Table of Contents
Risk Assessment and Risk Management Implications of Hormesis

Carl J. Paperiello, Ph.D., Director

Office of Nuclear Material Safety and Safeguards, U.S. Nuclear Regulatory Commission

Tel: 301-415-7800, Fax 301-415-5370

As a regulatory agency the U.S. Nuclear Regulatory Commission's mission is to protect public health and safety and the common defense and security of the United States. One of the NRC's tasks is to ensure the safety and health of workers and the public from exposure to ionizing radiation from the use of those nuclear materials regulated under the Atomic Energy Act. The task of ensuring safety requires the NRC to take two actions. The first is to determine the risk from exposure to ionizing radiation while the second is to establish the acceptability of the risk. We can refer to these as risk assessment and risk management and they should not be confused. Risk assessment tends to be the more objective and scientific while risk management is value laden and strongly affected by societal and ethical views.

In the United States, the regulation of ionizing radiation protection standards is spread among a number of state and federal agencies. The scientific bases for radiation protection standards are included in the recommendations of several committees including: the International Commission of Radiological Protection(ICRP), the United Nations Scientific Committee on the Effects of Atomic Radiation(UNSCEAR), the United States National Research Council's Committee on the Biological Effects of Ionizing Radiation(BEIR) and the National Council on Radiation Protection and Measurements(NCRP). The assessment of risks from ionizing radiation by all of these committees is based on the Linear No-Threshold Model(LNT). In this model the stochastic consequences of radiation are considered to be a linear function of dose and except for one small correction, almost completely independent of dose rate.

The Linear No-Threshold Model, the current scientific paradigm for the effect of low level ionizing radiation, is used by the Nuclear Regulatory Commission for assessing radiation risks and performing cost-benefit analyses to assist in determining the consequences of regulatory actions. It is instructive to trace the evolution of this hypothesis through one committee in the United States because of the uncertainty in the model at some of the dose limits at which regulatory controls are imposed. The 1972 BEIR Report adopted the linear no-threshold model on practical grounds. It noted that “such estimates are fraught with uncertainty.” In the 1980 Beir III Report, the BEIR Committee noted that the scientific basis for making estimates of the carcinogenic risk of low-dose, low-LET radiation was inadequate but public policy and the exercise of regulatory authority required a position on the probable cancer risk. The Committee also added it did not know whether dose rates of about 100 mrads/yr were detrimental to man.

The BEIR Committee in 1990, in Report V, stated that the derivation of risk estimates for low doses and dose rates through any type of risk model involved assumptions that remained to be validated. The Committee stated the epidemiological data could not rigorously exclude the existence of a threshold in the millisievert(100 millirem) dose range. The Committee stated that the possibility that there were no risks from exposures comparable to natural background could not be ruled out and the lower range of uncertainty in the risk estimates extended to zero.

The arguments in support of the LNT model are based on plausibility. In addition, the concept of a collective dose, which is invalid without the LNT model, provides regulators with a seductively simple way to measure risk. However, science frequently discards plausible hypotheses. Proof is needed or evidence to the contrary. I see two possible approaches to test this hypothesis. The first is careful epidemiology. While the “background” cancer incidence may be too high to statistically demonstrate the validity of the LNT hypothesis at doses between background and occupational limits, statistical demonstration of the validity of an alternative hypothesis such as hormesis would invalidate the LNT hypothesis. The second approach is to calculate from first principles the observed cancer rates at high doses and dose rates. This would require a quantitatively predictive model based on fundamental molecular biology with a minimum of fitted parameters that could quantitatively predict the observed cancer rates, the differential organ sensitivities, and variation in age sensitivity. As I will note later, a quantitatively predictive model may be necessary for a regulator to exploit alternative models such as hormesis, a threshold or a radically nonlinear dose response model.

As discussed previously, regulation involves risk management as well as risk assessment. In our representative democracy, the Commission, appointed by the President and confirmed by the Senate, represents the people of the United States. The management of risks, as contrasted to the analysis of risks, is determined by a variety of considerations some of which may not always be enunciated. These include a variety of non-economic judgements concerning the value of an activity as well as formal or informal cost-benefit judgements. Depending on the scope of the activity, non-quantifiable individual, societal, and political values will influence risk management judgements. One can look into one's own risk judgements and see the validity of this statement. The public's toleration of risk is affected by the public's view of the availability and desirability of alternatives; the perceived degree of personal control over the risk; the dread of the consequences(i.e., painful or bizarre deaths are feared more than other causes of death); and the fairness or equity in the distribution of risks and benefits. The public through the political process informs the regulator what risk is tolerable. Risk assessment will aid the Commission in their decision making but the Commission must consider all factors in their decisions.

