The health effects of low doses of ionizing radiation, such as those received by radiation workers and most patients in healthcare, have been a controversial issue for decades.
A recent report from the Washington, DC-based National Research Council, referred to as the "BEIR VII" (Biological Effects of Ionizing Radiation) report, is the most up-to-date publication that addresses this issue. The National Research Council is the principal research arm of the U.S. National Academy of Sciences and the U.S. National Academy of Engineering. The report, which incorporates nearly 15 years of new data, was sponsored by the U.S. Departments of Defense, Energy and Homeland Security; the U.S. Nuclear Regulatory Commission; and the U.S. Environmental Protection Agency.
The most thoroughly studied population exposed to ionizing radiation is the survivors of the Hiroshima and Nagasaki atomic explosions. Smaller populations that were also considered in the BEIR VII report include persons medically and occupationally exposed to radiation. Excess cancers (the numbers of cancers above the expected level in the population) have been observed in these populations for individuals receiving doses of 100-4,000 mSv, which is 40-1,600 times greater than the average annual background exposure worldwide from all radiation sources. In cases of in utero exposure, excess cancers have been detected at doses as low as 10 mSv.
Data for solid tumors show an increasing rate of occurrence with increasing dose that is consistent with a linear hypothesis of radiation injury; i.e., doubling the dose causes a doubling in the number of cancers. Leukemia incidence displays a linear-quadratic relationship with dose, rather than a simple linear relationship. These data are statistically reliable, although their precision is compromised because cancer is a frequently occurring disease, and radiation-induced cancers do not differ from cancers resulting from other causes.
The challenge is to identify the appropriate model to extrapolate from bioeffects revealed at doses of 100-4,000 mSv to bioeffects that might occur at much lower doses (doses below 100 mSv are defined in the BEIR VII report as low doses). This extrapolation is the subject of the BEIR VII report, as it has been for previous BEIR reports (particularly BEIR III and BEIR V). Of particular interest are estimates of risk at doses of 10 mSv and below, which covers the range of most personnel and patient exposures in medicine.
Various models are examined in the BEIR VII report. The model used to date, and examined in detail in the BEIR VII report, is referred to as the linear no-threshold model. This model assumes that the risk of exposure at higher doses should be extrapolated in a straight-line fashion through the risk/dose origin. With this model, no dose is so low that a risk of radiation-induced cancer does not occur.
Another model, the so-called supralinear model, is based on the assumption that low doses are substantially more harmful that estimated by the linear no-threshold model. This model was discarded because the BEIR VII committee found no compelling evidence, either physical or biological, that supports it.
The hormesis or threshold model was also examined by the committee. This model presumes that low doses are substantially less harmful than estimated by the linear no-threshold model, and in fact may be nonexistent at very low doses. That is, a threshold may exist below which no effects occur, or below which exposure to ionizing radiation may even be healthful, an effect known as radiation hormesis.
Analogies to radiation hormesis are aspirin and wine. One aspirin or one glass of wine a day may improve health, whereas 1,000 aspirins or glasses a day would be lethal. After examining epidemiological, biological, and physical data, the committee concluded that existing evidence does not support a hormesis or threshold model for radiation injury.
The BEIR VII committee concluded that the linear extrapolation of risk from intermediate to low doses remains the most appropriate model for estimating radiation risk (i.e., the development of solid cancers) at low levels of exposure. As stated in the report: "The Committee judges that the balance of evidence from epidemiologic, animal and mechanistic studies tends to favor a simple proportionate relationship at low doses between radiation dose and cancer risk."
Evidence supporting this view was collected from animal studies and from models of radiation damage at the cellular and molecular levels.
The BEIR VII report concludes further that it is unlikely that a threshold exists for the induction of cancers, but notes that the risk of occurrence of radiation-induced cancers is small at low doses. Although other health effects (e.g., heart disease and stroke) may occur at intermediate doses and above, additional data are needed to determine and quantify whether there is a risk of these effects at low doses.
The committee confirmed that no health effects have been detected in humans that are attributable to genetic mutations. However, such mutations have been noted in mice and other organisms, and there is no reason to believe that humans are immune to such effects.
