7
Medical Radiation Studies

INTRODUCTION

Studies of patients irradiated for the treatment or diagnosis of diseases have provided considerable information for the understanding of radiation risks, particularly for specific cancer types, including thyroid and breast cancer (IARC 2000; UNSCEAR 2000b). Today, approximately 50% of cancer patients are treated using radiation (Ron 1998), and several million cancer survivors are alive in the United States, emphasizing the importance of investigating the long-term consequences of radiotherapy and examining the features of epidemiologic studies of medical radiation.

Large cohorts of radiation-treated patients who have been followed for long periods are available, allowing evaluation of cancer and other late effects. Population-based cancer registries in many countries have been used to identify these patients and to facilitate patient enrollment, thus allowing investigators to determine the risks of a second primary cancer after treatment with radiation for a primary cancer (Boice and others 1985). The characteristically detailed radiotherapy records for cancer patients and patients treated for nonmalignant conditions allow precise quantification of the doses to the organs of individuals, which in turn facilitates the evaluation of dose-response relationships. Frequently, patients with the same initial condition that receive treatments other than radiation are available for comparison, although the clinical indications for treatment may differ.

In most cases, patients received high doses of radiation on the order of 40–60 Gy to the targeted region, aimed at producing cell killing. These “high” doses would decrease with distance from the target tissue, and some tissues might receive doses that are referred to in this report as “low dose” (100 mGy or less). The use of such studies to estimate the effect of low-dose exposures raises a number of questions. The exposures were generally only partial-body exposures in persons who were ill, possibly resulting in a different risk than an equivalent whole-body uniform exposure. Because of their disease, patients may have a different sensitivity to radiation-induced disease than persons who do not have the disease. However, these studies are valuable and will likely become more important in the next decade, both for radiation protection of patients and for radiation protection in general because they provide a unique opportunity to address the following issues:

  • Effects of different radiation types

  • Risk of specific tumor types

  • Effects of potential risk modifiers, including sex, age, and exposure fractionation

  • Possible genetic susceptibility to radiation-induced cancer

In addition to studies of cancer survivors, long-term studies of patients who received radiation therapy for benign conditions such as enlarged tonsils and tinea capitis have also provided important information about radiation-induced cancer risk (UNSCEAR 2000b). These patients are particularly important in the evaluation of radiation risks in the absence of the possibly confounding effects of the malignant disease being treated and/or of concomitant therapy for cancer. Diagnostic radiation procedures, in contrast, generally result in small doses to target organs, and most studies of such exposure provide little information about radiation risks. A number of procedures, however, in particular repeated examinations of air collapse therapy for tuberculosis and of spine curvature for scoliosis, have resulted in sizable doses to specific tissues, and studies of patients who have undergone these examinations provide valuable information on radiation risks (UNSCEAR 2000b). It is noted that, although no informative studies are available, the recent use of computed tomography (CT) can deliver sizable doses, typically of the order of tens of millisieverts per examination (Brenner and Elliston 2004); UNSCEAR (2000b) reports cumulative doses of the order of 100 mSv for children.

As in the other review chapters in this report, studies were judged to be informative for the purpose of radiation risk



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 155
7 Medical Radiation Studies INTRODUCTION radiation-induced disease than persons who do not have the disease. However, these studies are valuable and will likely Studies of patients irradiated for the treatment or diagno- become more important in the next decade, both for radia- sis of diseases have provided considerable information for tion protection of patients and for radiation protection in the understanding of radiation risks, particularly for specific general because they provide a unique opportunity to ad- cancer types, including thyroid and breast cancer (IARC dress the following issues: 2000; UNSCEAR 2000b). Today, approximately 50% of cancer patients are treated using radiation (Ron 1998), and • Effects of different radiation types several million cancer survivors are alive in the United • Risk of specific tumor types States, emphasizing the importance of investigating the long- • Effects of potential risk modifiers, including sex, age, term consequences of radiotherapy and examining the fea- and exposure fractionation tures of epidemiologic studies of medical radiation. • Possible genetic susceptibility to radiation-induced can- Large cohorts of radiation-treated patients who have been cer followed for long periods are available, allowing evaluation of cancer and other late effects. Population-based cancer reg- In addition to studies of cancer survivors, long-term stud- istries in many countries have been used to identify these ies of patients who received radiation therapy for benign patients and to facilitate patient enrollment, thus allowing conditions such as enlarged tonsils and tinea capitis have investigators to determine the risks of a second primary can- also provided important information about radiation-induced cer after treatment with radiation for a primary cancer (Boice cancer risk (UNSCEAR 2000b). These patients are particu- and others 1985). The characteristically detailed radio- larly important in the evaluation of radiation risks in the ab- therapy records for cancer patients and patients treated for sence of the possibly confounding effects of the malignant nonmalignant conditions allow precise quantification of the disease being treated and/or of concomitant therapy for can- doses to the organs of individuals, which in turn facilitates cer. Diagnostic radiation procedures, in contrast, generally the evaluation of dose-response relationships. Frequently, result in small doses to target organs, and most studies of patients with the same initial condition that receive treat- such exposure provide little information about radiation ments other than radiation are available for comparison, al- risks. A number of procedures, however, in particular re- though the clinical indications for treatment may differ. peated examinations of air collapse therapy for tuberculosis In most cases, patients received high doses of radiation and of spine curvature for scoliosis, have resulted in sizable on the order of 40–60 Gy to the targeted region, aimed at doses to specific tissues, and studies of patients who have producing cell killing. These “high” doses would decrease undergone these examinations provide valuable information with distance from the target tissue, and some tissues might on radiation risks (UNSCEAR 2000b). It is noted that, al- receive doses that are referred to in this report as “low dose” though no informative studies are available, the recent use of (100 mGy or less). The use of such studies to estimate the computed tomography (CT) can deliver sizable doses, typi- effect of low-dose exposures raises a number of questions. cally of the order of tens of millisieverts per examination The exposures were generally only partial-body exposures (Brenner and Elliston 2004); UNSCEAR (2000b) reports in persons who were ill, possibly resulting in a different risk cumulative doses of the order of 100 mSv for children. than an equivalent whole-body uniform exposure. Because As in the other review chapters in this report, studies were of their disease, patients may have a different sensitivity to judged to be informative for the purpose of radiation risk 155

