ered at low dose rates, this nonneoplastic component of life shortening has not been observed (Ullrich and Storer 1979a). A few instances have been reported of apparent radiation-induced life lengthening following exposure to low levels of single or protracted doses of radiation (NCRP 1980). Statistical analyses of the distribution of deaths in these studies indicate that control animals usually show a greater variance around the mean survival time than the groups exposed to low doses of radiation (NCRP 1980). In addition, the longer-living irradiated animals generally have a reduced rate of intercurrent mortality from nonspecific and infectious diseases during their early adult life, followed by a higher mortality rate later in life (NCRP 1980). Since these studies were conducted under conditions in which infectious diseases made a significant contribution to overall mortality, the interpretation of these studies with respect to radiation-induced cancer or other chronic diseases must be viewed with caution.
Experiments designed to address questions of low-dose risk using life shortening have used two different experimental approaches (NCRP 1980). One approach has been to deliver radiation doses at different dose rates over the entire life span of the animals. A second approach has been to develop dose-response relationships following acute, fractionated, and low-dose-rate exposures delivered as defined radiation doses. In such studies, a range of radiation doses have been delivered, generally to young adult animals. In the case of fractionated or low-dose-rate exposure regimens, the exposures were terminated at specific total doses delivered over a well-defined fraction of their life span.
For purposes of understanding risks from low-dose-rate exposures, it is important to make a clear distinction between dose-rate effects (which involve terminated exposures) and protraction effects (which involve radiation exposures over the entire life span). With few exceptions, dose-response relationships derived from life-shortening data following single acute radiation doses, fractionated exposures, and terminated low-dose-rate exposures all suggest linear dose-responses over wide range of doses (NCRP 1980). This apparent linearity in the dose-response for life shortening may reflect the integration of a variety of tumor types whose individual dose-responses may vary widely.
The exceptions are generally related to instances in which a single tumor type is the principal cause of death following radiation exposure. The primary effect of fractionating the radiation dose or reducing the rate at which the dose is delivered is to reduce the slope of the linear response.
Importantly, experiments using multiple, low-dose-rate, terminated exposures suggest a limiting linear slope in all cases (Storer and others 1979; NCRP 1980; Carnes and others 1989). Once this limiting linear response is reached, no further reduction in effect is seen if dose rate is reduced further. However, for protracted exposures that involve irradiation over the entire life span, a further reduction in life shortening per unit dose has been observed (NCRP 1980). This further reduction in slope has been attributed to so-called wasted radiation. According to this concept, radiation injury induced late in life does not have sufficient time to express itself, thereby reducing the slope of any dose-effect relationship.
In fact, both dose-rate effects and protraction effects are more complicated than they appear at first glance. Analysis of cause of death and tumor incidence data indicates that reducing the rate at which a radiation dose is delivered reduces the frequency of radiation-induced tumors and alters the spectrum of neoplastic disease (Storer and others 1979; NCRP 1980). First, the frequencies of early appearing radiation-induced neoplasms such as leukemia and lymphoma are reduced. This effect alone has a major impact on life shortening by switching the spectrum of disease to more late-occurring solid cancers. Second, a reduction in the frequency of late-appearing tumors when compared to animals receiving a single acute exposure is also observed. Depending on the exposure regimen, this effect on solid tumor frequencies may be a result of dose-rate effects in the case of terminated exposures, as well as a protraction effect in the case of lifetime exposures. This duality of effect tends to amplify dose-rate or protraction effects seen for individual tumors. Regardless of the fine structure of dose-rate and protraction effects, it is important to note that all of the data support a linear dose-response for radiation-induced life shortening at low doses and low dose rates over a wide range of doses.
Application of the linear-quadratic dose dependence, αD+βD2, and a wide range of molecular, cellular, and animal data have been used to argue that data on radiation-induced cancer in human populations derived from studies following acute radiation exposures tend to overestimate radiation risks at low doses and low dose rates. In this regard, analyses of the animal studies examining dose-response and dose-rate effects described above have been particularly important. In an attempt to quantify the degree to which extrapolation of acute high-dose data might overestimate risks at low doses and low dose rates, a number of groups have used a similar approach. The approach taken has been relatively simple. Essentially, the effectiveness per unit dose for acute exposures has been determined using a linear interpolation of data in the 2–3 Gy dose range and control data at 0 Gy. The rationale for using only the high-dose data and not data at lower doses was based on the assumption that this would simulate analyses of risks from epidemiologic studies where most of the available data were for single acute exposures at relatively high doses. Except in instances where threshold dose-responses were observed, effects per unit dose following low-dose-rate exposures were derived by calculating the slope of the entire dose-response (not just in the 2–3 Gy dose range).