Land, C. E., Tokunaga, M., Koyama, K., Soda, M., Preston, D. L., Nishimori, I. and Tokuoka, S. Incidence of Female Breast Cancer among Atomic Bomb Survivors, Hiroshima and Nagasaki, 1950–1990. Radiat. Res. 160, 707–717 (2003). An incidence survey of the Life Span Study (LSS) population found 1093 breast cancers among 1059 breast cancer cases diagnosed during 1950–1990. As in earlier breast cancer surveys of this population, a linear and statistically highly significant radiation dose response was found. In the analysis, particular attention was paid to modification of radiation dose response by age at exposure ( e ) and attained age ( a ). Dose-specific excess relative risk (ERR 1Sv ) decreased with increasing values of e and a. A linear dose–response model analysis, with e and a as exponential age modifiers, did not conclusively discriminate between the two variables as modifiers of dose response. A modified isotonic regression approach, requiring only that ERR 1Sv be monotonic in age, provides a fresh perspective indicating that both e and a are important modifiers of dose response. Exposure before age 20 was associated with higher ERR 1Sv compared to exposure at older ages, with no evidence of consistent variation by exposure age for ages under 20. ERR 1Sv was observed to decline with increasing attained age, with by far the largest drop around age 35. Possible explanations for these observations are discussed, along with research approaches that might provide more information.

A wide-ranging search for benign and malignant tumors of the major and minor salivary glands among members of the Life Span Study sample of the Radiation Effects Research Foundation identified 41 malignant and 94 benign incident tumors, including 14 malignant and 12 benign tumors of the minor salivary gland, plus 10 major gland tumors of unknown behavior. Dose-response analyses found statistically significant increases in risk with increasing A-bomb dose for both cancer and benign tumors. Estimated relative risks at 1 Sv weighted tissue kerma (<tex-math>${\rm RR}_{1{\rm Sv}}$</tex-math> with 90% confidence interval in parentheses) were 4.5 (2.5-8.5) for cancer and 1.7 (1.1-2.7) for benign tumors. When analyzed by histological subtype within these two broad groups, it appeared that most of the dose response for malignant tumors was provided by an exceptionally strong dose response for mucoepidermoid carcinoma [11 exposed cases with dose estimates, <tex-math>${\rm RR}_{1{\rm Sv}}$</tex-math> = 9.3 (3.5-30.6)], and most or all of that for benign tumors corresponded to Warthin's tumor [12 cases, <tex-math>${\rm RR}_{1{\rm Sv}}$</tex-math> = 4.1 (1.6-11.3)]. There was a marginal dose response for malignant tumors other than mucoepidermoid carcinoma [<tex-math>${\rm RR}_{1{\rm Sv}}$</tex-math> = 2.4 (0.99-5.7)] but no significant trend for benign tumors other than Warthin's tumor [<tex-math>${\rm RR}_{1{\rm Sv}}$</tex-math> = 1.3 (0.9-2.2)]. Re-examination of the original data from published studies of other irradiated populations may shed new light on the remarkable type specificity of the salivary tumor dose response observed in the present study.

An incidence survey among atomic bomb survivors identified 807 breast cancer cases, and 20 second breast cancers. As in earlier surveys of the Life Span Study population, a strongly linear radiation dose response was found, with the highest dose-specific excess relative risk (ERR) among survivors under 20 years old at the time of the bombings. Sixty-eight of the cases were under 10 years old at exposure, strengthening earlier reports of a marked excess risk associated with exposure during infancy and childhood. A much lower, but marginally significant, dose response was seen among women exposed at 40 years and older. It was not possible, however, to discriminate statistically between age at exposure and age at observation for risk as the more important determinant of ERR per unit dose. A 13-fold ERR at 1 Sv was found for breast cancer occurring before age 35, compared to a 2-fold excess after age 35, among survivors exposed before age 20. This a posteriori finding, based on 27 exposed, known-dose, early-onset cases, suggests the possible existence of a susceptible genetic subgroup. Further studies, involving family histories of cancer and investigations at the molecular level, are suggested to determine whether such a subgroup exists.

More than 30 years ago, population-based tumor registries were established in Hiroshima and Nagasaki. This report, the first of a series of papers on cancer incidence, describes methodological aspects of the tumor registries and discusses issues of data quality in the context of the Life Span Study (LSS) cohort, the major atomic bomb survivor population. The tumor registries in Hiroshima and Nagasaki are characterized by active case ascertainment based on abstraction of medical records at area hospitals, augmented by tissue registries operational in the area and a number of clinical and pathological programs undertaken over the years among the atomic bomb survivors. Using conventional measures of quality, the Hiroshima and Nagasaki tumor registries have a death certificate-only (DCO) rate of less than 9%, a mortality/incidence (M/I) ratio of about 50%, and a histological verification (HV) rate in excess of 70%, which place these registries among the best in Japan and comparable to many established registries worldwide. All tumor registry data pertaining to the LSS population were assembled, reviewed and handled with special attention given to the quality and uniformity of data based on standardized procedures. Special studies and monitoring programs were also introduced to evaluate the quality of the tumor incidence data in the LSS. Analyses were performed to examine the quality of incidence data overall and across various substrata used for risk assessment such as age, time and radiation dose groups. No significant associations were found between radiation dose and data quality as measured by various indices. These findings warrant the use of the present tumor registry-based data for studies of cancer incidence in the atomic bomb survivors.

