Technical Report No. 12-87

Life Span Study Report 11. Part 1. Comparison of risk coefficients for site-specific cancer mortality based on the DS86 and T65DR shielded kerma and organ doses

Shimizu Y, Kato H, Schull WJ, Preston DL, Fujita S, Pierce DA

Editor’s note:  

The following journal articles, based on this ABCC technical report, were published in the scientific literature: 
Shimizu Y, Kato H, Schull WJ, Preston DL, Fujita S, Pierce DA: Studies of the mortality of A-bomb survivors. 9. Mortality, 1950-1985: Part 1. Comparison of risk coefficients for site-specific cancer mortality based on the DS86 and T65DR shielded kerma and organ doses. Radiat Res 118:502-24, 1989

Summary

In March 1986, as a result of a comprehensive reevaluation of the exposures of the atomic bomb survivors of Hiroshima and Nagasaki, a new method for the estimation of individual doses was introduced, termed the Dosimetry System 1986 (DS86). Briefly, the differences between this system and the old (T65DR) can be summarized as follows: 1) The DS86 free-in-air gamma dose increases somewhat in Hiroshima, but decreases in Nagasaki in comparison with the T65DR; whereas the neutron dose decreases in both cities, to about 10% of its former value in Hiroshima and 30% in Nagasaki. 2) The transmission factor for gamma rays in wooden Japanese structures is smaller, around 51% and 59% of the T65DR value, on average, in Hiroshima and Nagasaki, respectively. 3) As a consequence, the average DS86 total shielded kerma (the sum of the individual shielded gamma and neutron doses) for those survivors exposed to 10 mGy and over decreases to 69% and 76% of the T65DR values in Hiroshima and Nagasaki, respectively.

The present analysis embraces a total of 75,991 persons, hereafter termed the DS86 subcohort, including the distally exposed (59,784 individuals) to whom doses are assigned in most instances and 16,207 among a total of 19,387 proximally exposed subjects (i.e., survivors within 1,600 m in Hiroshima and 2,000 m in Nagasaki, based on T65DR distances), mostly individuals within Japanese houses or tenements, on whom DS86 doses can be directly calculated.

Emphasis here is on cancer site-specific comparisons of the DS86 cancer risk estimates and those obtained by using the T65DR on the same group of survivors (the DS86 subcohort). The rationale for thus restricting the comparison is to assess, as directly as possible, the effect of changes in the risk estimates, unconfounded by a change in the cohort definition. In another report, the primary aim was to compare, but only for broad classes of cancer, the conclusions which have been drawn from the T65DR dosimetry to those that now seem appropriate. Comparison was thus made to the full T65DR cohort; substantial emphasis in this report was placed on the effects of assumptions about relative biological effectiveness (RBE) and possible nonlinearities in the dose-response. As a result of all of these factors, the changes due to the new dosimetry described there are somewhat greater than those indicated here.

The risk coefficients for cancer mortality for the period 1950-1985 have been compared, using 10 sites or groups of sites of cancer where the increase in mortality is statistically significant, with both the DS86 and T65DR doses, viz, leukemia, all cancers except leukemia, cancer of the stomach, colon, lung, female breast, esophagus, ovary, bladder, and multiple myeloma. Three measures of risk–the excess relative risk (per gray), absolute risk (excess cancer deaths per 10⁴ person-year-gray [PYGy]), and the attributable risk (%)–have been estimated. In general, in terms of shielded kerma, these measures increase as expected given that the DS86 doses decrease in contrast to the T65DR values. The excess relative risk over the various sites ranges from 1.35 to 1.51-fold higher and the excess numbers of cancer deaths per 10⁴ PYGy are increased from 1.38 to 1.61-fold under the DS86 system. The attributable risks do not differ significantly between the two dosimetric systems (the ratio of the DS86 and T65DR estimates varies from 0.95 to 1.08).

Under the DS86 system, the computed average transmission factors for specific organs increase generally but those for house shielding decrease as compared with the T65DR. Accordingly, in terms of organ-absorbed dose (gamma and neutron), assuming a linear dose-response model, the difference in risk coefficients between the two dose systems is smaller than that for shielded kerma; the changes in the transmission factors for house shielding and organ tissue tend to nullify one another. The excess numbers of cancer deaths per 10⁴ PYGy with the DS86 organ-absorbed doses (and ratio to T65DR) are 2.94 (0.95), 10.13 (0.73), 2.42 (0.72), 0.81 (0.83), 1.68 (0.89), 1.20 (1.33), 0.45 (0.92), 0.71 (1.11), 0.66 (0.81), and 0.26 (0.90) for leukemia, all cancers except leukemia, cancer of the stomach, colon, lung, female breast, esophagus, ovary, bladder, and multiple myeloma, respectively. It should be noted that the risk coefficients based on the DS86 organ-absorbed doses are generally smaller than those based on the T65DR values. However, the risk coefficient for cancer of the female breast increases with the DS86 even at the organ-absorbed dose level. Parenthetically, it should be noted that in the DS86 system posture and orientation are taken into account; this was not true in the T65DR.

The difference in cancer mortality between Hiroshima and Nagasaki is smaller with the DS86 than with the T65DR and no longer statistically significant for either shielded kerma or organ-absorbed dose. The magnitude of the effects of such modifiers of radiation-induced cancer as age at exposure, sex, and time since exposure does not change between the two dose systems for either shielded kerma or organ-absorbed dose.

