or at first childbirth were at greater risk of radiation-induced thyroid cancer.
The risk of benign thyroid adenomas was also studied in more detail (Shore and others 1993b). There were 86 pathologically confirmed thyroid adenomas among the irradiated group and 11 in the sibling controls. The estimated ERR was 6.3 Gy−1 (90% CI 3.7, 11.2) overall and 7.8 Gy−1 when restricted to subjects with doses less than 6 Gy. Adenoma rates were elevated even at lower doses, with a significant increase in the lowest-dose group (<0.25 Gy). The risk continued to be elevated to the end of follow-up.
Analyses of the risk of breast cancer in relation to radiation dose were also carried out in this population. Hildreth and colleagues (1989) reported on the follow-up to 1985 of 1200 women who received X-ray treatment and their 2469 nonirradiated sisters. Twenty-two breast cancer cases were diagnosed in the irradiated group and twelve in the control group. The estimated average dose to the breast was 0.69 Gy. A linear dose-response was observed, with an ERR of 2.48 Gy−1 (95% CI 1.1, 5.2) and an EAR of 5.7 per 104 PY per gray (95% CI 2.9, 9.5).
Two Swedish cohort studies have been performed of patients treated for skin hemangioma in infancy. In the first study (Lundell and others 1994), the cohort consisted of 14,351 infants (less than 18 months of age) treated between 1920 and 1959 at Radiumhemmet, Stockholm, who were followed up for cancer incidence over the period 1958–1986. Radiotherapy was given with -particles, X- and/or -rays, and usually, with some type of 226Ra applicator. Individual organ doses were calculated using treatment information and, for 226Ra needles and tubes, phantom simulations. Seventeen thyroid cancers were registered in this cohort during the follow-up period. The mean dose to the thyroid was 1.07 Gy (range <0.01, 4.34 Gy). A significant excess thyroid cancer incidence was seen in this cohort, starting 19 years after treatment and persisting at least 40 years after irradiation. A significant dose-relationship was observed, with an ERR of 4.92 Gy−1 (95% CI 1.26, 10.2) and an EAR of 0.90 per 104 PY per gray.
Lundell and Holm (1995) also studied the risk of other solid tumors in this cohort. Statistically significantly increased SIRs were seen for cancer of the pancreas and tumors of the endocrine glands, based on small numbers of cases (9 and 16, respectively). For lung cancer (mean dose 0.12 Gy), a nonsignificant ERR of 1.4 Gy−1 was reported (confidence interval not given) and an EAR of 0.33 per 104 PY per gray, based on 11 cases. For stomach cancer (mean dose 0.09 Gy), both the ERR and the EAR were negative (values not reported), based on five cases.
Lundell and colleagues (1996) reported more specifically on the risk of breast cancer among women from this cohort. The mean absorbed dose to the breast was 0.39 Gy (range <0.01, 35.8 Gy). During the follow-up period, 75 breast cancer cases were found in the cohort. A significant linear dose-response relationship was observed, with an ERR of 0.38 Gy−1 (95% CI 0.09, 0.85) and an EAR of 0.41 per 104 PY per gray. This was not modified by age at exposure or by dose to the ovaries. The ERR increased significantly with time since exposure, however, with an ERR at 1 Gy of 2.25 (95% CI 0.59, 5.62) 50 years or more after exposure. The EAR was 22.9 per 104 PY per gray.
In an analysis of leukemia mortality in the same cohort, 20 deaths from leukemia were observed (11 in childhood and 9 among adults). The weighted bone marrow dose was 0.13 Gy on average (range <0.01–4.6 Gy). There was no association between radiation dose and leukemia (childhood or adult) in this cohort. Among those who received more than 0.1 Gy, the ERR was estimated to be 5.1 Gy−1 (95% CI 0.1, 15) for childhood leukemia, −0.02 Gy−1 (95% CI −0.8, 1.9) for adult leukemia, and 1.6 Gy−1 (95% CI −0.6, 5.5) overall.
The second Swedish hemangioma study included 11,807 patients treated with 226Ra between 1930 and 1965 at Sahlgrenska University Hospital in Göteborg (Lindberg and others 1995). The cohort was followed up for cancer incidence over the period 1958–1989. Doses to 11 organs were calculated on the basis of 226Ra activity, location of the hemangioma, and estimated absorbed dose rate in these organs per unit activity in a phantom the size of a 5–6-month-old child. No correction was made for different body sizes according to the age of the child at the time of treatment. A total of 248 malignancies were observed during the study period. A significantly increased risk of cancer was seen overall, as well as tumors of the CNS (34 cases), thyroid (15 cases), and other endocrine glands (23 cases). The mean absorbed dose to the thyroid in this cohort was 0.12 Gy; the ERR for thyroid cancer was estimated to be 7.5 Gy−1 (95% CI 0.4, 18.1) and the EAR 1.6 per 104 PY per gray.
Karlsson and others (1997) studied the risk of intracranial tumors in this cohort further in a cohort and a case-control study. Dose estimation was similar to that described above for subjects in the cohort study, although a correction was made for different age groups (0–4, 5–11, 12–18, and 18+ months). Activity was considered to be uniformly distributed over the treatment area. In the case-control study, the dose at the exact tumor site was calculated by considering the exact distance between the treatment location, according to the record, and the site of the tumor. For the controls, the dose was calculated at the location of the tumor in the corresponding case. In the cohort, 47 intracranial tumors developed in 46 individuals. An excess was found for many histopathological subgroups but was significant only for gliomas and meningiomas. The mean dose to the brain was 0.072 Gy (median 0.023 Gy; range <0.001–2.4 Gy). There was an excess of brain tumors in all dose categories, but no clear dose-response relationship. When analyses were restricted to subjects treated before the age of 7 months, both a linear and a