Projected lifetime cancer risks from exposure to regional radioactive fallout in the Marshall Islands

Abstract

Radioactive fallout from nuclear test detonations during 1946-1958 at Bikini and Enewetak Atolls in the Marshall Islands (MI) exposed populations living elsewhere in the MI archipelago. A comprehensive analysis, presented in seven companion papers, has produced estimates of tissue-specific radiation absorbed dose to MI residents at all historically inhabited atolls from internal (ingested) and external irradiation resulting from exposure to radioactive fallout, by calendar year, and by age of the population at time of exposure. The present report deals, for the first time, with the implications of these doses for cancer risk among exposed members of the MI population. Radiation doses differed by geographic location and year of birth, and radiation-related cancer risk depends upon age at exposure and age at observation for risk. Using dose-response models based on committee reports published by the National Research Council and the National Institutes of Health, we project that, during the lifetimes of members of the MI population potentially exposed to ionizing radiation from weapons test fallout deposited during the testing period (1948-1958) and from residual radioactive sources during the subsequent 12 y (1959-1970), perhaps 1.6% (with 90% uncertainty range 0.4% to 3.4%) of all cancers might be attributable to fallout-related radiation exposures. By sub-population, the projected proportion of cancers attributable to radiation from fallout from all nuclear tests conducted in the Marshall Islands is 55% (28% to 69%) among 82 persons exposed in 1954 on Rongelap and Ailinginae, 10% (2.4% to 22%) for 157 persons exposed on Utrik, and 2.2% (0.5% to 4.8%) and 0.8% (0.2% to 1.8%), respectively, for the much larger populations exposed in mid-latitude locations including Kwajalein and in southern locations including Majuro. By cancer type, point estimates of attributable risk varied, by location, between 12% and 95% for thyroid cancer, between 2% and 78% for leukemia, and between 0.8% and 55% for all cancers combined. The largest projected risks pertain to the Rongelap Island community and the lowest risks pertain to the populations resident on the southern-most atolls. While the projected cancer risks are smaller than those estimated by the National Cancer Institute in a more simplistic analysis conducted in 2004, these estimates of cancer risk are the best available as they are based on the most detailed dose reconstruction to date and comprehensively include populations at all locations and dose contributions from all nuclear tests.

Fig. 1

Fitted Bezier cubic splines expressing yearly interpolated, gender-specific population numbers for the entire Marshall Islands population, based on census results for 1935, 1958, 1967, and 1973.

Fig. 2

Estimated population sizes over time for Majuro, 13 other southern atolls, Kwajalein, 6 other mid-latitude atolls, and the Rongelap and Utrik communities including members not present at the time of the Castle BRAVO test on March 1, 1954.

Fig. 3

Estimated cumulative thyroid doses for different communities, by year of birth, drawn from Table 6 in Simon et al., 2009b. *Rongelap community members exposed to BRAVO fallout on Rongelap Island and on Ailinginae on March 1, 1954. ** Dose estimates for persons born in 1931 also pertain to persons born earlier.

Fig. 4

IREP uncertainty distributions for the dose and dose rate effectiveness factors (DDREF) to be applied at low doses and low dose rates to risk estimates for (a) breast and thyroid cancer and (b) solid cancers other than breast and thyroid.

Fig. 5

Results of a Monte Carlo simulation to evaluate the effects of adjusting an uncertain thyroid cancer excess relative risk (ERR) projection distributed as lognormal with geometric mean (GM) = 0.1023 and geometric standard deviation (GSD) = 3.625, by the DDREF with uncertainty distribution shown in Figure 4, panel 1. The simulated uncertainty distribution is approximately lognormal with GM = 0.0688 and GSD = 3.92.

Fig. 6

U.S. male and female life tables used in the preparation of this report. Drawn from U.S. Decennial Life Tables for 1989–1991, Vol. 1, Number 1, United States Life Tables. Hyattsville, MD. U.S. Dept. of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics (PHS) 97-110-1; 1997.

Fig. A1

(Appendix). Sigmoid (S-shaped) function representing the multiplicative adjustment factor (and its uncertainty) applied to the risk of stomach cancer, colon cancer and all solid cancers (less thyroid and non-melanoma) as group, due to the effect of minimum latency period at early times since exposure.