Back to 2015 Joint Annual Meeting

Introduction:Renal cell carcinoma (RCC) is a common, immunogenic urological malignancy. Missense mutations may result in increased tumor-specific neoantigens and tumor immunoreactivity. The relationship between tumor-specific neoantigens and pathology is unknown. Our aim was to evaluate the relationship between neoantigens and clinicopathologic outcomes using The Cancer Genome Atlas (TCGA).
Materials & Methods:TCGA was queried for RCC data that contained whole exome sequencing (WES), transcriptome sequencing (RNA-seq), and clinicopathologic outcomes. For each sample, all possible missense mutations were identified and all possible mutated peptides (neoantigens) were evaluate for immunogenicity. HLA type was computationally predicted from RNA-seq data. HLA-antigen binding affinity was computed using the neural-network based software. All neoantigens with an IC50 < 500nM were considered capable of eliciting an immune response. Pathologic characteristics of the tumor were correlated with neoantigen count.
Results:147 RCC cases met inclusion criteria. Overall, neoantigen counts ranged from 0-198 (mean=36, SD=39). Low-grade (Fuhrman 1&2) tumors had significantly more immunogenic neoantigens than high-grade (Fuhrman 3&4) tumors (41 neoantigens vs 31, p=0.05). Localized tumors (T1 or T2) had significantly more neoantigens than non-localized (T3 or T4) tumors (40 vs 29, p=0.05). Low-grade, localized tumors had significantly more neoantigens compared to high-grade and/or non-localized (45 vs. 32, p=0.03).
Conclusions:Computational prediction of potentially-immnunogenic neoantigens demonstrates that lower-risk tumors are associated with an increased number of neoantigens. Experimental validation of neoantigens is necessary for further refinement of their prediction and development of immnotherapeutics.