Deep learning using tumor HLA peptide mass spectrometry datasets improves neoantigen identification.

Neoantigens, which are expressed on tumor cells, are one of the main targets of an effective antitumor T-cell response. Cancer immunotherapies to target neoantigens are of growing interest and are in early human trials, but methods to identify neoantigens either require invasive or difficult-to-obtain clinical specimens, require the screening of hundreds to thousands of synthetic peptides or tandem minigenes, or are only relevant to specific human leukocyte antigen (HLA) alleles. We apply deep learning to a large (N = 74 patients) HLA peptide and genomic dataset from various human tumors to create a computational model of antigen presentation for neoantigen prediction. We show that our model, named EDGE, increases the positive predictive value of HLA antigen prediction by up to ninefold. We apply EDGE to enable identification of neoantigens and neoantigen-reactive T cells using routine clinical specimens and small numbers of synthetic peptides for most common HLA alleles. EDGE could enable an improved ability to develop neoantigen-targeted immunotherapies for cancer patients.

Randomized, multicenter, phase II study of CO-101 versus gemcitabine in patients with metastatic pancreatic ductal adenocarcinoma: including a prospective evaluation of the role of hENT1 in gemcitabine or CO-101 sensitivity.

PURPOSE: Gemcitabine requires transporter proteins to cross cell membranes. Low expression of human equilibrative nucleoside transporter-1 (hENT1) may result in gemcitabine resistance in pancreatic ductal adenocarcinoma (PDAC). CO-101, a lipid-drug conjugate of gemcitabine, was rationally designed to enter cells independently of hENT1. We conducted a randomized controlled trial to determine whether CO-101 improved survival versus gemcitabine in patients with metastatic PDAC (mPDAC) with low hENT1. The study also tested the hypothesis that gemcitabine is more active in patients with mPDAC tumors with high versus low hENT1 expression. PATIENTS AND METHODS: Patients were randomly assigned to CO-101 or gemcitabine, after providing a metastasis sample for blinded hENT1 assessment. An immunohistochemistry test measuring tumor hENT1 was developed. To dichotomize the population, an hENT1 cutoff value was defined using primary PDAC samples from an adjuvant trial, and a high/low cutoff was applied. The primary end point was overall survival (OS) in the low hENT1 subgroup. RESULTS: Of 367 patients enrolled, hENT1 status was measured in 358 patients (97.5%). Two hundred thirty-two (64.8%) of 358 patients were hENT1 low. There was no difference in OS between treatments in the low hENT1 subgroup or overall, with hazard ratios (HRs) of 0.994 (95% CI, 0.746 to 1.326) and 1.072 (95% CI, 0.856 to 1.344), respectively. The toxicity profiles in both arms were similar. Within the gemcitabine arm, there was no difference in survival between the high and low hENT1 subgroups (HR, 1.147; 95% CI, 0.809 to 1.626). CONCLUSION: CO-101 is not superior to gemcitabine in patients with mPDAC and low tumor hENT1. Metastasis hENT1 expression did not predict gemcitabine outcome.