linear-quadratic model described the data. The estimated ERR for the entire cohort was 1.05 Gy−1; the EAR was 1.20 per 104 PY per gray. In the case-control study, the mean absorbed dose at the site of the tumor was 0.031 Gy for cases and 0.09 for controls. The estimated OR was 1.65 Gy−1 (95% CI 0.63, 4.32).
Pooled analyses of the data on breast cancer and intracranial tumors from the two Swedish hemangioma cohorts were also carried out. In the pooled breast cancer analyses (Lundell and others 1999), 245 breast cancer cases diagnosed between 1958 and 1993 were available. The ERR was estimated to be 0.35 Gy−1 (95% CI 0.18, 0.59) and the EAR 0.72 per 104 PY per gray (95% CI 0.39, 1.14). There was no evidence of an effect of time since exposure on the ERR; the EAR, however, increased with time since exposure. Neither age at exposure, dose rate, nor ovarian dose appeared to have an effect on the ERR.
In the pooled analysis of intracranial tumors (Karlsson and others 1998), 88 tumors were found in 86 individuals between 1958 and 1993. There was a significant dose-response relationship, and increasing age at exposure decreased the magnitude of the risk. The ERR was 2.7 Gy−1 (95% CI 1.0, 5.6) overall; 4.5 Gy−1 for exposure before the age of 5 months; 1.5 Gy−1 for exposures between 5 and 7 months; and 0.4 Gy−1 for exposures at older ages. The overall EAR was 2.12 per 104 PY per gray (95% CI 0.27, 4.38). There was no effect of time since exposure on the ERR, while the EAR increased with time since exposure.
Cancer mortality was studied in a cohort of 7037 patients less than 15 years of age treated for a skin hemangioma between 1940 and 1973 at the Institut Gustave Roussy, near Paris, France (Dondon and others 2004). Among them, 4940 had received radiotherapy. The cohort was followed up from 1969 to 1997, during which time 16 patients died of cancer; 14 of these had received radiotherapy. A nonsignificant excess of cancer-related mortality was observed for irradiated patients compared to the general population (SMR 1.53; 95% CI 0.86, 2.48). The excess was highest among those treated with 226Ra (RR 2.53; 95% CI 0.84, 7.07), in comparison to those who did not receive radiotherapy. No estimate of risk per dose is presented.
In 1974, a prospective follow-up program was set up at the Michael Reese Hospital in Chicago to screen patients who had received X-ray treatment between 1939 and 1962 during childhood for benign head and neck conditions (primarily enlarged tonsils). During the screening, more than 35% of the subjects were found to have thyroid nodules (Schneider and others 1985). Analyses of dose-response relationships for thyroid cancer and thyroid nodules in this cohort were conducted by Schneider and colleagues (1993) with a mean follow-up of 33 years. Individual doses were estimated for study subjects on the basis of treatment records and experiments with an anthropomorphic phantom of a 6-year-old child, together with conversion factors for children of different ages. The average dose to the thyroid was estimated to be 0.6 Gy. Overall uncertainty in thyroid dose estimates for an individual subject is of the order of 50% and is related to the child’s movements during treatment and deviations in height and weight. The cohort included 4296 patients, of whom 3843 had estimated dose to the thyroid and 2634 could be followed up. A total of 1043 nodules and 309 thyroid cancers were diagnosed in the cohort. The ERR/Gy was 3.0 overall; it decreased with increasing age at exposure (from 3.6 for exposures below the age of 1 year to 1.4 for exposures between ages 5 and 15). There was no apparent difference between men and women. The slope of the dose-response relationship appeared to reach a maximum 25–29 years after exposure, but response continued to be elevated 40 years after exposure. The ERR appeared to be greater for cases diagnosed before 1974, when the screening program started (ERRs 9.2 and 1.8, respectively, for the period before 1974 and for 1974 and later, based on 109 and 200 cases), but this difference was not significant (p = .4).
From 1975 through 1982, a follow-up study was performed in Sweden for patients treated with X-rays for cervical tuberculous adenitis (Fjalling and others 1986). Of these patients, 444 underwent thyroid examination on average 43 years after their initial treatment. 101 had undergone surgery for thyroid nodules, including 25 for thyroid cancer. The absorbed dose to the thyroid was estimated to range from 0.4 to 51.0 Gy. A dose-response relationship was seen both for thyroid cancer and for nodules. No estimate of risk per dose is presented.
No study has focused specifically on populations exposed to 131I in childhood or adolescence for the treatment of hyperthyroidism. As indicated in the earlier section on treatment of adult benign thyroid diseases, the number of subjects under age 20 at diagnosis in the hyperthyroidism cohorts is very small. In a review paper, Shore (1992) carried out an analysis of risk in those exposed below age 20 in the Swedish and U.S. studies. The total population was estimated to be 602, with an approximate average follow-up of 10 years and a mean dose to the thyroid of about 88 Gy. Two cases of thyroid cancer were reported compared to about 0.1 expected. The estimated ERR was 0.3 Gy−1 (90% CI 0.0, 0.9) and the EAR 0.1 per 104 PY per gray (90% CI 0.0, 0.2).
Studies of children treated with radiation (X-rays and γ-rays) for benign disease also provide valuable information about the carcinogenicity of low-LET radiation. Studies of patients treated for tinea capitis, enlarged thymus, and benign head and neck diseases have provided much of the quantita-