For individuals exposed to radiation in middle age, standard models and evaluations of radiological cancer risk (13
) have suggested the patterns of risk observed for exposures in childhood and young adulthood, in which the excess cancer risks decrease with increasing age at exposure, are maintained for exposure in middle age. However, as discussed below, the weight of epidemiological evidence now suggests that, for adult exposures, radiation-induced cancer risks do not generally decrease with increasing age at exposure. Here we have investigated whether this pattern of radiation-induced cancer risks that do not decrease monotonically with increasing age at exposure, as observed in the atomic bomb survivors (see ), is biologically plausible. We have shown that a model of radiation-induced cancer that uses realistic parameters and includes both initiation and promotion components (23
) can reproduce the observed dependencies of radiation-induced cancer risks as a function of age at exposure, both for individual cancers as well as for all cancers combined. This is because initiation, which dominates at younger ages, results in risks that decrease with increasing age at exposure, whereas promotion, which dominates at older exposure ages, does not (). Thus, different balances between initiation and promotion (the ratio of parameters X
in ) will produce different dependencies of radiation risk as a function of age at exposure.
We conclude that the observed patterns of radiation-induced cancer risks as a function of age of exposure are not consistent with standard models of radiation carcinogenesis in which radiation solely initiates premalignant cells but are consistent with models of radiation carcinogenesis that include both radiation-induced initiation and promotion. This conclusion is conceptually important because many commonly used biologically based models of radiation-induced carcinogenesis, such as various derivatives of the original Armitage–Doll model (24
), describe only the initiating component of radiation carcinogenesis, and do not describe potential radiation-induced promotional effects. More recently, however, several investigators have also published mechanistically based models of radiation-induced carcinogenesis that consider, in effect, both radiation-induced initiation and promotion (3
Our second goal was to use an initiation- and promotion-based model, with radiation parameters estimated from fitting the atomic bomb survivor ERR data, to generate absolute lifetime radiation-induced cancer risks per unit dose in the US population, as a function of age at exposure. The results, summarized in , suggest that the radiation-related cancer risk for an exposure at, for example, age 50 years, could be twice as high as estimated using standard models in which radiation-induced cancer risks are constrained such that the ERRs cannot increase with increasing age at exposure.
Practically speaking, there could be considerable societal consequences if the excess lifetime cancer risks for radiation exposure in middle age are somewhat higher than previously estimated, for example, in the recent BEIR-VII (14
) or International Commission on Radiological Protection (13
) reports. The majority of the radiation exposures in the population, both medical and occupational radiation exposures, occur in individuals who are older than 30 years (41
). The relevant regulatory occupational exposure limits for ionizing radiation are derived almost entirely from analyses of atomic bomb survivors who were exposed in adulthood (13
). Thus, an increase in the best estimate of the excess lifetime cancer risk after radiation exposure in middle age might be reflected in a corresponding change in occupational radiation exposure limits. The practical implications of such a change would probably not be wide ranging because it is unusual for radiation workers to be exposed to doses close to the regulatory limits; however, there could be implications for those activities in which individuals are occasionally exposed to doses near these limits, such as for staff in interventional radiology facilities (45
) or for some emergency responder scenarios (46
A far more common source of radiation exposure in middle age is from diagnostic radiology (41
). The medically related component of the US population exposure to ionizing radiation has increased sixfold in the past three decades (47
), mostly because of the rapid increase in computerized tomography (CT) imaging (48
). The most common ages at which individuals undergo CT examinations are approximately 35 to 50 years (49
). When a CT scan is medically warranted, its benefits far outweigh any radiation risks, so that even increasing the estimated risks by a factor of 2 would not materially affect the risk–benefit balance (48
). However, the risk–benefit balance is potentially relevant for CT-based screening of asymptomatic “healthy” adults. Specifically, routine CT screening of the colon (50
), lung (52
), and heart (56
) are increasingly being advocated. Lung and cardiac CT screening are of particular relevance here because in both cases, the most important organ in terms of radiation risk is the lung, and we have shown here that there is good evidence ( and ) that the excess lifetime risks of lung cancer do not decrease in middle age and indeed may peak at around age 50 years, the most likely age for individuals to undergo CT screening.
This study has several limitations, the primary one being the statistical uncertainties associated with the underlying data (5
) from the atomic bomb survivors (see error bars in ). As can be seen in , because the data are stratified by age at exposure, the individual confidence intervals are quite wide, particularly when further stratification by cancer site is made. Little (6
) has shown that the data for all solid cancer cancers combined, as shown in , are inconsistent with an ERR that decreases monotonically with increasing age at exposure. Although data for cancer mortality are not analyzed here, similar empiric conclusions have been reached for the variation in radiation-induced cancer mortality with age at radiation exposure (6
). The corresponding atomic bomb survivor data for individual cancer sites () did not reach statistical significance, which is not surprising given the decreased statistical power (6
) but, as shown in , the same trends, for adult exposure (radiation risks not monotonically decreasing with increasing age at adult exposure), were consistently seen for liver, colon, lung, stomach, and bladder, but not for breast cancer. A recently detailed analysis of combined radiation and smoking effects among atomic bomb survivors (59
) strongly suggests that the observed increase in ERR with age at exposure for lung cancer is present irrespective of smoking status. This same pattern of increasing ERR with increasing age at adult exposure was seen in a large study of cancer risks in more than 400
000 radiation workers in the nuclear industry (30
), where a lower ERR for all solid cancers was observed in workers exposed at ages 20–35 years compared with workers exposed at older ages (P
= .09). This pattern of increasing ERR with increasing age at adult exposure was not, however, seen in a smaller study of cancer mortality in 20
000 radiation workers at the Russian Mayak nuclear complex (60
), although a study of approximately 30
000 individuals exposed to protracted radioactive contamination from the Mayak complex (61
) showed increased (P
= .08) cancer mortality per unit dose with increasing age at first exposure. Overall, the weight of the epidemiological evidence suggests that for adult exposures, radiation risks do not generally decrease with increasing age at exposure, and the mechanistic underpinning described here provides this conclusion with some biological plausibility.
Another limitation is that although we have hypothesized that the risk patterns modeled here reflect the influence of promotion as well as initiation, other interpretations are possible. For example, the data may be consistent with an abrupt age-dependent increase in smoking and/or drinking patterns among survivors after the atomic bomb explosions; however, data for Japan as a whole do not show age-specific changes in smoking patterns in the immediate post–World War II period (62
). There are also possible alternative biological interpretations of the data. For example, organ-specific stem cells might have greater sensitivity to radiogenic initiation and/or promotion in older individuals than in younger individuals. Alternatively, an increased cancer risk in middle age may be due to stimulation of tumor progression by radiation, for example, by activation of microscopic dormant tumors, which may accumulate in older individuals (as opposed to promotion of premalignant cells, as discussed here) (63
). We do not, however, have any direct evidence for these alternative explanations, and alternative biological explanations consistent with the data shown in would not substantively affect the absolute age-dependent risks shown in , nor would they change the societal implications of the increased radiation risks for adult exposure.