OCR for page 155
156 BEIR VII estimation if (1) the study design was adequate (see Chap- TABLE 7-1 Estimated Range of Effective Doses from ter 5 concerning informative study designs and limitations); Diagnostic Radiation Exposures (2) individual quantitative estimates of radiation dose to the organ of interest were available for the study subjects; (3) if Procedure Type of Examination Range of Doses so, the details of the dose reconstruction approach were Conventional Chest films 0.02–10 mGy evaluated; and (4) a quantitative estimate of disease risk in simple X-rays X-rays of bones and skull relation to radiation dose—in the form of an estimated rela- X-ray of abdomen tive risk (ERR) or excess absolute risk (EAR) per gray—was provided. Conventional GI series 3–10 mGy Overall, more than 100 studies of patients receiving diag- complex X-rays Barium enema Intravenous urogram nostic or therapeutic radiation have evaluated the associa- tion between exposure to radiation and risk of cancer at mul- Computed Head injuries 5–15 mGy tiple sites (IARC 2000; UNSCEAR 2000b). Studies that tomography (CT) Whole-body examinations provide information about the size of radiation cancer risks are reviewed in detail in this chapter. Articles included in Spiral CT Head injuries 10–20 mGy Whole-body examinations this chapter were identified principally from searching the PubMed database of published articles from 1990 through Angiography Coronary, aortic, peripheral, 10–200 mGy December 2004. Searches were restricted to human studies carotid, abdominal and were broadly defined: key words included radiation; neoplasms; cancers; radiation-induced; medical exposures; Interventional Angioplasties with stent 10–300 mGy procedures placement radiotherapy; diagnostic radiation; and iodine-131. Articles Percutaneous dilatations, were also identified from UNSCEAR (2000b), from the ref- closures, biopsy procedures erences cited in papers reviewed, and from direct contacts with some of the main scientists who have been involved Internal emitters Radioisotope studies 3–14 mSv with studies of medical exposures in recent years. The data and confidence intervals are those given in the cited papers. Although doses of single procedures are typically low, MEDICAL USES OF RADIATION there is concern that populations of pediatric patients who Medical use of radiation usually occurs under three cir- may need repeated exams over time to evaluate their pulmo- cumstances: (1) treatment of benign disease, (2) diagnostic nary, cardiac, urinary, or orthopedic conditions may receive examination, and (3) treatment of malignant disease (Ta- relatively high cumulative doses. Similarly, adult patients ble 7-1). Diagnostic imaging using X-rays goes back to the may also require repeated examinations to evaluate fracture time of Roentgen’s discovery in 1896. Diagnostic proce- healing, or progression of pulmonary disease, or the regres- dures, particularly the widespread use of X-rays, continue to sion or progression of neoplastic lesions. be the most common application of radiation in medicine, In contrast, therapeutic exposures are less frequent, and even as non-ionizing radiation methods—ultrasound and the dose levels are higher in view of the different purpose. magnetic resonance imaging—have become more generally Currently, radiotherapy is used mainly for the treatment of accepted. Approximately 400 million diagnostic medical cancer, where the intention is to deliver a lethal dose to ma- examinations and 150 million dental X-ray examinations are lignant tissue within a well-defined target volume, while performed annually in the United States (Mettler and others minimizing the irradiation of surrounding healthy tissue. In 1996). On average, each person receives at least two exami- the past, high doses of radiation have also been used for the nations per year. The annual individual and collective effec- treatment of a number of benign conditions, such as enlarged tive doses from diagnostic medical X-rays have been esti- thymus and ringworm of the scalp (tinea capitis). Doses from mated as 0.5 mSv and 130,000 person-Sv (UNSCEAR radiotherapy to the target organs are generally above 1 Gy 2000b). (and typically in the range of 50–60 Gy for the treatment of The range of X-ray techniques used includes radiogra- malignant diseases). Radiotherapy involves mainly partial- phy, fluoroscopy, CT, interventional radiology, and bone body irradiation, however; hence very different doses are densitometry. These procedures are intended to provide delivered to different organs or tissues of the body. Doses to diagnostic information and in principle are conducted with distant organs are generally considerably lower (of the order the lowest practicable levels of patient dose to meet clinical of fractions of a gray), and studies of cancer risk in these objectives. Ranges of typical doses from various medical organs are therefore potentially informative for the assess- diagnostic exposures are shown in Table 7-1. ment of risks associated with low-level exposure. Further, many of the patients treated with radiotherapy received frac-

