Background
Endogenous retrovirus (ERV) elements are genomic footprints of ancestral retroviral infections within the human genome. While the dysregulation of ERV transcription has been linked to immune cell infiltration in various cancers, its relationship with immune checkpoint inhibitor (ICI) response in solid tumors, particularly metastatic clear-cell renal cell carcinoma (ccRCC), remains inadequately explored.
Conclusions
Our findings introduce a dual ERV-based stratification system that effectively categorizes patient risk and predicts clinical outcomes for ccRCC patients undergoing ICI therapy. Beyond enhancing the predictive precision of existing transcriptomic models, this system paves the way for more targeted and individualized approaches in the realm of precision oncology.
Methods
This study analyzed patients with metastatic ccRCC from two prospective clinical trials, encompassing 181 patients receiving nivolumab in the CheckMate trials (-009 to -010 and -025) and 48 patients treated with the ipilimumab-nivolumab combination in the BIONIKK trial. ERV expression was quantified using the ERVmap algorithm from RNA sequencing data. Our primary objective was to correlate ERV expression with progression-free survival, with overall survival and time-to-second-treatment survival as secondary endpoints. We used bootstrap methods with univariate Cox regression on 666 substantially expressed ERVs to evaluate their prognostic significance and stability.
Results
Our analysis centered on two ERVs, E4421_chr17 and E1659_chr4, which consistently exhibited opposing prognostic impacts across both cohorts. We developed a stratification system based on their median expression levels, categorizing patients into four ERV subgroups. These subgroups were further consolidated into a three-tier risk model that significantly correlated with ICI treatment outcomes. The most responsive ERV risk category showed enhanced endothelial cell infiltration, whereas the resistant category was characterized by higher levels of myeloid dendritic cells, regulatory T cells, myeloid-derived suppressor cells, and markers of T-cell exhaustion. Notably, this ERV-based classification outperformed traditional transcriptomic signatures in predicting ICI efficacy and showed further improvement when combined with epigenetic DNA methylation markers. Conclusions: Our findings introduce a dual ERV-based stratification system that effectively categorizes patient risk and predicts clinical outcomes for ccRCC patients undergoing ICI therapy. Beyond enhancing the predictive precision of existing transcriptomic models, this system paves the way for more targeted and individualized approaches in the realm of precision oncology.