Ionizing radiation: future etiologic research and preventive strategies.

Estimates of cancer risks following exposure to ionizing radiation traditionally have been based on the experience of populations exposed to substantial (and known) doses delivered over short periods of time. Examples include survivors of the atomic bombings at Hiroshima and Nagasaki, and persons treated with radiation for benign or malignant disease. Continued follow-up of these populations is important to determine the long-term effects of exposure in childhood, to characterize temporal patterns of excess risk for different types of cancer, and to understand better the interactions between radiation and other host and environmental factors. Most population exposure to radiation occurs at very low dose rates. For low linear energy transfer (LET) radiations, it often has been assumed that cancer risks per unit dose are lower following protracted exposure than following acute exposure. Studies of nuclear workers chronically exposed over a working lifetime provide data that can be used to test this hypothesis, and preliminary indications are that the risks per unit dose for most cancers other than leukemia are similar to those for acute exposure. However, these results are subject to considerable uncertainty, and further information on this question is needed. Residential radon is the major source of population exposure to high-LET radiation. Current estimates of the risk of lung cancer due to residential exposure to radon and radon daughters are based on the experience of miners exposed to much higher concentrations. Data indicate that lung cancer risk among miners is inversely associated with exposure rate, and also is influenced by the presence of other lung carcinogens such as arsenic in the mine environment. Further study of populations of radon-exposed miners would be informative, particularly those exposed at below-average levels. More direct evidence on the effects of residential exposure to radon also is desirable but might be difficult to come by, as risks associated with radon levels found in most homes might be too low to be quantified accurately in epidemiological studies.


Introduction
Epidemiological studies of populations exposed to ionizing radiation have provided considerable quantitative information concerning the risks of radiation-induced cancer (1-3)-more, perhaps, than is available for any other human carcinogen. Such data are invaluable for setting radiation protection policy and evaluating the late effects of medical exposures. Epidemiological data also complement experimental evidence as a basis for learning about mechanisms by which cancers develop. Yet, important PY, person-year; RR, relative risk; Sv, Sievert; UK, United Kingdom; UNSCEAR, United Nations Scientific Committee on the Effects of Atomic Radiation; WL, working level; WLM, working level month. unanswered questions remain regarding lifetime risks of radiation-induced cancer, the magnitude of risks from chronic lowdose-rate exposures, risk from low-level radon exposures in the home, determinants of host susceptibility, and molecular mechanisms of radiation carcinogenesis. The epidemiological data are strongest for populations exposed to high doses of radiation at high dose rates, but such exposures are uncommon in the general population. Implementation of effective preventive strategies requires better understanding of cancer and other health risks associated with low-level exposures. Full understanding of lifetime cancer risks attributable to irradiation requires long-term follow-up of exposed populations, including those irradiated at young ages.

Atomic Bomb Survivors
The single most important source of information about radiation-induced cancer in humans has been the Life Span Study (LSS) cohort of survivors of the atomic bomb explosions in Hiroshima and Nagasaki, for which the latest comprehensive reports document cancer incidence through 1987 (4,5). Continued follow-up of this study population is essential if we are to understand lifetime risks of exposure during childhood and adolescence. This is illustrated in Figure 1, which is for solid tumors occurring among males (5). Panel A shows the fitted excess relative risk (ERR) of solid tumors at a dose of 1 Sv as a function of time since exposure, separately for three different ages at exposure. For ages 30 and 50 years, the ERR was approximately constant with increasing time, but for age 10 at exposure, it decreased with time from initially high levels. Some observers have interpreted this decrease in relative risk as evidence that excess risk disappears with time among persons exposed during childhood. However, the pattern looks different when data for the same people are plotted on the absolute risk scale (Figure 1 C). Even for those age 10 at exposure, the absolute risk increased with time since exposure. Data in Figure 1D are expressed in terms of attained age rather than in years since exposure. The absolute excess risk at a given attained age ( Figure  1D) was almost identical for those who were age 10 at exposure and those age 30 at exposure. The age 10 at exposure group had only reached age 50 or so as of the last follow-up, and we do not know how patterns of risk will change as these people Environmental Health Perspectives  pass through ages of high cancer incidence. Studies of atomic bomb survivors now under way at the Radiation Effects Research Foundation in Japan should be continued so that this opportunity for lifelong follow-up is not missed.