How might a change in the LNT paradigm change the regulation of radiation exposure in the United States? This could occur not only through the demonstration of hormesis but also by the demonstration of a threshold or a severe nonlinear dose response curve. We need to recognize that a shift in the paradigm would need the support of the established scientific community. I will not try to define who this community is. Clearly, the ICPR, NCRP, UNSCEAR or the BEIR committees would qualify. However, one could not exclude some other set of organizations driving a change in the paradigm. However, I doubt that any responsible regulator would change the scientific basis for regulation on their own and without substantial scientific support.

Assuming the above did occur, and the LNT model were replaced with an hormetic model, what would happen? First, the threshold for transition from benefit to harm would have to be determined along with the risk coefficient above that threshold. If the threshold were not high enough, there might be no practical value in hormesis. Alternatively, the threshold might be high enough to raise public and environmental limits but not affect occupational exposure standards. Secondly, dose rate dependence would have to be determined. The LNT model conservatively assumes that risk is almost completely independent of a dose rate. For chemical hormesis at least there appears to be dose rate dependence(i.e., heavy trace metals). Thirdly, any synergistic effect with hormetic carcinogens such as chemicals might have to be considered. For example if radiation effects below a certain point x were hormetic and exposure to a certain chemical below a point y were also hormetic, would a combined exposure slightly below x plus y also be hormetic or would the threshold into harm be crossed? While a similar problem exists for linear models, the risks are usually added absent empirical evidence to the contrary although synergism also is frequently discussed.

Of technical note, the entire ICRP dosimetry system in place since 1977, and used by all U.S.regulators, as well as foreign counterparts is based on absolute linearity, not only for the total human organism but for individual organs. Loss of the LNT model would result in 20 years worth of calculational work being discarded as well as every environmental analysis dependent upon this dosimetry.

Lastly, for the NRC, sources of radiation other than those from NRC regulated activities might have to be considered. For example, if the point for a transition from a beneficial dose to a harmful dose were five centigray( 5 rad), the additive risk to an individual from a four centigray( 4 rad) occupational exposure would be different from the additive risk from the same occupational exposure to another individual additionally receiving a three-centigray ( 3 rad) medical exposure. True, this last scenario is likely to be complicated by consideration of dose rate effects and timing of the doses. Given enough spacing between exposure to provide time for repair, both individuals may remain below the threshold for harm. However, this would have to be determined. In the linear model, although the individual receiving both the occupational exposure and the medical exposure would have the higher risk, the assumed additive risk from identical occupational exposure would be equal. Depending on the transition threshold, a similar example could be created for manmade environmental exposures in the presence of natural background.

I would hope that these observations do not imply that I am skeptical of all radiation effect models that are not LNT. I am certainly not. However, a determination that the biological response to ionizing radiation follows a hormetic rather than an LNT dose response model may be difficult to translate into a revision of radiation protection standards. It will make the regulator's task more difficult. The above questions and probably many more will be raised. Recall that the linear hypothesis was originally formulated as a conservative assumption that placed an upper limit on risk for the purpose of informing regulators as risk managers. Some may say that the regulator should maintain conservative limits based on the linear model until the regulator knows exactly what happens. Much research might be needed before an alternative paradigm could be implemented fully. Speaking for myself, I suspect that a demonstration of hormesis, or a threshold, or significant non-linearity will initially stop the movement toward increasingly more stringent radiation protection dose limits. It could significantly reduce emphasis on As-Low-As Reasonably Achievable (ALARA) practices. Depending on the magnitude of the threshold dose, environmental dose limit changes derived from natural background levels may be more likely than occupational dose limit changes. In addition, since hormetic effects are based on repair stimulation by many small doses, the trend in the past 50 years to base occupational standards on annual limits rather than weekly or monthly limits might be reversed.

I do not think the challenge is easy. The current LNT model evolved over a period of 50 years. If past history is predictive of the future, change could be comparably slow.

The opinions above represent the thoughts of the author and do not necessarily represent those of the NRC.