Human Model To Predict The Outcome Of Cancer Vaccine Clinical Trials.

e14298 Background: Cancer vaccines activate T-cells with peptides presented by HLA alleles. We hypothesized that peptides binding to multiple HLA alleles of a patient are more likely to induce T-cell responses than epitopes presented by a single HLA. To prove this hypothesis we predicted the outcome of vaccine clinical trials in a Model Population of subjects with complete HLA genotype. Methods: We collected the immune- and objective response rate (IRR, ORR) results from 64 clinical trials conducted in 1,790 patients treated with 42 different vaccines. From these 42 vaccines, we identified personal epitopes (PEPIs) with strong binding affinity to at least 3 HLA alleles of each subject in the Model Population and calculated the proportion of subjects with PEPIs (PEPI Score), multiple PEPIs (MultiPEPI-Score) and PEPIs against multiple vaccine antigens (MultiAg PEPI-Score). Results: Highly significant correlation between the PEPI Score and IRR (p = 0.001) suggests that peptides binding to at least 3 HLA alleles of the subject induce T-cell responses. MultiPEPI-Score correlated with ORR when the tumor of patients expressed the target antigen (p = 0.01). MultiAg PEPI-Score correlated with ORR of clinical trials conducted with vaccines containing multiple antigens (p = 0.01). Conclusions: Our results suggest that (i) a patient’s HLA genotype is the main determinant of vaccine response and (ii) T-cell responses against multiple antigens expressed in the patient’s tumor favors tumor shrinkage. We introduced a novel pre-clinical human model that accurately predicts the outcome of clinical trials, will reduce the risk, cost and time of vaccine development for the prevention and treatment of cancer.