Medically Irradiated Populations
It is important to continue to follow other irradiated groups in addition to those exposed at Hiroshima and Nagasaki. Studies of persons given radiotherapy for benign and malignant disease also have provided valuable information about cancer risks due to low-LET radiation. Ankylosing spondylitis patients treated with external-beam X-rays were c first such study populations to be (6), and follow-up recently was (7). Figure 2 shows relative risk time since irradiation for men wh years old at exposure. The solid sents lung cancer and the dotted solid tumors. Whereas the RR wa less constant with increasing ti exposure among the Japanese atoi survivors who were age 30 at exp RR for spondylitics decreased, for lung cancer but also to a cert for other solid tumors. A differei was seen for cancer of the bladdi ing radiotherapy with X-rays ft pathia hemorrhagica (Table condition that typically occui women between the ages of 4' These treatments delivered larg4 pelvic organs. The RR for bladc 7-l 70 80 increased with time following irradiation, but had this population been followed for only 15 to 20 years, this radiation effect might have been missed. Other pelvic organs displayed a different pattern of excess risk over time (Table 1) (8).
Results for different study populations complement each other, as they provide information about risks associated with different types of radiation exposure and possible modifying effects of host characteristics. Identical results are not necessarily to be expected among studies, and differences can provide important insights about mechanisms of induction of cancer by radiation. bomb survivors and persons given radiotherapy usually are of short-duration, occupational, environmental, and diagnostic medical exposures to ionizing radiation usually involve chronic or repeated expoatomic bomb sure to low doses. These are the types of )sure groups. exposure that account for most of the radiation exposure in the general population (9), but cancer risks associated with protracted, highly fractionated exposures are ne of the less well understood. described

Nuclear Workers
Because animal data indicate that s updated cancer risks from penetrating, low-LET c (RR) by radiation are lower when the dose is accu-Lo were 35 mulated over a prolonged period than for line repre-acute exposure, risk estimates derived from line other human populations exposed briefly at high S more or doses often have been divided by a doseime since rate effectiveness factor when applied to mic bomb populations receiving protracted exposures. osure, the Animal data have been interpreted to indiespecially cate that this factor might lie in the range 2 ain extent to 10, with greater support for the low end nt pattern of this range (2). We are just now begin-.er followning to have enough data available on or metro-nuclear workers with protracted radiation 1) (8), a exposures to test this hypothesis. Prelimrs among inary data do not support dose-rate factors 5 and 50. as large as 10 (   Nuclear workers (U.K. and U.S.) 1.7 (< 0, 5.9) Atomic bomb survivors (men exposed at ages > 20) 6.2 (2.7, 13.8) CLL, chronic lymphocytic leukemia. Data from UNSCEAR (3). workers and atomic bomb survivors, while leukemia risk estimates differ by a factor of about three to four (4,5,10,11). Again, these data are preliminary, and the estimates have wide confidence intervals. It is important to continue studies of nuclear workers and make further comparisons of this nature.

