Technical Report No. 5-88

Life Span Study Report 11. Part 2. Cancer mortality in the years 1950-85 based on the recently revised doses (DS86)

Shimizu Y, Kato H, Schull WJ

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: Studies of the mortality of A-bomb survivors. 9. Mortality, 1950-1985: Part 2. Cancer mortality based on the recently revised doses (DS86). Radiat Res 121:120-41, 1990
• Shimizu Y, Schull WJ, Kato H: Cancer risk among atomic bomb survivors. The RERF Life Span Study. JAMA 264:601-4, 1990
• Shimizu Y, Kato H, Schull WJ: A review of forty-five years study of Hiroshima and Nagasaki atomic bomb survivors. Mortality among atomic bomb survivors. J Radiat Res (Tokyo) 32S:212-30, 1991

Summary

ABCC and its successor, RERF, have followed since 1959 and retrospectively to 1950 the mortality in a fixed cohort of survivors of the atomic bombings of Hiroshima and Nagasaki, the so-called Life Span Study sample. The present study, the 11th in a series that began in 1961, extends the surveillance period three more years and covers the period 1950-1985. It is based on the recently revised dose system, called DS86, that has replaced previous estimates of individual exposures. The impact of the change from the old system of dosimetry, the T65DR, to the new on the dose-response relationships for cancer mortality was described in the first of this series of reports. Here, the focus is on cancer mortality among the 76,000 A-bomb survivors within the LSS sample for whom DS86 doses have been estimated, with the emphasis on biological issues associated with radiation carcinogenesis.

Briefly, the following was found: The excess in leukemia mortality has continued to decline with time, but remains slightly but significantly elevated in the most recent years of study, 1981-1985 in Hiroshima. For cancers other than leukemia, excess deaths continue to increase with time in proportion to the natural cancer mortality for the attained age, and the relative risk (RR) seems unchanged over time for specific age-at-the-time-of-the-bombing (ATB) cohorts except for the youngest cohort, i.e., 0-9 years ATB. For the latter cohort, unlike the older cohorts, the time from exposure to death (a measure of the latent period) is shortened for all cancers combined except leukemia in the high (1 Gy and over) dose group and the RR decreases with time under 30 years of age and then levels off. The present analysis still supports, in the main, the RR model rather than the absolute risk (AR) model in estimating lifelong risk.

For the same attained age, both the RR and AR are higher for younger age-ATB cohorts than for older ones.

Since the cohort is aging, the trend in estimated risk with time for the entire cohort has been examined, either with an adjustment for age-ATB effects or within specific age-ATB cohorts. The excess RR does not vary statistically over time to a significant degree for all cancers except leukemia, nor for cancers of the stomach, lung, and breast. However, at face value, a slightly decreasing trend with time is seen for cancer of the lung, and a slightly increasing trend is noted for all cancers except leukemia, and for cancers of the stomach and breast. Further surveillance will be necessary to clarify the pattern of temporal change in radiation-induced cancer.

Some sex differences in estimated risk are observed. As in the most recent previous report (1950-1982), there is no statistically significant difference in excess deaths between males and females except for leukemia, though the RR is higher for females than for males, significantly so for cancers of the esophagus and lung, reflecting the higher background cancer rate for males. An extensive lung cancer analysis, based on the cohort for the present analysis using both radiation and smoking information, revealed that the sex difference in excess RR observed without adjustment for smoking, is no longer significant after adjustment.

As previously observed, a statistically significant increase in the frequency of deaths with increasing dose is observed for leukemia, cancers of the esophagus, stomach, colon, lung, breast, ovary, urinary bladder, and multiple myeloma. No significant increase is demonstrable as yet for cancers of the rectum, gallbladder, pancreas, uterus, and prostate and malignant lymphoma. In the present report, the study of the frequency of deaths related to ionizing radiation has been extended to include cancers of the bone, pharynx, nose and larynx, and skin except melanoma, but none of these sites showed a significant increase with dose. Mortality tends to increase with dose for tumors of the central nervous system other than the brain (0.05 < p < 0.10), but not to increase for brain tumors.