Most persons involved in radiation protection support the use of the linear no-threshold model for establishing standards for protection against ionizing radiation. When establishing exposure standards and limits, a conservative model of risk, such as the linear no-threshold model, should be used. That is, the model should overestimate rather than underestimate the risk in situations where personnel exposures are being controlled and the prediction of risk is inexact.
Exposure guidelines for radiation workers and members of the public are established by using the linear no-threshold model to set exposure limits so that the risks to exposed individuals are acceptably small, and further reductions would be time-consuming and financially costly. This approach to setting dose limits is known as ALARA (as low as reasonably achievable). Radiologists and other healthcare providers should be aware of and always follow the ALARA principle.
From data at intermediate doses and the linear no-threshold model, an estimate of excess cancer risk per unit dose can be computed. In BEIR VII, this estimate is one out of every 100 persons receiving a dose of 100 mSv would develop a radiation-induced fatal or nonfatal cancer (solid cancer or leukemia) over a 70-year lifetime. Since 42 of these 100 people will develop cancer for other reasons, the relative risk of radiation-induced cancer is 2.4%. A dose of 100 mSv is quite large, an order of magnitude or more beyond the doses received annually by radiation workers and most patients. For a dose of 10 mSv at the upper edge of most personnel and patient exposures, the relative risk of radiation-induced cancer is 0.24%. It must be emphasized that these risk estimates are speculative -- they are only as valid as the higher-dose data and the linear no-threshold model from which they are established.
The estimated risks of solid cancers and leukemia are about the same in BEIR VII as they were in BEIR V, and as they are reported in recent United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR, Report Vol. II: "Effects, Sources and Effects of Ionizing Radiation," New York City, United Nations, 2000) and International Commission on Radiological Protection ("Biological and Epidemiological Information on Health Risks Attributable to Ionizing Radiation: A Summary of Judgments for the Purposes of Radiological Protection of Humans") reports. BEIR VII data demonstrate a higher risk (about 62% higher) of solid cancers in exposed women compared with men, probably because of a higher risk of cancer of the breast and female organs.
It is one thing to use a conservative model to estimate risk and establish protection standards, and another to use the risk estimates to estimate the numbers of excess cancers and cancer deaths induced in large populations of exposed individuals. BEIR VII reports that in a population of 100,000 persons receiving a dose of 100 mSv (the report uses the unit mGy for this calculation, but for the x and g radiations employed routinely in health care, 1 mGy = 1 mSv), 2,100 excess cases of solid cancers will occur, as will 170 excess cases of leukemia. Cases in the absence of radiation would be 82, 400, and 1,420, respectively. That is, the relative cancer risk of 2.4% per 100 mSv noted previously would, according to these numbers, cause 2,270 additional persons to experience cancer over their lifetimes in a population of 100,000 persons if each received a dose of 100 mSv.
Such numbers are the substance of headlines -- invariably reported as fact rather than as highly speculative estimates. Because they are reported as fact, they contribute to a societal fear of radiation that has yielded a decades-long moratorium on nuclear power that reinforces our continued dependence on imported oil -- which is a real, not a speculative, risk. This fear also causes some individuals to defer medical radiation procedures to detect and diagnose diseases that constitute a real threat to their health.
The BEIR VII report offers several recommendations for additional research into the health effects of radiation exposure. To the degree that this research adds new knowledge about fundamental biological mechanisms, additional support is justified. However, additional research conducted simply to refine risk estimates and further justify the linear no-threshold model of radiation injury should be reconsidered. With ever-tightening medical research budgets, we should instead be making investments into biomedical research that could ultimately have a greater societal impact.
By William R. Hendee, Ph.D.
AuntMinnie.com contributing writer
November 18, 2005
William Hendee, Ph.D., is a guest editor of health policy for the American Roentgen Ray Society (ARRS). This article originally appeared in the American Journal of Roentgenology (October 2005, Vol. 185:4). Reprinted by permission of the ARRS.
Copyright © 2005 American Roentgen Ray Society