OCR for page 155
MEDICAL RADIATION STUDIES 157 tionated doses, and studies of these patients provide the po- plete follow-up (hundreds of cases and controls, with follow- tential to study the effects of exposure fractionation and pro- up of 10–20 years or more); chemotherapy was rarely used; traction. and the existence of radiotherapy records facilitated the development of a comprehensive dose reconstruction sys- tem to estimate individual doses. Radiotherapy for Malignant Disease In an expanded case-control study nested within this in- Studies of second cancer following radiotherapy have ternational cohort (Boice and others 1988), radiation doses generally focused on patients treated for cervical cancer, for selected organs were reconstructed from original radio- breast cancer, Hodgkin’s disease (HD), and childhood can- therapy records. Very high doses, of the order of several cers (i.e., patients that generally have a favorable long-term hundred grays delivered to the cervix, significantly increased prognosis). Survivors of these cancers may live long enough the risks for cancers of the bladder, rectum, and vagina and to develop a second, treatment-related malignancy. It should possibly bone, uterine corpus, cecum, and non-Hodgkin’s be noted that chemotherapy and/or hormonal therapy used in lymphoma (NHL). Doses of several grays increased the risks the treatment of cancers is a potential confounding factor in for stomach cancer and for leukemia. The ERR1 for stomach investigations of the risk of a second primary cancer. cancer was 0.54 Gy–1 (90% CI 0.05, 1.5), with an excess attributable risk of 3.16 per 104 person-years (PY) per gray (0.05, 10.4), based on 348 cases and 658 controls. A nonsig- Cervical Cancer nificant twofold increase in the risk of thyroid cancer was The treatment of cervical cancer involves external beam observed, with an average dose of 0.11 Gy (43 cases and 81 radiotherapy or radium or cesium in applicators to deliver controls). high local doses of X-rays and gamma rays to the cervix More detailed dose-response investigations were carried uteri and adjacent organs in the abdomen and pelvic area. out for leukemia and breast cancer after treatment for cervi- Treatment is usually successful, and patients survive for cal cancer. The case-control study of leukemia risk (Boice years after radiotherapy. Although doses to the cervix are and others 1987) included 195 cases and 745 controls, of very high (typically 40–150 Gy), doses to distant organs are whom 181 and 672, respectively, had received radiotherapy. significantly lower: of the order of 0.1 Gy to the thyroid, Radiation dose to the active bone marrow was estimated 0.3 Gy to the breast and the lung, 2 Gy to the stomach, and from detailed radiotherapy records of the subjects. Radiation 7 Gy to the active bone marrow (Kleinerman and others exposure did not affect the risk of chronic lymphocytic leu- 1995). kemia (CLL; 52 cases). For other forms of leukemia taken Most of the information on second cancers following ra- together (143 cases), there was a significant twofold increase diotherapy for cervical cancer comes from an international in risk associated with radiotherapy; the risk increased with cohort study of approximately 200,000 women treated for increasing dose up to about 4 Gy and then decreased at cervical cancer. The study involved the follow-up, based on higher doses and was modeled adequately by a linear-expo- 15 cancer registries in eight countries (Canada, Denmark, nential function. The linear term of this model for leukemia Finland, Norway, Sweden, the United Kingdom, the United other than CLL provides an estimate of the ERR per gray in States, and Yugoslavia [Slovenia]), of a multinational co- the low-dose range, where cell killing is negligible; this esti- hort of nearly 200,000 women patients treated for cancer of mate is 0.88 Gy–1 (standard error = 0.69). the cervix after 1960. In 1985, Boice and colleagues reported The case-control study of breast cancer included 953 on 5146 second cancers that were diagnosed in this cohort cases and 1806 controls (Boice and others 1989). Radiation up to 1980 and showed an increased risk of cancer following doses to the breast (average 0.31 Gy) and ovaries (average radiotherapy at a number of sites (Boice and others 1985). 32 Gy) were reconstructed from original radiotherapy Kleinerman and coworkers (1995) extended the follow-up records. Overall, there was no association between radio- of this cohort, adding an additional 10 years of incident therapy and risk of breast cancer. Among women with intact cases. Several registries from the original study were re- ovaries (561 cases), radiotherapy was associated with a sig- tained, and other registries were added to increase the num- nificant reduction of risk, probably attributable to cessation ber of nonexposed comparison subjects. A total of 7543 cases of ovarian function. Among women with no ovaries, there were included. This study confirmed earlier findings of in- was a slight increase in breast cancer risk and a suggestion of creased risk of malignancies following radiotherapy and the a dose-response with a relative risk (RR)2 of 1.0, 0.7, 1.5, persistence of increased risk over time. and 3.1, respectively, for the dose groups 0, 0.01–0.24, 0.25– Case-control studies of specific cancer types, nested 0.49, and 0.5 + Gy. From these data, UNSCEAR (2000b) within this cohort, allowed the reconstruction of individual doses to specific organs and the estimation of site-specific 1ERR is the rate of disease in an exposed population divided by the rate cancer risks (Boice and others 1987, 1988, 1989). These of disease in an unexposed population minus 1.0. studies are based on incidence data; the numbers of exposed 2RR is the rate of disease in an exposed population divided by the rate of and unexposed patients were large; there was long and com- disease in an unexposed population.

