Bayesian methods for the evaluation of Tritium: Relative Biological Effectiveness and cancer risk

ALL RIGHTS RESERVED iii ABSTRACT Ghassan Badri Hamra: Bayesian methods for the evaluation of Tritium: Relative Biological Effectiveness and cancer risk. Tritium is a radioisotope of hydrogen that releases -energy, which is a form of ionizing radiation. Tritium is understudied due to a lack of epidemiological data on human exposure, despite the fact that exposure to low energy ionizing radiation is ubiquitous in the environment. In occupational studies of nuclear workers, tritium is usually aggregated with gamma radiation and examined under the assumption that the cancer risk per unit exposure of each is equivalent. However, a recent systematic review of the literature suggests that beta radiation is more biologically effective than gamma radiation. We utilize Bayesian methods to inform estimation of cancer risk for tritium as well as the RBE of tritium compared to gamma radiation using information compiled at the Savannah River Site (SRS) nuclear fuel facility in Aiken, SC. SRS staff maintained detailed records of personnel dosimetry that have been utilized in previous epidemiological studies. Included is information on employment history, radiation exposure, including separate gamma and tritium radiation dosimetry records, as well as vital status information. We calculate the excess relative rate of leukemia and leukemia excluding chronic iv lymphocytic leukemia (CLL) associated with tritium and gamma radiation. The ERR/10mGy (90% HPD) of leukemia associated with tritium and gamma radiation are relative biological effectiveness of tritium relative to gamma radiation (RBE) of 6.24 the range of plausible values suggested by others. Our results utilize evidence from in vitro and in vivo research to inform estimation of the risk of cancer associated with tritium by incorporating knowledge of the direction and magnitude of tritium " s relationship to cancer compared to gamma radiation. This illustrates a simple approach for using Bayesian methods to integrate external knowledge into epidemiological studies without the need to specify estimates of the risk based on research that cannot be easily translated from experimental animal and cellular models into human risk models. v ACKNOWLEDGEMENTS I am grateful for a number of people without whom this dissertation would not have been possible. First and foremost, my wife, Leah Hope Schinasi, for explaining epidemiologic concepts to me before I even knew I would grow up to be an epidemiologist, and for providing mental support through this process that was much needed. Next, I would like to thank my committee chair, Dr. David Richardson, who has been …