e13132 Background: Association between certain HLA types and cancer is well known. We hypothesized that the number of epitopes of tumor antigens presented by autologous HLAs characterizes a patient’s capacity to kill tumor cells. These CD8+ T cell epitopes are presented by 6 out of >13,000 known HLA class I alleles and induce extremely variable tumor-specific T-cell responses. Methods: We predicted HLA-binding epitopes from 48 frequently expressed tumor antigens in subjects characterized with 6 HLA class I alleles. To develop the “HLA Score” cancer risk predictor we used epitopes binding to multiple HLAs of a subject. The predictor was trained on 5,789 non-American subjects. To identify populations with high and low immunogenetic risk to develop cancer we compared the HLA Score of American cancer subjects to a general population of American subjects (1,400 subjects). The performance of the predictor was characterized with AUC and Risk Ratio. Due to Bonferroni correction, AUC values with p-value p<0.007 was accepted as significant. Results: “HLA Score” predictor significantly separated cancer patients from the general population in six out of the seven investigated cancer types. The Risk Ratios between the most protected and most at risk subpopulations ranged between 2.38 and 5.67 (Table). Cancer risk prediction with HLA Score. Conclusions: Subjects with certain HLA class I alleles have high risk of developing cancer. The novel “HLA Score” predictor we introduced here could complement current testing used for determination of the genetic risk of cancer.[Table: see text]
Characterization of the peptide binding specificity of the HLA class I alleles B*38:01 and B*39:06
