Early-phase dose and H-ARS severity-based response category algorithms using hematological biomarkers in baboon gamma and mixed-field neutron/gamma radiation models

INTRODUCTION: The purpose of this study was to determine early phase CBC markers that could predict radiation dose and hematopoietic acute radiation syndrome (H-ARS) severity in a male baboon radiation model, and to determine the translational relevancy of these findings derived with known radiation doses as it pertains to the human algorithms based on reconstructed doses. METHODS: A panel of hematologic biomarkers was measured from blood-cell samples collected at 0 to 5d after receiving 0-8 Gy of TBI exposures of gamma or mixed field (neutron/gamma) using a standard blood cell counter. The database was used in an initial stepwise multiple-regression model for estimating radiation dose using data from 1-3d. A logistic-regression model-fitting approach was also applied for estimating H-ARS severity using TBI and PBI radiation data consisting of gamma only or mixed field collected from 1-5d. RESULTS: = 0.6760 (p < .001) and accuracy ∼75%. The logistic regression H-ARS severity algorithm consisted of lymphocytes (p < .001) and platelets-natural log (p < .0001) with overall accuracy of 93.4 (0 vs 2.5 & 2 H-ARS), 96.6 (0 vs 2.5 H-ARS), and 99.3 (0 vs 3 H-ARS) %. The predicted H-ARS severity outcome for the 0-5-day models showed area under the curves of 0.974, 0.990, 0.995 respectively by ROC curve analysis. DISCUSSION: The resultant study supports the proof-of-concept that initial dose estimation and H-ARS severity-response-category algorithms using standard blood-cell parameters can provide both rapid triage and radiation injury assessment following gamma and mixed (neutron/gamma)-field exposures. CONCLUSION: These findings, based on a baboon radiation model, provide confirmatory evidence of the utility of hematological biomarkers predicting radiation dose and injury severity.