to identify compounds that might specifically target the vulnerabilities of high-risk tumors.
The six-gene risk score achieved an AUC of 0.8 at 1 year for predicting which patients would survive, a strong performance that outperformed individual clinical variables such as tumor stage. High-risk patients had significantly shorter overall survival and progression-free survival compared to low-risk patients, and these differences held even after adjusting for age, tumor stage, and grade in multivariate analysis.
Immune microenvironment analysis confirmed that high-risk tumors had significantly higher infiltration of MDSCs and Tregs, validating the biological premise of the signature. These tumors also showed higher expression of multiple immune checkpoint molecules, including PD-L1, which is the target of several widely used immunotherapy drugs. However, despite higher checkpoint expression, high-risk patients showed worse outcomes with nivolumab (an anti-PD-1 immunotherapy) compared to low-risk patients.
This paradox, higher checkpoint expression but poorer immunotherapy response, likely reflects the deeply immunosuppressive environment in high-risk tumors. When so many MDSCs and Tregs are present, even removing the PD-1 checkpoint may not be enough to restore effective anti-tumor immunity. These patients may need combinations of therapies that also target MDSC and Treg activity directly.
WDFY4 is involved in antigen presentation, the process by which immune cells display pieces of pathogens or cancer cells to trigger an immune response. Altered WDFY4 expression may impair how well the immune system recognizes and targets tumor cells. IL16 is a cytokine that attracts T cells to sites of inflammation; its dysregulation in the tumor microenvironment can skew which types of T cells are recruited, potentially favoring Tregs over cancer-killing cytotoxic T cells.
FCGR1B encodes a receptor found on immune cells that binds antibodies, playing a role in antibody-dependent immune responses. NOD2 is an intracellular receptor that detects bacterial components and activates immune signaling pathways. Its activity in the tumor microenvironment connects innate immune sensing with the broader immunosuppressive state. RELT is a member of the TNF receptor family involved in immune cell signaling and survival.
MKI67 encodes the Ki-67 protein, which is a widely used marker of cell proliferation. Higher MKI67 expression indicates that tumor cells are dividing more rapidly, and its inclusion in the signature connects the immunosuppressive tumor environment with proliferative tumor activity. Tumors that are both rapidly proliferating and heavily immunosuppressive represent the most challenging combination for current treatments.
The drug screening analysis identified 12 potential FDA-approved compounds that showed predicted activity against high-risk tumors based on their known mechanisms of action and the molecular vulnerabilities of high-risk ccRCC. While these are computational predictions that require experimental validation, they provide a prioritized list for researchers to test in laboratory and clinical studies.
For patients classified as high-risk by this six-gene score, the findings suggest that standard immunotherapy alone may be insufficient. Combination strategies that simultaneously target immune checkpoints and suppress MDSC or Treg activity may be more effective. Several clinical trials are already exploring such combinations, and this risk score could help select the patients most likely to benefit from enrollment in those trials.
The practical implementation of this signature would require measuring the expression of six genes from a tumor tissue sample, a routine laboratory procedure in centers with molecular pathology capabilities. As tumor profiling becomes more standardized in cancer care, incorporating this kind of immune risk signature into clinical decision-making becomes increasingly feasible. Prospective validation studies are the essential next step before clinical adoption.