A special issue of the journal Radiation Research, published in February, is devoted to RERF analyses of cancer incidence, a complement to RERF’s periodic Life Span Study reports that focus on mortality among the atomic-bomb survivors.

by Kiyohiko Mabuchi, Department of Epidemiology, RERF, and Dale Preston, Department of Statistics, RERF

This article was originally published in RERF Update 6(1):3-4, 1994. Complete bibliographic information for all four papers published in this special issue are listed at the end of this article.

---

The RERF Life Span Study (LSS), comprising 93,000 atomic-bomb (A-bomb) survivors and 27,000 unexposed persons, is a major source of epidemiological data for cancer-risk assessment. Periodic analyses of LSS mortality data have been published by RERF, beginning in 1961. However, the few reports addressing cancer incidence have almost exclusively dealt with selected individual cancer sites. Four RERF reports published in a special issue of Radiation Research in February provide the first comprehensive overview of cancer incidence in the LSS. What follows is a brief summary of this four-part special issue.

Use of the tumor registries

Part I, by K Mabuchi et al, describes the methodological aspects of the Hiroshima and Nagasaki tumor registries and addresses data-quality issues relevant to the incidence studies, in the context of the LSS cohort. Population-based tumor registries, established in 1958 in Hiroshima and Nagasaki, are characterized by active case ascertainment based on abstracting medical records at local hospitals. Efforts to improve the quality and usefulness of this cancer-incidence data already have been described by K Mabuchi and M Soda in RERF Update 2(2):5-6, 1990.

The quality of the Hiroshima and Nagasaki registries ranks among the best in Japan and is comparable to registries established in other countries, because of a death-certificate-only rate of less than 9% and a histological verification rate greater than 70%. LSS cancer cases identified from the registries were reviewed and processed using standardized procedures to ensure complete ascertainment, data quality, and consistency. Special studies and monitoring programs were introduced. The analyses performed indicated the uniformity of the data across various strata, justifying the use of incidence data for cancer risk assessment in the LSS.

 

Solid tumors

Solid-cancer-incidence data are presented in Part II, by D Thompson et al. Included in the analyses are 8613 first occurrences of primary solid tumors diagnosed among 79,972 individuals during the period 1958-87. A standard set of analyses based on a general excess-relative-risk (ERR) model was performed for all cancer sites as well as for each of 21 organs or systems, using Dosimetry System 1986 (DS86) organ doses. Analyses performed for each cancer site involved fitting of a set of models: the background model with no dose effect, a linear-dose-response model with no effect modifiers, a linear-quadratic dose-response model with no effect modifiers, and a series of linear-dose-response models that included each of the covariates (sex, age at exposure, time since exposure, attained age, and city) individually as effect modifiers. Because the tumor registries ascertain cancers in the registry catchment areas only, an adjustment was made for the effect of migration (RERF Update 3[3]:10, 1991).

Part II begins with the analysis of all solid-tumor-incidence data. For all cancer sites, a strong linear dose response was found, with no significant difference between the responses for Hiroshima and Nagasaki. The ERR is about twice as high for females as for males and decreases with increasing age at radiation exposure, as previously found in the mortality studies. ERR for all solid tumors decreased with time for the groups exposed when young but remained virtually constant for the cohorts exposed when older; averaged over all ages at exposure, ERR decreased with time since exposure.

Part II also describes the results of cancer-site-specific analyses. In agreement with the previous findings, a statistically significant excess risk was found for cancers of the stomach, colon, lung, breast, ovary, urinary bladder, and thyroid. An effect of A-bomb radiation on salivary-gland tumors was also observed, strengthening earlier incidence findings. For the first time in the LSS cohort, radiation has been associated with liver cancer and nonmelanoma skin cancer. No significant dose response was found for cancers of the oral cavity and pharynx, esophagus, rectum, gallbladder, pancreas, larynx, uterine corpus and cervix, prostate, kidney and renal pelvis. No city differences were found for any of the cancer sites examined; females had a significantly higher ERR compared with males, for cancers of the respiratory system, including the lung, and cancers of the urinary system. The ERR decreased with increasing age at exposure for the following cancer sites: salivary glands, stomach, nonmelanoma skin, breast, and thyroid.

 

Leukemia, lymphoma, and multiple myeloma

Leukemia incidence has been of interest since the earliest studies of A-bomb survivors, and the analysis of the incidence data on leukemia as well as on cancers of the lymphoid tissues is presented in Part III, by D Preston et al. This analysis adds 9 additional years of follow-up for leukemia and 12 for multiple myeloma to the last comprehensive incidence reports on these cancers, and it is the first analysis of lymphoma incidence in the LSS cohort. Unlike Part II, which is based solely on solid-tumor data from the Hiroshima and Nagasaki tumor registries, Part III also includes incident cases ascertained from the leukemia registry. The availability of the leukemia-registry data made it possible to extend the study period back to 1950. The total numbers of cases diagnosed during the present study period (1950-87) are 290 for leukemia, 229 for lymphoma, and 73 for multiple myeloma, with analyses restricted to the first occurrences of primary tumors among the residents of Hiroshima and Nagasaki assigned DS86 dose estimates between 0 Gy and 4 Gy.