Radon
Another topic causing much public concern at the present time is the risk of lung cancer due to exposure to radon (222Rn) in the home. Radon and its short-lived daughter products are the largest source of radiation exposure to the general population (2). Until now, estimates of the risk of radon-induced lung cancer have been based on studies of underground miners, most of whom were exposed at relatively high levels. Extrapolation of these risk estimates down to the dose range more typical of residential exposures would suggest that there are, perhaps, 15,000 radon-induced lung cancer deaths per year in the United States (12). However, there are enormous uncertainties in those estimates.
One of the uncertainties concerns the effect of exposure rate. Table 3 summarizes results from six studies of underground miners listed in decreasing order of average exposure rate. The Port Radium (Northwest Territories, Canada) and Newfoundland miners worked in mines that, on average, had the highest ambient concentrations of radon, and the Malmberget (Sweden) miners and Beaverlodge (Saskatchewan) miners on average experienced the lowest concentrations of radon. There is a nearly monotonic inverse association between average exposure rate and estimates of the excess relative risk per cumulative workinglevel month (WLM). (A working level is defined as any combination of short-lived radon daughters in 1 liter of air that will result in the ultimate emission of 1.3 x 105 MeV of potential o energy, which approximately equals the x energy released from the decay of daughters in equilibrium with 100 picocuries of 222Ra (1). A workinglevel month is defined as the exposure resulting from inhalation of air with a concentration of 1 working level of radon daughters for 170 working hours) (1). Most of the public's exposure to radon is accrued at a very low exposure rates, lower than any of the rates for miners shown in Table 3. Failure to take account of a doserate effect could, therefore, result in a substantial underestimate of risk to the general population.
Detailed analyses of lung cancer mortality among 4320 uranium miners from West Bohemia revealed a 20-fold decrease in the excess relative risk per WLM with increasing average exposure rate (Table 4) (19). The men in this study worked in 19 different mine shafts, and radon concentrations varied widely, both from shaft to shaft and also within any particular shaft over time, as engineering changes were introduced to improve ventilation and reduce the radon concentration. Most men worked in a variety of mine shafts, and detailed data were available about the average radon concentration to which each man was exposed for every month he was employed in the mines. In trying to understand and model the exposure rate effect among these miners, it was discovered that it was entirely attributable to the small proportion of men who were ever employed in a mine shaft with a concentration of radon daughters of 10 working levels (WL) or more, which is an extremely high concentration.
When these men were omitted from the analysis, the association between RR and exposure rate completely disappeared, even though the average exposure was still quite high (Table 4). It would be interesting to determine whether a similar phenomenon occurs in other studies of radon-exposed miners. If it does, then a better data set for extrapolating to the general population might be obtained by omitting those men who at any time were exposed to a very high concentration.
Even if exposure rate effects can be sorted out, there are other difficulties associated with using radon risk estimates for miners to extrapolate to residential exposures. Among these is the possible role of other carcinogens in the mine environment. In a large study of Chinese tin miners exposed to both radon and arsenic, it was shown that the apparent risks of radon exposure were substantially reduced when adjustment was made for arsenic exposure (20). The role of arsenic and other carcinogens until now has not been studied in great detail among radon-exposed miners. However, Tomasek et al. (19) were also able to examine the question to a certain extent for the cohort of West Bohemian miners. There were no estimates of arsenic exposure for individual miners, but data were available on the proportion of arsenic in the dust at the two mines where the men were employed. At Jaichymov, which was the main mine, arsenic levels in the dust were considerable, while at Horni Slavkov, a subsidiary mine, they were negligible. Miners were grouped according to whether they spent more or less than 20% of their time at the Jaichymov mine. For equivalent cumulative radon exposures, the risk of lung cancer was discernibly greater among those who were employed at Jachymov, the arsenic mine, than at Horni Slavkov (Figure 3), both when the entire cohort was considered and when analysis was restricted to men who were only exposed at rates below 10 WL. Although this does not prove that arsenic played a role in the etiology of lung cancer in these men, it is consistent with what one would expect if it did and points to the need for more thorough exploration of the possible role of other exposures among radonexposed miners in the future. Venitt and Biggs (21) suggested that mycotoxins might also contribute to lung cancer excesses observed among miners. Other sources of uncertainty in generalizations from the experience of miners to the residential environment include risks of exposure to infants and children, and to females as well as males, and joint effects of radon exposure and smoking (22).
Because extrapolation of radon risks from exposed miners to the general population clearly is going to be difficult and uncertain however carefully it is done, it seems important to collect data bearing directly on the question of risks of residential radon. Such studies need be large, as radon exposures tend to be low. In a recently published Swedish study (23), investigators obtained an overall risk estimate that is very much in line with the predictions based on the miner data, but the confidence interval was wide (Table 5). Several other large case-control studies are in progress. It probably will be necessary to pool results from multiple studies to get a more definitive answer about risks from residential radon. Even then, however, it is questionable whether risks associated with radon levels encountered in most homes can be evaluated directly because it is difficult for epidemiological studies to distinguish low-level effects from bias (24). The task would be simplified if radon-induced tumors could be identified, such as through a characteristic mutation in a particular gene (25)(26)(27).

Chernobyl
The reactor accident at Chernobyl in April 1986 resulted in the release of large quantities of radioactive materials to the environment and was a catastrophe of enormous proportions in terms of disruptions in peoples' lives (28). The full extent of the health effects will be difficult to document for several reasons, including uncertainties about exposures for individuals and difficulties in ascertaining health outcomes in an unbiased manner. Nearly half the surveyed people from villages between 30 and 300 km from Chernobyl reported having had an illness that they attributed to radiation exposure, but early clinical and laboratory studies did not find evidence to corroborate these perceptions (28). With regard to cancer, particular attention should be paid to the risk of thyroid cancer among people exposed as children. There is concern that a large increase in thyroid cancer seen among children in Belarus might have been caused by exposure to radioactive iodines from the accident (29).

Summary
Future epidemiological studies of radiationexposed populations should aim to cover new ground and not simply revisit the question ofwhether radiation causes cancer, which has been answered in the affirmative for most though not all types of cancer (2). There is particular need for good quantitative information about risks from chronic low-dose-rate exposures, including radon in the home. An essential component of risk quantification is good dosimetry.
The public often overestimates risks of exposure to ionizing radiation relative to other common and more hazardous exposures. Residential, occupational, and diagnostic medical exposures are potentially controllable but only at a cost. While it is desirable to avoid unnecessary radiation exposures, it might be contrary to society's best interest to undertake extreme protective measures that would effect only a small reduction in risk.