OCR for page 155
158 BEIR VII estimated an ERR per gray of 0.33 (< –0.2, 5.8) for women cases having been diagnosed within 10 years of their initial with no ovaries and of –0.2 (< –0.2, 0.3) overall. disease (Kaldor and others 1992). A cohort study of second cancer risk following radiation In 1995, Boivin and collaborators published results of a therapy for cancer of the uterine cervix was also carried out joint Canada-U.S. study of second cancer risk among 10,472 in Japan among 11,855 patients (Arai and others 1991). Sig- patients treated for HD between 1940 and 1987. A total of nificant excesses of leukemia and of cancers of the rectum, 122 leukemia and 438 solid tumors were found, and nested bladder, and lung were observed. No estimation of organ case-control studies were carried out. Significant increases dose is available. in the risk of cancers of the respiratory system, intrathoracic organs, and female genital system were observed among patients followed for 10 years or more after surgery. Esti- Hodgkin’s Disease mates of organ doses were not available, and analyses by The large radiation therapy fields used to treat HD and level of radiation dose are not shown. the young age and long survival of patients provide an Van Leeuwen and collaborators (1995) conducted a case- opportunity to study the risk of second cancer after exposure control study of lung cancer nested in a cohort of 1939 pa- to ionizing radiation. Most patients, however, in the past tients treated for HD between 1966 and 1986 in the Nether- 20 years, have been treated with a combination of radio- lands. Radiation dose to the parenchyma, bronchi, and therapy and chemotherapy. trachea were estimated for patients who had received radio- Following a first report by Arseneau and collaborators therapy (30 cases and 82 controls). A statistically significant (1972), a number of authors have studied the risk of second increase in the risk of lung cancer was observed, with an RR cancer following treatment for HD (Boivin and others 1984). of 9.6 (95% CI 0.93, 98.0) for patients who had received Initial reports focused mainly on the risk of leukemia fol- 9 Gy or more compared to patients who had received less lowing this treatment, but as longer follow-up periods were than 1 Gy. The increase was greater among those who either considered, an excess risk of a number of solid cancers (in continued smoking or started smoking after diagnosis, and a particular breast and lung) became apparent. multiplicative interaction was observed between radiation The results of the first multinational study were published dose and tobacco smoking. in 1987 by Kaldor and collaborators. The study involved the Swerdlow and collaborators (2001) carried out a nested follow-up (based on 11 cancer registries in seven countries: case-control study of lung cancer in a cohort of 5519 pa- Denmark, Finland, Norway, Sweden, Slovenia, Canada, and tients with HD treated in Britain between 1963 and 1993. the United Kingdom) of a cohort of 28,462 patients treated The study included 88 cases and 176 controls for whom treat- for HD between 1950 (in the earliest countries) and 1984. ment and other risk factor information was abstracted from Increases in the risk of NHL, leukemia, lung, bladder, and medical records. An increased risk of lung cancer following breast cancer were reported in this cohort. No treatment radiotherapy was observed. No individual reconstruction of information was available in this study, and no information dose to the lung was carried out. is provided on radiation risks. Nested case-control studies of Travis and colleagues (2002) carried out a case-control leukemia and of lung cancer were carried out, allowing study of lung cancer nested within a multinational cohort of reconstruction of individual doses for the subjects and esti- 19,046 HD patients diagnosed between 1965 and 1994 and mation of site-specific cancer risks (Kaldor and others 1990a, reported to population-based cancer registries in Connecti- 1992). cut, Iowa, Denmark, Finland, the Netherlands, Sweden, and The case-control study of leukemia included 163 cases Canada (Ontario). The study included 222 cases and 444 and 455 controls. Radiation dose to the active bone marrow matched controls. Nineteen of the cases were included in the was estimated for subjects who had undergone radiotherapy, previous case-control study by Kaldor and coworkers (1992). and doses were classified into three categories (<10, 10–20, Dose to the specific location of the lung where the tumor had and 20+ Gy). Among patients who did not receive chemo- developed (and to a comparable location for matched con- therapy, a significant increase in the risk of leukemia was trols) was calculated from radiotherapy records. The mean seen at doses of more than 20 Gy (Kaldor and others 1990a). dose was 27.2 Gy in cases and 21.8 Gy in controls. In sub- Another case-control study from the same collaborative jects who had not undergone chemotherapy, a significantly group involved 98 cases of lung cancer occurring between increased risk of lung cancer was observed (odds ratio [OR]3 1960 and 1987 and 259 matched controls (Kaldor and others 5.9; 95% CI 2.7, 13.5) for a dose of 5 Gy or more. A signifi- 1992). Radiation dose to the lung as a whole was estimated cant trend in risk was observed with increasing dose. for the 60 cases and 275 controls who had undergone radio- In a follow-up to this study, Gilbert and colleagues (2003) therapy, and doses were classified into three categories (<1, analyzed radiation effects among 227 lung cancer cases and 1–2.5, and 2.5+ Gy). Among patients treated with radio- therapy alone, there was a nonsignificant increase in risk in relation to radiation dose level. It is noted that the follow-up 3OR is the odds of being exposed among diseased persons divided by the was short in this study, with three-quarters of the lung cancer odds of being exposed among nondiseased persons.

OCR for page 155
MEDICAL RADIATION STUDIES 159 455 controls (the 199 cases and 393 controls from the Travis dose, chemotherapy, and hormonal factors in breast cancer 2002 study who had adequate radiation dose information and following HD. As in the Travis study, the risk of breast can- 28 cases and 62 controls from the Dutch study of van cer increased with radiation dose up to at least 40 Gy. A Leeuwen and others 1995). Doses to the lung ranged from substantial risk reduction was associated with chemotherapy, zero to more than 60 Gy; the distribution of doses was bimo- which affects menopausal age, suggesting that ovarian hor- dal, with most subjects having received doses of less than 5 mones promote tumorigenesis after radiation-induced initia- or more than 30 Gy. To account for a possible latent period tion. No estimate of ERR or EAR4 per gray is given. Little if between radiation exposure and lung cancer resulting from any increased risk was seen for patients treated after age 30. that radiation exposure, only doses received more than Most recently, Dores and colleagues (2002) studied the 5 years in the past were considered. Among the 146 cases risk of second cancers in general among 32,581 HD patients and 271 controls who had received radiotherapy more than (including 1111 25-year survivors of HD) registered in 16 5 years in the past, a significant association was seen be- population-based cancer registries in North America and tween radiation dose and risk of lung cancer, with an ERR Europe. A total of 2153 second cancers were observed be- per gray of 0.15 (95% CI 0.06, 0.39). There was little evi- tween 1935 and 1994. As before, significant increases in the dence for nonlinearity of the dose-response, despite the fact risk of a number of second malignancies were observed. that the majority of patients received doses to the lung in Although the elevated risks of cancers of the stomach, breast, excess of 30 Gy. Information about smoking and radio- and uterine cervix appeared to persist for 25 years or more, therapy was available for the study subjects. A multiplica- an apparent decrease in the risk of other solid tumors is sug- tive interaction was seen between radiation dose and tobacco gested. These cohort studies, although they provide impor- smoking and an additive interaction with chemotherapy. The tant information concerning treatment-related second neo- ERRs for men and women were respectively 0.18 (0.063, plasms and their patterns of risk over time, do not provide 0.52) and 0.044 (–0.009, 0.53); the difference between the quantitative information on the risk of radiation-induced sexes was not statistically significant. cancer because of the absence of individual dose estimates. Breast cancer following treatment for Hodgkin’s disease The risks of breast, thyroid, and gastrointestinal cancers has also been studied in a number of cohorts. Travis and were also investigated in patients treated for HD at Stanford collegues (2003) carried out a nested case-control study of University Medical Center (Hancock and others 1991, breast cancer in a cohort of 3817 one-year survivors of HD 1993a; Birdwell and others 1997). Increases in these dis- diagnosed at age 30 years or younger between 1965 and 1994 eases were observed, but no dose estimates were available. and included in cancer registries in Iowa, Denmark, Finland, Hancock and colleagues (1993b) also investigated mortality Sweden, the Netherlands, and Ontario, Canada. Individual from heart disease following treatment for HD in a cohort of doses to the area of the breast from which the tumor arose 2232 patients treated from 1960 to 1991 with an average were reconstructed using detailed radiotherapy records and follow-up of 9.5 years. The RR for mortality due to heart results of experiments with phantoms. Mean dose delivered diseases was 3.5 (95% CI 2.7, 4.3) among those who re- to the location of the breast where cancer developed was ceived mediastinal radiation doses of more than 30 Gy. The 25.1 Gy (range: 12.0–61.3 Gy) in cases and 21.1 Gy (range increased risk was highest for exposures that occurred before 0–56.0 Gy) in controls. The study included 105 cases and the age of 20 and increased with time since exposure. No 266 controls. A significant increase in the risk of breast can- increased risk was observed among subjects who received cer was seen following doses of 4 Gy or more (OR 3.2; 95% doses lower than 30 Gy. In a separate study, Heidenreich CI 1.4, 8.2); the increase remained significant even follow- and coworkers (2003) found a high prevalence of asymp- ing very high doses (OR 8, 95%; CI 1.6, 26.4, of 40 Gy or tomatic heart disease—specifically aortic valvular disease— more). No significant association between age at exposure following mediastinal irradiation. or reproductive history was seen in this study, but the risk was lowered among women who received 5 Gy or more to Breast Cancer the ovaries or who were also treated with alkylating agents. The estimated ERR per gray for women who did not receive Leukemia, lung cancer, soft tissue sarcoma, and contralat- alkylating agent chemotherapy or high radiation doses to eral breast cancer have been studied in patients receiving their ovaries was 0.15 (95% CI 0.04, 0.74). radiotherapy for breast cancer. Van Leeuwen and colleagues (2003) also studied the risk A case-control study of leukemia (excluding CLL) was of breast cancer among female survivors of HD treated in carried out nested within a cohort of 82,700 women with the Netherlands. The study included 48 cases who devel- breast cancer in the United States. A total of 90 cases and oped breast cancer 5 years or more after HD diagnosis and 264 controls were included with individual estimates of dose 175 matched controls. It should be noted that 40 of the 48 cases in the study of van Leeuwen and colleagues were also included in the study by Travis and coworkers (2003). The 4EAR is the rate of disease in an exposed population minus the rate of object of the study was to evaluate the joint roles of radiation disease in an unexposed population.