Since the DS86 neutron dose is very small, even in Hiroshima, attempts to measure the separate effects of neutrons through an analysis of the dose-response using the gamma and neutron doses separately or to estimate their RBE do not give meaningful results. The data prove compatible with a wide range of plausible RBE values.

With the DS86, the excess numbers of cancer deaths per 10⁴ person-year-sievert (PYSv) attributable to gamma rays for an assumed constant neutron RBE of 1, 10, and 20 in organ dose equivalent, adjusted for the effects of age and sex, are 2.95, 2.67, 2.40 for leukemia, 10.1, 9.41, 8.76 for all cancers except leukemia, 2.63, 2.36, 2.10 for stomach cancer, 0.76, 0.73, 0.69 for colon cancer, 1.80, 1.59, 1.42 for lung cancer, and 1.22, 1.00, 0.82 for female breast cancer. Thus, the risk associated with gamma rays does not differ materially with different assumed values of the neutron RBE. With the T65DR doses, the estimated excess deaths per 10⁴ PYSv is much more sensitive to the RBE value that is assumed, and the disparity between the two dosimetries grows larger as the assumed RBE increases, reflecting the relative importance of the neutron component in the two systems. At an RBE of 10, for the five specific cancers, i.e., female breast, colon, leukemia, lung, and stomach, the increase in excess number of deaths per 10⁴ PYSv under the DS86 varies from 12% (colon) to 133% (female breast).

Briefly, this reevaluation of the exposures does not change the list of radiation-related cancers. Some city differences in dose-response previously thought to be real when the T65DR doses were used, such as leukemia, are no longer significant with the DS86 doses. Assuming a linear dose-response, and using estimated organ-absorbed doses, the risk coefficients derived from the two dosimetries are very similar, whereas those based on shielded kerma are about 40% higher with the new dosimetry. If larger RBE values are assumed, the disparity between the two dosimetries increases because the neutron dose is much greater in the revised T65 dosimetry.

Editor’s note: 

The following components of this report contain data on communicable disease frequencies, allergies, malignancies, and many other symptoms that may be of interest from a public health standpoint.

List of Tables

1. Number of study subjects by exposure status, cohort, and city
2. Number of subjects, person-years, and cancer deaths by DS86 shielded kerma
3. Mean shielded kerma and organ-absorbed dose (mGy) among survivors exposed to 0.01 Gy and over
4. Distribution of proximally exposed survivors by T65DR dose estimation method and city
5. Excess relative risk per 1 Gy (DS86 shielded kerma) by T65DR dose estimation method–DS86 subcohort
6. Comparison between DS86 and T65DR in summary measures of radiation dose-response for mortality based on shielded kerma–DS86 subcohort [both cities, sexes (unless otherwise noted), all ages ATB combined]
7. Comparison between DS86 and T65DR in summary measures of radiation dose-response for mortality based on organ-absorbed dose–DS86 subcohort [both cities, sexes (unless otherwise noted), all ages ATB combined]
8. Comparison of radiation effect modification based on DS86 and T65DR estimates–Excess relative risk per gray
9. Comparison of the deviance among various dose-response models based on total organ-absorbed dose
10. Comparison of the deviance among various dose-response models based on organ-absorbed gamma rays and neutron doses
11. Ratio of the Hiroshima to Nagasaki risk for fixed RBE values–Organ dose equivalent
12. Comparison of excess death per 10⁴ person-year-Sv for selected RBE values using the DS86 and T65DR doses
13. Proportion of subjects with T65DR shielded kerma dose of 6 Gy and over by DS86 shielded kerma

List of Figures

1. Free-in-air kerma by ground distance, city, and dosimetry system for gamma rays and neutrons
2. Shielded kerma by ground distance, city, and dosimetry system for gamma rays and neutrons
3. Shielded kerma and organ-absorbed dose-response curves for mortality from leukemia by city and dosimetry system
4. Shielded kerma and organ-absorbed dose-response curves for mortality from all cancer except leukemia by city and dosimetry system
5. Shielded kerma and organ-absorbed dose-response curves for mortality from stomach cancer by city and dosimetry system
6. Shielded kerma and organ-absorbed dose-response curves for mortality from colon cancer by city and dosimetry system
7. Shielded kerma and organ-absorbed dose-response curves for mortality from lung cancer by city and dosimetry system
8. Shielded kerma and organ-absorbed dose-response curves for mortality from female breast cancer by city and dosimetry system

 

List of Appendix Tables

1. Average house transmission factors by dose system and city
2. Comparison of T65DR and DS86 shielded kerma dose estimates by city
3. Average organ dose transmission factors by dose system
4. Comparison of T65DR and DS86 organ-absorbed dose estimates by city–Bone marrow
5. Ditto–Large intestine
6. Ditto–Lung
7. Ditto–Stomach
8. Ditto–Female breast
9. Ditto–Bladder
10. Ditto–Ovary
11. Summary measures of radiation dose-response for mortality: Both cities, both sexes (unless otherwise stated), all ages ATB combined, 1950-1985 (DS86-subcohort, DS86 shielded kerma)
12. Ditto (DS86-subcohort, T65DR shielded kerma)
13. Ditto (Full-cohort, T65DR shielded kerma)
14. Comparison of the deviance among various dose-response models based on total dose–Shielded kerma
15. Comparison of the deviance among various dose-response models based on gamma rays and neutrons–Shielded kerma
16. Ratio of the Hiroshima to Nagasaki risk by fixed RBE values–Shielded kerma
17. Comparison of risk coefficients between models with or without age, sex adjustment–Organ-absorbed dose