Gene expression signatures that predict radiation exposure in mice and humans

Background: The capacity to assess environmental inputs to biological phenotypes is limited by methods that can accurately and quantitatively measure these contributions. One such example can be seen in the context of exposure to ionizing radiation. Methods and Findings: We have made use of gene expression analysis of peripheral blood (PB) mononuclear cells to develop expression profiles that accurately reflect prior radiation exposure. We demonstrate that expression profiles can be developed that not only predict radiation exposure in mice but also distinguish the level of radiation exposure, ranging from 50 cGy to 1,000 cGy. Likewise, a molecular signature of radiation response developed solely from irradiated human patient samples can predict and distinguish irradiated human PB samples from nonirradiated samples with an accuracy of 90%, sensitivity of 85%, and specificity of 94%. We further demonstrate that a radiation profile developed in the mouse can correctly distinguish PB samples from irradiated and nonirradiated human patients with an accuracy of 77%, sensitivity of 82%, and specificity of 75%. Taken together, these data demonstrate that molecular profiles can be generated that are highly predictive of different levels of radiation exposure in mice and humans. Conclusions: We suggest that this approach, with additional refinement, could provide a method to assess the effects of various environmental inputs into biological phenotypes as well as providing a more practical application of a rapid molecular screening test for the diagnosis of radiation exposure. © 2007 Dressman et al.