This report presents, for the first time, comprehensive data on the incidence of solid cancer and risk estimates for A-bomb survivors in the extended Life Span Study (LSS-E85) cohort. Among 79,972 individuals, 8613 first primary solid cancers were diagnosed between 1958 and 1987. As part of the standard registration process of the Hiroshima and Nagasaki tumor registries, cancer cases occurring among members of the LSS-E85 cohort were identified using a computer linkage system supplemented by manual searches. Special efforts were made to ensure complete case ascertainment, data quality and data consistency in the two cities. For all sites combined, 75% of the cancers were verified histologically, 6% were diagnosed by direct observation, 8% were based on a clinical diagnosis, and 12.6% were ascertained by death certificate only. A standard set of analyses was carried out for each of the organs and organ systems considered. Depending on the cancer site, Dosimetry System 1986 (DS86) organ or kerma doses were used for computing risk estimates. Analyses were based on a general excess relative risk model (the background rate times one plus the excess relative risk). Analyses carried out for each site involved fitting 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 effects of migration. In agreement with prior LSS findings, a statistically significant excess risk for all solid cancers was demonstrated [excess relative risk at 1 Sv <tex-math>$({\rm ERR}_{1\ {\rm Sv}})=0.63$</tex-math>; excess absolute risk (EAR) per 10 4 person-year sievert (PY Sv) = 29.7]. For cancers of the stomach (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=0.32$</tex-math>), colon (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=0.72$</tex-math>), lung (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=0.95$</tex-math>), breast (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=1.59$</tex-math>), ovary (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=0.99$</tex-math>), urinary bladder (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=1.02$</tex-math>) and thyroid (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=1.15$</tex-math>), significant radiation associations were observed. There was some indication of an increase in tumors of the neural tissue (excluding the brain) among persons exposed to the bombs before age 20. For the first time, radiation has been associated with liver (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=0.49$</tex-math>) and nonmelanoma skin (<tex-math>${\rm ERR}_{1\ {\rm Sv}}=1.0$</tex-math>) cancer incidence in the LSS cohort. The present analysis also strengthened earlier findings, based on a smaller number of cases, of an effect of A-bomb radiation on salivary gland cancer. There was no significant radiation effect for cancers of the oral cavity and pharynx as a group, esophagus, rectum, gallbladder, pancreas, larynx, uterine cervix, uterine corpus, prostate, kidney and renal pelvis. Analyses of solid tumors individually and in combination revealed no appreciable differences between Hiroshima and Nagasaki (P > 0.5). The combined solid tumor analysis demonstrated a twofold greater relative risk for females than males and a trend for a decreasing relative risk with increasing age at exposure (P < 0.001). Females had a higher relative risk of cancers of the lung, total respiratory system and urinary system than males. The excess relative risk decreased with increasing age at exposure for combined digestive, stomach, nonmelanoma skin, breast and thyroid cancers. For solid cancers combined, the excess cancer risk increased with increasing attained age and was proportional to the background incidence rate. Unadjusted for age at exposure, the excess relative risk for most sites tended to decrease with increasing attained age. For some cancers (colon, breast, central nervous system and kidney) models that allowed the excess relative risk to vary with attained age fit at least as well as models that included age-at-exposure effects. For all solid tumors, excess cancers increased with time since exposure, based on an absolute excess risk model. Averaged over all ages at exposure, the relative risk decreased with time since exposure. Examination of temporal patterns by age-at-exposure groups suggested that the excess relative risk decreased with time for the younger age-at-exposure groups and remained virtually constant for the older cohorts. The LSS has served as one of the major sources of data used for cancer risk estimation. Previous studies focused primarily on the association between cancer mortality and radiation exposure. Although these mortality studies are extremely valuable, the accuracy of cancer diagnoses is limited, and death certificates do not provide adequate information on cancers with relatively high survival rates. Although incidence data also have their limitations (e.g., incomplete case ascertainment and partial reliance on death certificate diagnoses), they can provide more complete data on cancers with good survival, on histological type and on time from exposure to cancer onset. Thus future analyses of atomic bomb survivors should focus on both cancer mortality and incidence.

A binational panel of Japanese and American pulmonary pathologists reviewed tissue slides of lung cancer cases diagnosed among Japanese A-bomb survivors and American uranium miners and classified the cases according to histological subtype. Blind reviews were completed on slides from 92 uranium miners and 108 A-bomb survivors, without knowledge of population, sex, age, smoking history, or level of radiation exposure. Consensus diagnoses were obtained with respect to principal subtype, including squamous-cell cancer, small-cell cancer, adenocarcinoma, and less frequent subtypes. The results were analyzed in terms of population, radiation dose, and smoking history. As expected, the proportion of squamous-cell cancer was positively related to smoking history in both populations. The relative frequencies of small-cell cancer and adenocarcinoma were very different in the two populations, but this difference was accounted for adequately by differences in radiation dose or, more specifically, dose-based relative risk estimates based on published data. Radiation-induced cancers appeared more likely to be of the small-cell subtype, and less likely to be adenocarcinomas, in both populations. The data appeared to require no additional explanation in terms of radiation quality (α particles vs γ rays), uniform or local irradiation, inhaled vs external radiation source, or other population difference.

Ascertainment of breast cancer incidence among the cohort of the RERF Life Span Study extended sample identified 574 breast cancers among 564 cases diagnosed during 1950-1980 of which 412 cancers were reviewed microscopically. There were no dose-dependent differences with respect to diagnostic certainty or histological type. As in previous studies, the dose response appeared to be roughly linear and did not differ between the two cities. The most remarkable new finding was the emergence of a radiation-related excess among women under 10 years of age at exposure. The risk of radiogenic breast cancer appears to decrease with increasing age at exposure, whether expressed in relative or absolute terms. These results suggest that exposure of female breast tissue to ionizing radiation at any time during the first four decades of life, even during the premature stage, can cause breast cancer later in life, and that the length of time that tumor promoters such as endogenous hormones operate following exposure has an important influence on the development of radiation-induced breast cancer. An unresolved question is whether breast cancer risk is increased by radiation exposure at ages older than 40.