The dose-response curve for all cancers except leukemia, the linear (L) model shows a good fit, although other nonlinear models (except the quadratic one) fit, statistically, the data now available as well as the L model. For leukemia, the L model fits well over the entire dose range, however, when the high dose range is excluded, the linear quadratic (LQ) model fits better than the L model.

Examination of the dose-response relationship at low doses (under 0.50 Gy), using much finer dose categories than have been employed previously, failed to reveal significant differences in the regression coefficients between the low and higher dose ranges except for leukemia, where the estimated risk coefficient for survivors exposed to under 0.5 Gy is lower than that for those exposed to 0.5 Gy and over.

The lifetime risk was estimated employing a method similar to that used in the BEIR III report. The estimated value, based on a L model, for both leukemia and all cancers except leukemia, is about two times higher than that estimated in the BEIR III report. The ratio of the present estimates to the BEIR III estimates under LQ model is much larger than the ratio of the two estimates under L model.

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 subjects in the LSS E85 and the DS86 samples
2. A. Summary measures of radiation dose response for cancer mortality by site; both cities, both sexes (unless otherwise stated), all ages ATB, 1950-1985 (shielded kerma)
   B. Summary measures of radiation dose response for cancer mortality by site, for certain specific cancers; both cities, both sexes, all ages ATB, 1950-1985 (shielded kerma)
3. Mortality from CNS tumors by type
4. Summary measures of radiation dose response for mortality of statistically significant site; both cities, both sexes, all ages ATB, 1950-1985 (organ-absorbed dose)
5. A. Relative risk at 1 Sv by age ATB and sex for certain site of cancer (organ-absorbed dose)
   B. Excess deaths (per 10⁴ Sv) by age ATB and sex for certain site of cancer (organ-absorbed dose)
6. Relative risk at 1 Gy by age ATB and age at death for various sites of cancer (shielded kerma)
7. Excess deaths (per 10⁴ PYGy) by age ATB and age at death (ATD) for various sites of cancer (shielded kerma)
8. Goodness of fit of the relative and absolute risk models, for all cancers except leukemia as measured by the deviance (shielded kerma)
9. Mean years (its standard error) since exposure to death by radiation dose and age ATB (shielded kerma)
10. Risk coefficients for age ATB under 10
11. Relative risk at 1 Gy observational period for various sites of cancer (shielded kerma)
12. Risk coefficients by sex (shielded kerma)
13. Radiation effect by sex for lung cancer (shielded kerma)
14. A. Difference in deviance between the linear (L) model and other nonlinear models; linear-quadratic (LQ) model, linear with cell-killing (L-K) model, linear-quadratic with cell killing (LQ-K) model, and between quadratic (Q) and LQ models for selected site of cancer
    B. Difference in deviance between the linear (L) model and linear-quadratic (LQ) model, and quadratic (Q) and LQ models by city for leukemia and all cancers except leukemia
15. Excess relative risk per gray in low-dose range
16. Estimated relative risk compared to the 0 Gy group
17. First appearance of a significant increase in mortality attributable to specific cancers
18. Estimated age- and sex-specific regression coefficient, RBE=10, dose range 0-6 Gy
19. Lifetime excess death from leukemia and all cancers except leukemia per million persons following a single exposure to 0.1 Sv

List of Figures

1. A. Relative risk at 1 Gy (shielded kerma) and 90% confidence interval, 1950-1985
   B. Relative risk at 1 Gy (shielded kerma) and 90% confidence interval for certain special cancers, 1950-1985
2. Cumulative mortality rate from all cancers except leukemia and 90% confidence interval by time since exposure and radiation dose (shielded kerma) for those exposed under age 10
3. Cumulative distribution of death from all cancers except leukemia by time since exposure, age ATB, and radiation dose (shielded kerma)
4. A. Variation in relative risk at 1 Gy (shielded kerma) of mortality by time interval and selected cancers; both sexes, all ages ATB
   B. Variation in relative risk at 1 Gy (shielded kerma) of mortality by time interval and selected cancers; both sexes, all ages ATB
5. Cumulative mortality rate from lung cancer by radiation (shielded kerma), smoking, and sex
6. Observed and fitted organ-absorbed dose-response curves for leukemia and for all cancers except leukemia
7. Shielded kerma and organ-absorbed dose-response curves for mortality from leukemia and all cancers except leukemia by city and dosimetry system