losis patients having received a highly fractionated dose (with 92 fluoroscopy sessions on average) at a low dose rate.
The ERR/Gy from the studies of acute high-dose-rate exposures are compatible and in the range 0.1–0.4 Gy−1. The study of HD survivors showed little evidence for nonlinearity of the dose-response, despite the fact that the majority of patients received very high doses to the lung (in excess of 30 Gy). A multiplicative interaction was seen in this study between radiation dose and tobacco smoking, with a smaller ERR/Gy among women than men (difference not statistically significant).
It is difficult to evaluate the effects of age at exposure or of exposure protraction based on these studies because only one study (the hemangioma cohort) is available in which exposure occurred at very young ages and protracted low-dose-rate exposures were received. Risk estimates from that study are higher than those seen in other studies, but the difference is not statistically significant.
The study of tuberculosis patients, based on a very large number of lung cancer deaths, appears to indicate that substantial fractionation of exposure leads to a reduction in risk.
Breast cancer is the most commonly diagnosed cancer and cause of cancer mortality among women in North America and Western Europe. Incidence rates are lower in Asian countries. Ionizing radiation is well documented as a cause of breast cancer in women, especially when exposures occur in childhood and around puberty (UNSCEAR 2000b).
Of all the studies reviewed in the medical uses of radiation section above, only 11 provide dose-specific estimates of ERR and/or EAR. Table 7-3 and Figure 7-2 summarize the results from these studies. In the figure, results are shown for all studies as well as restricted to studies in which the average dose to the breast was less than 1 Gy.
In the incidence studies, the ERR/Gy ranges from 0.15 Gy−1 in women who received very high doses for HD radiotherapy (mean among cases = 25 Gy) to 2.5 Gy−1 in populations irradiated for enlarged thymus in infancy. The range in mortality studies is similar: from 0.08 Gy−1 among patients irradiated for treatment of ankylosing spondylitis to 2.7 Gy−1 in women repeatedly exposed to X-rays to monitor scoliosis. As indicated previously, in the international cervical cancer follow-up study (Boice and others 1988) the significant reduction of risk seen among women with intact ovaries was probably attributable to the cessation of ovarian function related to radiotherapy; only the risk estimates in women with no ovaries are considered here. Similarly, the results from Travis and colleagues are restricted to women who had chest radiotherapy only (and hence exclude women with high doses to the ovaries).
Although the risk estimates from these studies vary considerably, confidence intervals are very large and the estimates shown are therefore statistically compatible, except for the study of HD patients where the doses to the breast (up to more than 60 Gy) will have led to cell killing and hence a reduction of risk per gray.
The situation is somewhat different for EAR. Few studies have reported risk estimates in terms of EAR. The estimates shown in Table 7-3 and Figure 7-2 are quite variable, and several of the confidence intervals do not overlap, indicating heterogeneity in risk estimates across these studies. In reviewing these results, differences in study populations and exposure patterns must be taken into account. These include the following:
The thymus and hemangioma studies relate mainly to patients who were irradiated in infancy; in the scoliosis study, the mean age at first exposure was 10.1 years, while in the other studies, the majority of subjects were adults at the time of radiation exposure. Mean ages ranged from 25 to 52 years, respectively, in the Massachusetts fluoroscopy study and the cervical cancer survivor study.
Exposure patterns ranged from very protracted low-dose-rate or fractionated exposures in the scoliosis and tuberculosis studies, where diagnostic radiation was used to monitor the evolution of the disease, to high-dose-rate, acute or much less fractionated exposures received for treatment of disease.
Exposures in childhood and adolescence, particularly in the time around puberty (Doody and others 2000), have been shown to be associated with higher risks of radiation-induced breast cancer than exposures later in life. Figure 7-3 shows the relation between ERR/Gy and average age at exposure. The ERR appears to decrease with increasing age at exposure.
Exposure in infancy led to an ERR of 2.5 Gy−1 in the Rochester thymic irradiation study, based on 22 exposed cases, but the ERR in the pooled analysis of data on hemangioma patients was much lower, 0.35 Gy−1. A previous report of a much higher ERR in the Stockholm hemangioma cohort 50 or more years after exposure was not confirmed in the pooled analysis.
Exposure at age 15 in the Massachusetts fluoroscopy study was estimated to result in an ERR/Gy of 1.0, and the ERR for breast cancer mortality was 2.7 Gy−1 in the scoliosis study (with average age at first exposure of 10.1 years) and 0.9 Gy−1 overall in the Canadian fluoroscopy study for exposures between 15 and 25 years of age. Exposures at older ages tended to result in lower risk estimates, ranging from 0.33 among cervical cancer patients with no ovaries to about 0.61 in women exposed to fluoroscopy, and to about 1.63 following high-dose-rate irradiation for benign breast disease. The ERR was lower, but not statistically incompatible, in the ankylosing spondylitis study, probably related to the fact that the dose to the ovary was high among these women (ERR not shown in Figure 7-3).
Exposure fractionation does not appear to be an important determinant of risk per gray in the fluoroscopy studies.