Serum Circulating MicroRNAs as Functional Biodosimeters in Patients Undergoing Total Body Irradiation

Abstract

In a radiation exposure event, life-threatening doses of ionizing radiation may be delivered in a total-body irradiation (TBI) pattern. To triage victims and implement countermeasures, we need clinically feasible biodosimeters. Previously, we used mouse and non-human primate models to show that serum microRNAs may predict the biological impact of lethal and sublethal radiation doses. We hypothesized that these results can be replicated in humans subjected to TBI. We obtained sera from 25 patients who underwent allogeneic stem-cell transplantation conditioned with TBI (2 Gy/fx, total 12 Gy) in two radiation oncology centers. The sera were collected at the following timepoints: no irradiation, low dose (2 or 4 Gy; n = 25) and high dose (12 Gy, n = 9). We profiled microRNA expression in these samples using next-generation sequencing (miRNA-seq). Additionally, we isolated exosomes from paired serum samples of 6 patients at 0 Gy and 12 Gy and performed miRNA-seq on their contents. After quality control (filtering out samples with <350 miRNAs with non-zero expression), we deemed 68 of 71 samples suitable for downstream analysis. Differential expression results were largely consistent with previous non-human studies, with miR-150-5p significantly downregulated and miR-375 together with miR-200 and miR-30 families upregulated following exposure. Serum microRNA signatures showed near-perfect separation of irradiated and non-irradiated samples in the low dose setting (in cross-validation AUC 0.87, 95% confidence interval 0.76-0.98). Model built using 9 high dose vs 22 non-irradiated samples yielded in cross-validation AUC 0.86 (95% confidence interval 0.68-1.0). Paired analyses in low- and high dose settings allowed us to identify several microRNAs, including miR-150-5p, whose expression decreased in a linear manner with respect to radiation dose, identifying them as potential biodosimeters. Additionally, many of the differentially expressed miRNAs have been previously shown to directly target candidate biodosimeter proteins such as ATM (miR-181 family) or TP53 (miR-150-5p, miR-375, miR-30 families). MicroRNAs identified as differentially expressed in exosomes were largely consistent in terms of change with that observed in serum samples, however, due to low power, only miR-130a-3p was significantly up- and miR-193a-5p significantly downregulated (FDR<0.05). We confirm that serum microRNAs reflect radiation dose exposure of humans undergoing TBI and thus may be used as functional biodosimeters for precise identification of people exposed to clinically significant amounts of radiation. These findings have important implications in emergency medicine and in radiation oncology.