OCR for page 155
160 BEIR VII to the active bone marrow. A significant dose-response was mor were estimated from detailed radiotherapy records. seen for acute nonlymphocytic leukemia after adjustment for More than 40% of the irradiated patients received a local the amount of chemotherapy, with an RR of 2.4 among those dose of less than 1 Gy. A significant quadratic dose-response who received radiotherapy alone (Curtis and others 1992). was found in this study, with an excess risk of all second No information was provided on the magnitude of the risk malignant neoplasms combined of 0.2% (95% CI 0.05, per gray or on the risk of other forms of leukemia. 0.5%) at 1 Gy. A case-control study of contralateral breast cancer was Darby and coworkers (2003) studied cardiovascular mor- carried out nested within a cohort of 41,109 women in Con- tality in a cohort of 89,407 Swedish women identified from necticut diagnosed with breast cancer between 1935 and the Swedish cancer registry as having had unilateral breast 1982. A total of 655 cases and 1189 controls were included. cancer at the ages of 18 to 79 years between 1970 and 1996. The average dose to the contralateral breast was 2.8 Gy. A Mortality from cardiovascular disease was higher in women significant increased risk was seen only among women who who had left-sided tumors (OR 1.10; 95% CI 1.03, 1.18) received radiotherapy before age 45 (RR 1.59; 95% CI 1.07, 10 years or more after the diagnosis of breast cancer; for 2.36, based on 78 exposed cases); a significant dose-response ischemic heart disease, the OR was 1.13 (95% CI 1.03, 1.25). was observed in this group (Boice and others 1992). No dose estimates were available in this record linkage study, No excess risk of contralateral breast cancer was seen in a but the fact that the increase was restricted to women with cohort of 14,000 women treated between 1946 and 1982 in tumors in the left breast and that no increase in mortality Denmark (Basco and others 1985). The study included 194 from other causes (except breast cancer) was seen in this cases with individual dose estimates (mean doses ranging population lends plausibility to the hypothesis of a radiation from 1.4 to 3.3 depending on the type of radiotherapy and effect on the risk of heart disease. the field considered). The RR per 100 cGy was 0.99 (95% CI 0.76, 1.30); little difference was seen for those diagnosed 5– Ovarian Cancer 10 years or more after their first tumor. A case-control study of leukemia within an international A case-control study of contralateral breast cancer was cohort of 99,113 survivors of ovarian cancer showed no sig- conducted among women with primary breast cancer entered nificant excess risk for leukemia associated with radio- in the Danish Cancer Registry from 1943 to 1978 (Storm therapy alone (Kaldor and others 1990b). A more recent and others 1992). A total of 529 cases and 529 controls were case-control study was carried out, nested within an interna- included, and individual doses to the contralateral breast tional cohort of 28,971 patients in whom ovarian cancer was were estimated from detailed radiotherapy records for all diagnosed between 1980 and 1993 (Travis and others 1999). subjects who had received radiotherapy. The mean dose to The study included 96 leukemia cases and 272 controls. In- the contralateral breast was estimated to be 2.5 Gy. There dividual dose to the active bone marrow was estimated for was no significantly increased risk of contralateral breast the 26 cases and 79 controls who had received radiotherapy. cancer in this study (RR = 1.04; 95% CI 0.74, 1.46). The median dose to the bone marrow was 18.4 Gy. Radio- A case-control study of lung cancer was conducted based therapy increased the risk of leukemia following platinum- on the Connecticut Tumor Registry (Inskip and others 1994). based chemotherapy. No increased risk of leukemia was ob- The study included 61 cases of lung cancer and 120 controls. served in subjects who had radiotherapy alone; the data are Cases were diagnosed between 1945 and 1981 among sparse: one exposed case and 36 exposed controls. women who had been treated for invasive breast cancer be- tween 1935 and 1971 and survived at least 10 years. Indi- Testicular Cancer vidual radiation dose to different segments of the lung was estimated from detailed radiotherapy records. Average dose Travis and colleages (1997) studied second cancer inci- to the lung was 15.2 Gy to the ipsilateral lung and 4.6 Gy to dence in a multinational cohort of 28,843 men who had been the contralateral lung. Patients who received radiotherapy diagnosed with testicular cancer between 1935 and 1993 in had a 1.8 times higher risk of developing lung cancer than the United States, Denmark, Finland, the Netherlands, Swe- those who did not (95% CI 0.8, 3.8). The risk increased with den, and Canada (Ontario). Cases of second cancers occur- time since exposure and appeared to be higher among women ring between 1965 and 1994 in this cohort were identified exposed under age of 45, although this was not significant. from population-based cancer registries in these countries. The risk was highest for the ipsilateral lung. The ERR was Significantly increased risks of second cancers in general, as estimated to be 0.2 Gy–1 to the affected lung (95% CI –0.62, well as of leukemia (64 cases) and stomach cancer (93 cases), 1.03), based on 15 exposed cases. were observed among patients who had received radio- A nested case-control study of second malignant neo- therapy 5 years or more in the past. No individual doses were plasms was carried out for a cohort of 7771 women initially available. treated for breast cancer between 1954 and 1983 at the Travis and colleagues (2000) conducted a case-control Institut Gustave Roussy near Paris, France (Rubino and oth- study of leukemia nested within a multinational cohort of ers 2003). Individual doses to the location of the second tu- 18,567 patients diagnosed with testicular cancer between