Time-dependent models for the excess absolute risk were applied for all the studied diseases and for specific leukemia types, including acute lymphocytic leukemia (ALL), acute myelogenous leukemia (AML), chronic myelocytic leukemia (CML), and adult T-cell leukemia (ATL), which is endemic in Nagasaki. Too few cases of chronic lymphocytic leukemia were found to allow meaningful statistical analysis. The report emphasizes the importance of considering age, time, and sex, as well as specific leukemia types, in estimating risk. Strong evidence exists of radiation-induced risks for all types of leukemia except ATL, but significant leukemia-type differences in dose response and patterns of risk by age at exposure, sex, and time since exposure have been found. The AML dose-response function was nonlinear, whereas no evidence against nonlinearity was found for the other types of leukemia. The results suggest that men and women differ significantly with regard to the level and temporal patterns of the excess risk. The present analysis shows no evidence of an excess risk of multiple myeloma, whereas weak evidence was found of an increased risk of non-Hodgkin’s lymphoma in men.

 

Comparing incidence and mortality data

The significance of considering incidence data in risk assessment is made clear by comparing the major differences between incidence and mortality findings, highlighted in Part IV, by E Ron et al. The incidence data derived from the Hiroshima and Nagasaki tumor registries are limited to the period after 1958 and the catchment area of the registries, whereas mortality data are available through nationwide family registries, dating from 1950 onward. During the period 1958-87 in Hiroshima and Nagasaki, 9014 first-primary-incident tumor cases (including hematopoietic cancers) were identified among the LSS cohort members, compared with 7308 deaths nationwide attributable to cancer for the period of 1950-87 and 5859 cancer deaths for 1958-87 in Hiroshima and Nagasaki. For certain cancer sites, such as the oral cavity and pharynx, skin, breast, female and male genital organs, urinary bladder, and thyroid, the number of incident cases was more than double the number of mortality cases. The overall conclusions, regarding which cancer sites provide significant evidence of dose response, are consistent for incidence and mortality data. Both incidence and mortality data show significant excess risks for all solid cancers of the stomach, colon, liver (defined as primary liver cancer or liver cancer not otherwise specified on the death certificate), lung, breast, ovary, and urinary bladder; no significant risk is seen for cancers of the pharynx, rectum, gallbladder, pancreas, nasal cavity, larynx, uterus, prostate, or kidney in either the incidence or mortality series. Disagreements are confined to the risk of esophageal cancer–significantly elevated on the basis of mortality data–but not on the basis of incidence data and the risk of nonmelanoma skin cancer, which is significantly elevated only on the basis of incidence data. Cancers of the salivary glands and thyroid, found to be in excess in the incidence series, were not considered in the earlier mortality analyses.

For all solid tumors, the estimated ERR based on incidence data is about 40% higher than the mortality-based ERR. Furthermore, the incidence-based EAR is 2.7 times higher than the mortality-based estimate. For some cancer sites, the differences between the risks for incidence and mortality are greater. These differences reflect several factors but are largely due to the better representation in the incidence series of relatively nonfatal cancers, such as breast, thyroid, and skin cancers.

 

Both incidence and mortality data are important

These recently completed comprehensive analyses of the LSS cancer-incidence data provide valuable new information on the cancer risk associated with radiation. Especially for fatal cancers, incidence data support the major LSS mortality findings, emphasizing the importance of continued mortality surveillance of A-bomb survivors. The virtually complete ascertainment of deaths among the LSS and other RERF cohorts by means of the unique koseki* system is one of the major strengths of the RERF follow-up investigations. Incidence data are not problem free and are liable to bias unless special care is taken to collect the data systematically.

The present studies demonstrate the usefulness of the tumor-registry-based incidence data, which are crucial to fully evaluating the full impact of radiation effects. Incidence data provide a powerful means of evaluating the risk at the time of disease onset and provide information more useful for cancer etiology. In view of the rapid progress in the treatment and control of cancer, the importance of continued incidence surveillance will increase. With the availability of the two complementary sets of cancer data–incidence and mortality, a broader spectrum of research on radiation effects and risk assessment is now possible.

 

Radiation Research Special Issue:

1. Mabuchi K, Soda M, Ron E, Tokunaga M, Ochikubo S, Sugimoto S, Ikeda T, Terasaki M, Preston DL, Thompson DE: Cancer incidence in atomic bomb survivors. Part I: Use of the tumor registries in Hiroshima and Nagasaki for incidence studies. Radiation Research 137S:1-16, 1994
2. Thompson DE, Mabuchi K, Ron E, Soda M, Tokunaga M, Ochikubo S, Sugimoto S, Ikeda T, Terasaki M, Izumi S, Preston DL: Cancer incidence in atomic bomb survivors. Part II: Solid tumors, 1958-1987. Radiation Research 137S:17-67, 1994
3. Preston DL, Kusumi S, Tomonaga M, Izumi S, Ron E, Kuramoto A, Kamada N, Dohy H, Matsuo T, Nonaka H, Thompson DE, Soda M, Mabuchi K: Cancer incidence in atomic bomb survivors. Part III: Leukemia, lymphoma and multiple myeloma, 1950-1987. Radiation Research 137S:68-97, 1994
4. Ron E, Preston DL, Mabuchi K, Thompson DE, Soda M: Cancer incidence in atomic bomb survivors. Part IV: Comparison of cancer incidence and mortality. Radiation Research 137S:98-112, 1994