Low-level Radiation Harmed Humans Near Three Mile Island
Wing et al. (
1
) analyze data from the area nearest the Three Mile Island nuclear installation, showing elevated cancer incidence rates 5 years after the 1979 accident. Wing et al. refute earlier assumptions that low-level radioactive emissions from the accident were too minute to produce observable effects (
2
). The critical analysis by Wing et al. (
1
) shows excess cases for all cancers combined, lung cancers, and leukemia.
Figure 1
. Correlation between the ratio of observed to expected lung cancer cases (1975-1979 versus 1981-1985) and the log values of the dose at Three Mile Island.
r
= 0.88.
|
Leukemia rates rise with increasing doses, but the small number of cases hamper the significance of the findings. Lung cancer rates also rise, and the larger number of observations make the findings significant. The ratio of observed to expected (O/E) lung cancer cases is three or four times higher at the greatest exposures (still considered low-level) than at the lowest exposures, within 5 years after the accident.
An examination of Wing's data yields a specific finding about effects at low doses. Plotting the O/E ratios by level of exposure shows that the dose response does not conform to a linear model. Rather, a logarithmic or supralinear curve describes the relationship more accurately, as the greatest per-dose effects occur at the very lowest levels of exposure. Perhaps this logarithmic relationship is best seen in lung cancer, where the ratios of the observed to the expected number of cases show a significant rise from 1975 through 1979 to 1981 through 1985. In order to focus on the more rapid rise at low levels of exposure, I demonstrate in Figure 1 the significant correlation between the O/E ratios and the log values of the mean dose, omitting the observations involving the zero mean dose exposure and the highest exposure, the inclusion of which distorts the logarithmic relationship.
The supralinear dose-response scenario carries serious repercussions, not just for accidents like Three Mile Island but for all types of low-level radiation exposure to nuclear workers and the general population. The model has a basis, both in theory and observation. Canadian physician and biophysicist Abram Petkau demonstrated that it took smaller-than-expected amounts of radioactive
22
Na added to water to break cell membranes extracted from fresh beef brain (
3
). The supralinear dose response has been illustrated in the medical literature; most recently, such a relationship linking changes in U.S. newborn hypothyroid rates to Chernobyl fallout was reported (
4
).
Future research should not assume that there is a safe threshold for radiation exposure in humans. Because evidence that low levels of radiation exposure can be harmful is surfacing, it is imperative that studies of the biological mechanisms of low-level radiation begin immediately.
Joseph J. Mangano
Consultant
Radiation and Public Health Project
Brooklyn, New York
References
1. Wing S, Richardson D, Armstrong D, Crawford-Brown D. A reevaluation of cancer incidence near the Three Mile Island nuclear plant: the collision of evidence and assumptions. Environ Health Perspect 105:52-57 (1997).
2. Hatch MC, Wallenstein S, Beyea J, Nieves JW, Susser M. Cancer rates after the Three Mile Island nuclear accident and proximity of residence to the plant. Am J Public Health 81:719-724 (1991).
3. Petkau A. Effects of
22
Na
+
on a phospholipid membrane. Health Phys 22:239-244 (1972).
4. Mangano JJ. Chernobyl and hypothyroidism [letter]. Lancet 348:476-477 (1996).
Response
Mangano correctly notes that the dose- response relationships in our analyses of cancer incidence in relation to the Three Mile Island (TMI) accident (
1
) are supralinear. The goodness of fit statistics would have been larger (and corresponding
p
-values smaller) had we fit regression models using the log of dose.
The substantive issue raised by Mangano, however, concerns mechanisms that could account for a larger carcinogenic effect of radiation per dose unit at low levels than at higher levels. There are three important questions regarding Mangano's interpretation: 1) Do the cancer incidence patterns reflect low dose radiation? 2) Is the study design appropriate for distinguishing the shape of radiation dose-response relationships at low levels? and 3) Is the original scaling (vs. magnitude) of dose estimates correct? We answer no to the first two questions and discussed reasons for uncertainty regarding the third in our paper (1).
Figure 1
. Three Mile Island postaccident lung cancer rates for 1981-1985 (adjusted for age, sex, and preaccident incidence) in relation to the estimated distribution of radioactive emissions from the accident. Reprinted with permission from Endeavors (5).
Credit: Julia Bryan
We noted that patterns of cancer incidence were consistent with reports suggesting high doses of radiation. The appearance of large elevations in lung cancer incidence within 7 years of exposure is not consistent with previous studies of low-level radiation, where low might be defined as doses below the annual occupational limit of 50 mSv. Lung cancer elevations seen at TMI are much larger and appear with shorter latency than do elevations for workers with cumulative lung doses that are substantially above 50 mSv
(2-4).
Furthermore, the design of the TMI study is not well suited to distinguishing between shapes of dose-response associations. Such questions would be better addressed using epidemiological designs that include radiation dose measurements for individuals rather than estimates of average doses for geographic areas, within which there is heterogeneity of individual dose. Individual dose estimates should ideally include evaluation of both external penetrating radiation and internal exposures to ß- and *-emitting radionuclides, where relevant.
Subsequent to publication of our study, the lung cancer results reanalyzed by Mangano have been mapped for an article in the University of North Carolina's
Endeavors
magazine (5). Figure 1, which links the TMI postaccident lung cancer rates for 1981-1985 (adjusted for age, sex, and preaccident incidence) to a map of the estimated distribution of radioactive emissions from the accident, demonstrates Mangano's point that the dose response would be fairly linear on a log scale (note that the vertical axis is not on a log scale). Figure 1 also shows the large magnitude of the elevation when comparing the lowest and highest dose study areas. Given the uncertainty of dose estimates, heterogeneity within study blocks, and other limitations of study design, we caution against overinterpreting these findings in terms of low-level radiation's biological mechanisms.
Steve Wing
David Richardson
University of North Carolina School of Public Health
Chapel Hill, North Carolina
Donna Armstrong
State University of New York School of Public Health
Albany, New York
References
1. Wing S, Richardson D, Armstrong D, Crawford-Brown D. A reevaluation of cancer incidence near the Three Mile Island nuclear plant: the collision of evidence and assumptions. Environ Health Perspect 105:52-57 (1997).
2. National Research Council Committee on the Biological Effects of Ionizing Radiation (BIER IV). Health Risks of Radon and other Internally Deposited Alpha-emitters. Washington, DC:National Academy Press, 1988.
3. Hornung RW, Meinhardt TJ. Quantitative risk assessment of lung cancer in U.S. uranium miners. Health Phys 52:417-430 (1987).
4. Dupree EA, Watkins JP, Ingle JN, Wallace PW, West CM, Tankersley WG. Uranium dust exposure and lung cancer risk in four uranium processing operations. Epidemiology 6:370-375 (1995).
5. Dalrymple M. Science on the firing line. Endeavors 14(1):12-13 (1997).
Last Update: September 23, 1997