OCR for page 155
MEDICAL RADIATION STUDIES 161 1970 and 1993 and registered in cancer registries in Iowa, life expectancy of children with cancer and increased the Connecticut, New Jersey, Canada (Ontario), Denmark, Fin- chance of development of second cancers. Since childhood land, the Netherlands, and Sweden. The study included 36 cancer is rare, national and international groups such as the cases and 106 matched controls. Individual radiation dose to Late Effects Study Group (Tucker and others 1987a, 1987b, the active bone marrow was estimated from detailed radio- 1991) and several groups in the United Kingdom (Hawkins therapy records. In men who did not receive chemotherapy and others 1987) and France (de Vathaire and others 1989, (mean radiation dose to 12.6 Gy), a 3.1-fold elevation of leu- 1999) have combined their data to evaluate risks. Results kemia risk was observed (95% CI 0.7, 22). The risk increased from these cohort studies have indicated that the risk for with radiation dose to the active bone marrow, with an OR developing a second cancer in the 25 years after the diag- of 19.7 (95% CI 1.5, 59) for doses of 20 Gy or more (based nosis of the first cancer was as high as 12% (Tucker and on four exposed cases). No estimate of ERR or EAR per others 1991). Further, genetic predisposition appears to have gray is given. a substantial impact on risk of subsequent cancers. Among patients treated for hereditary retinoblastoma, the risk of developing a second cancer in the 50 years after the initial Thyroid Cancer diagnosis was as high as 51% (Wong and others 1997b). A cohort of 834 thyroid cancer patients treated with io- Three nested case-control studies including 64 cases of dine-131 and of 1121 thyroid cancer patients treated by other bone cancer and 209 controls (Tucker and others 1987a), 23 means in Sweden between 1950 and 1975 was followed for cases of thyroid cancer and 89 controls (Tucker and others cancer occurrence (Hall and others 1991). The average 131I 1991), and 25 cases of leukemia and 90 controls (Tucker and cumulative activity administered was 4.55 GBq. The aver- others 1987b) were conducted from the Late Effects Study age duration of follow-up was 14 years. A total of 99 second Group cohort of 9170 children who developed a second ma- cancers were found 2 years or more after 131I therapy among lignant tumor at least 2 years after diagnosis of the first tu- those treated with this modality and 122 among those treated mor. A significant increased risk of bone cancer was found by other means. The incidence of second malignancy was among patients who received radiation therapy (RR 2.7; 95% higher among those treated with 131I. Among women, the CI 1.0, 7.7), with a sharp dose-response gradient reaching a overall standardized incidence ratio (SIR)5 was 1.45 (95% fortyfold risk following doses to the bone of more than CI 1.14, 1.83), and significantly elevated SIRs were found 60 Gy. A significant increased risk of thyroid cancer was for tumors of the salivary glands, genital organs, kidney, and also found among patients who had received radiation adrenal gland. A significant trend was seen with increasing therapy; most of the increase was among those who had re- 131I activity, with a SIR of 1.80 (95% CI 1.20, 2.58) for ceived doses of 2 Gy or more. There was no evidence of a administered activities of 3.66 GBq and above. dose-response relationship for leukemia. A cohort of 1771 patients treated with 131I for thyroid In a U.K. cohort of 10,106 3-year survivors of childhood cancer was followed up for incidence of second cancers (de cancer, Hawkins and colleages (1987; Hawkins 1990) reported Vathaire and others 1997). The average 131I cumulative ac- an excess of second tumors among subjects who had received tivity administered was 7.2 GBq, resulting in an estimated radiotherapy in comparison with the general population. In average dose of 0.34 Sv to the bone marrow and 0.80 Sv to addition, two nested case-control studies of 59 cases of sec- the whole body. After a mean follow-up of 10 years, no case ond bone cancer and 220 controls (Hawkins and others 1996) of leukemia was seen. Eighty patients developed a second- and 26 cases of second leukemia and 96 controls (Hawkins ary solid cancer, including 13 colorectal cancers. The risk of and others 1992) were conducted within this cohort, with colorectal cancer was related to the total activity adminis- individual dose reconstruction to the organs of interest. The tered (ERR = 0.47 GBq–1; 95% CI 0.1, 1.6). The overall ERR risk of bone cancer increased substantially with increased for solid tumors in this study was 0.38 per estimated effective cumulative radiation dose to the bone (p < .001), although a sievert (95% CI –0.22, 1.2); when tumors of the digestive decline in risk was seen at doses equal to or greater than track were excluded, the ERR was reduced to –0.15 Sv–1 50 Gy. A nonsignificant increased risk of leukemia was ob- (95% CI –0.35, 0.22). served among those who had received radiotherapy (RR 8.4; 95% CI 0.9, 81.0 based on seven exposed cases). A signifi- cant dose-response relationship was observed. Childhood Cancers In a cohort study of 634 children treated for childhood The treatment for childhood cancers, often a combination cancer from 1942 and 1969 in the Institut Gustave Roussy in of both radiotherapy and chemotherapy, has prolonged the Paris, a twofold increase in the risk of second malignancy was seen after doses from radiotherapy of more than 25 Gy, based on two exposed cases (de Vathaire and others 1989). 5SIR is the ratio of the incidence rate of a disease in the population being A nonsignificant dose-response was seen based on 13 cases studied divided by the comparable rate in a standard population. The ratio is who had received radiotherapy alone. similar to an RR times 100.

OCR for page 155
162 BEIR VII In a French-British cohort study (de Vathaire and others tion with age at exposure in this study. A complex dose- 1999) that overlapped partially with the Late Effects Study, response was observed, with a relative risk of 1.8 (95% CI the French study, and British studies, described above, an 1.1, 2.8) following continuous exposures from brachy- excess of second cancers was seen among 1045 children who therapy6 at comparatively low doses and low dose rates received radiotherapy alone (based on 31 second malignant (mean total dose 1.7 Gy). The risk was of the same order neoplasms, including 8 brain cancers). Fourteen cases of thy- (RR 2.3; 95% CI 1.4, 3.7) after fractionated exposures at roid carcinoma were identified in the entire cohort of 4096 much higher doses and dose rates from external beam treat- 3-year survivors of childhood cancers. All 14 had received ment (mean total dose 9.9 Gy). radiotherapy. The average dose to the thyroid in this study was 7 Gy. A significant dose-response was observed for thy- Summary roid cancer in this study—RRs were 4.0 (90% CI 0.7, 44), Studies of second cancer following radiotherapy have 11.0 (90% CI 2.3, 123), 13.0 (90% CI 2.2, 141), and 26.0 generally focused on patients treated for malignant diseases (90% CI 3.4, 308) for doses within the ranges of 0.25 to with a favorable long-term prognosis, such as cervical can- <1 Gy (3 cases), 1.0 to <10 Gy (5 cases), 10 to <30 Gy (3 cer, breast cancer, HD, and childhood cancers. Because cases), and 30+ Gy (2 cases), respectively. many survivors of these cancers live long enough to develop In a joint analysis of data from childhood cancer survivor a second, treatment-related malignancy, these studies have cohorts from France, Britain, and Nordic countries, a nested provided valuable information on the magnitude of risk case-control study of melanoma was carried out. Radio- following radiation exposure. The cohort studies generally therapy appeared to increase the risk of melanoma for local do not provide quantitative information on the risk of doses greater than 15 Gy (OR 13.0; 95% CI 0.94, 174.0), radiation-induced cancer because of the absence of indi- based on three exposed cases; the ORs for doses less than vidual dose estimates. 1 Gy and of 1–15 Gy were 1.4 (95% CI 0.28, 7.0) and 3.2 Case-control studies of specific cancer types have been (95% CI 0.37, 27) based on very small numbers of exposed carried out, nested within cohorts of cancer survivors. In al- cases—five and two, respectively (Guerin and others 2003). lowing the reconstruction of individual doses to specific or- A partially nested case-control study of soft tissue sar- gans for the subjects, they have provided important informa- coma (STS) was carried within the French-U.K. cohort of tion for the estimation of site-specific cancer, even if the 4400 3-year survivors of childhood cancer survivors; 25 average doses to the target organs have generally been high. cases and 121 controls were included. Individual dose to the Studies of patients treated for HD have provided quantita- site of STS development was calculated. A significant in- tive estimates of the risk of cancers of the lung and breast— crease in the risk of STS was seen among those who re- organs that generally received fairly high doses (of the order ceived radiotherapy (OR 19.0; 95% CI 3.0, 60.0). The risk of 20 Gy on average) from the radiotherapy. Studies of pa- increased with the square of the radiation dose and was inde- tients treated for cancer of the cervix have provided esti- pendent of chemotherapy (Menu-Branthomme and others mates of the risk of breast cancer, leukemia, and stomach 2004). cancer (at average doses of 0.2, 7, and 2 Gy, respectively). Studies of women treated for a first breast cancer have pro- Other Cancers vided quantitative estimates of the risk of lung cancer, at average doses of the order of 5–15 Gy. These estimates are The health effects of radiotherapy for a number of other reviewed in detail, and compared with risk estimates derived cancer types have also been considered in single studies. from other medical exposure studies, in the section “Evalua- Travis and colleagues (1991) studied the risk of second can- tion of Risk for Specific Cancer Sites.” cers among 29,153 patients diagnosed with NHL between 1973 and 1987 in nine areas of the United States. Radiation therapy appeared to increase the risk of acute nonlympho- Radiotherapy for Benign Disease Among Adults cytic leukemia and possibly of cancers of the lung, bladder, In the past, radiotherapy has been used in different coun- and bone. No estimate of radiation dose was available. tries for the treatment of a number of benign conditions in Curtis and coworkers (1994) studied the risk of leukemia children (skin hemangioma, tinea capitis, enlarged thymus) following cancer of the uterine corpus in a cohort of 110,000 and adults (e.g., benign breast and gynecological disease, women assembled from nine population-based cancer regis- ankylosing spondylitis, peptic ulcer). Studies of patients tries in the United States, Canada, Denmark, Finland, and treated with radiation (X-rays and gamma rays) for benign Norway. Radiation doses were computed to 17 sections of disease provide valuable information about the carcinoge- the active bone marrow for 218 women who developed leu- nicity of low-LET (linear energy transfer) radiation. Doses kemia and 775 matched controls. There was no association between radiation dose and risk of CLL (RR 0.90; 95% CI 0.4, 1.9). For all leukemia excluding CLL, however, the RR 6Radiation therapy in which a radioactive material sealed in needles, or was 1.92 (95% CI 1.3, 2.9). There appeared to be no associa- wires, or other small delivery devices is placed directly into or near a tumor.

OCR for page 155
MEDICAL RADIATION STUDIES 163 used in the treatment of benign conditions were generally diotherapy compared to those who did not. Overall, a 50% not as high as those used to treat malignant disease so that increase in the risk of all cancers combined was observed. cell-killing effects do not predominate, survival after treat- Significant increases were seen for cancers of the stomach, ment is good because the conditions treated were generally pancreas, and lung; the average doses to the organs were not life-threatening, and there is minimal confounding from estimated to be 15, 13, and 1.7 Gy, respectively. For stom- concomitant treatment. ach cancer, a threefold increase in risk was observed in this study; the RR at 1 Gy was estimated to be 1.15, and the Benign Breast Disease absolute risk was 4.19 per 104 PY per gray. The estimated RR of lung cancer was 1.66 at 1 Gy. A U.S. cohort of 601 women treated with radiotherapy In an updated follow-up of this cohort up until 1997 (av- for acute postpartum mastitis and 1239 women treated by erage follow-up 25 years), Carr and colleagues (2002) also other means between 1940 and 1957 was followed for reported significant exposure-related increases in the risk of 29 years. The average dose to the breasts was 3.8 Gy. A sig- cancers of the stomach, pancreas, and lung among 1859 pa- nificant increase in the risk of breast cancer was seen among tients treated with radiotherapy. For stomach cancer, the women who had received radiotherapy, based on 51 exposed ERR was estimated to be 0.20 Gy–1 (95% CI 0, 0.73), based breasts with cancer. Using a linear multiplicative model the on analyses restricted to subjects who had received doses to risk increased by 0.4% per rad (ERR per Gy 0.4; 90% CI the stomach of 10 Gy or less (mean dose to the stomach 0.2, 0.7). A dose-response curve that appeared to be essen- 8.9 Gy; number of exposed cases 11). The corresponding tially linear up to about 7 Gy was demonstrated, and an in- estimate for cancer of the pancreas was 0.34 Gy–1 (95% CI creased risk for breast cancer was observed based on 56 cases 0.09, 0.89), with a mean dose of 8.2 Gy and 14 exposed can- (Shore and others 1986). cer cases. For lung cancer, the ERR was estimated to be A Swedish cohort of 1216 women treated for benign 0.43 Gy–1 (95% CI 0.12, 1.35) among subjects in the lowest- breast disease with radiotherapy and 1874 women treated by dose quartile (<1.4 Gy—mean dose 1.1 Gy), based on 21 other means from 1925 to 1954 was followed for an average deaths from lung cancer. Although the risk of pancreatic can- of 27 years for development of a subsequent cancer. Mean cer decreased with increasing age at exposure, no associa- absorbed doses to the breast were determined from detailed tion with age at exposure was observed for stomach and lung radiotherapy records and experiments with phantoms. The cancer. average dose to the breast was 5.8 Gy (range 0–50): 278 cases of breast cancer were diagnosed; 183 of these cases Benign Gynecological Diseases had received radiotherapy. A significant linear dose-response relationship was seen, with a downturn at approximately A U.S. cohort of 4153 women treated with intrauterine 10 Gy and higher. The estimated ERR for breast cancer was 226Ra between 1925 and 1965 for uterine bleeding disorders 1.63 Gy–1 (95% CI 0.77, 2.89, based on 47 exposed cases) was followed for an average of 27 years up to 1983 (Inskip among subjects with less than 3 Gy and 1.31 Gy–1 (95% CI and others 1990b). Individual organ doses were estimated 0.79, 2.04, based on 75 exposed cases) among subjects with based on detailed radiotherapy records and simulation of less than 5 Gy (Mattsson and others 1995). pelvic irradiation treatments on phantoms. A significantly Mattsson and colleagues (1997) also studied the risk of increased standardized mortality ratio (SMR)7 for death from malignancies other than breast cancer. Average doses were all cancers was seen in this population compared to the gen- estimated to 14 organs. A significant increase in the risk of eral population. In addition, significant increases were ob- all cancers combined (excluding breast) was observed. A served for deaths from colon and uterine cancer, cancers of significant linear dose-response was seen for stomach can- the female genital organs, and leukemia. Estimated ERR per cer: ERR per Gy 1.3 (95% CI 0.0, 4.4), based on 14 exposed Gy were 0.006 (90% CI –0.01, 0.05) for cancer of the uterus, cases and a mean dose to the stomach of 0.66 Gy (range 0– 0.41 (90% CI –0.69, 1.51) for other genital organs, 0.51 (90% 5.4). No significantly increased risk was seen for any other CI –0.08, 5.61) for colon cancer, and 0.20 (90% CI 0.08, cancer site, including leukemia, based on a small number of 0.35) for bladder cancer. exposed cases (Mattsson and others 1997). The estimated Inskip and colleagues (1990a) studied the risk of leuke- ERR for lung cancer was 0.38 (95% CI <0, 0.6), based on 10 mia in relation to radiation dose among 4483 of these exposed cases and a mean lung dose of 0.75 (range 0–9.0). women. Individual doses to various sections of the red bone marrow were calculated from detailed radiotherapy records. Peptic Ulcer The median dose to red bone marrow was 0.53 Gy. A sig- nificant excess of leukemia was observed; the risk was high- Cancer mortality up to 1985 was studied in a U.S. cohort est 2–5 years after treatment (SMR 8.1; 95% CI 2.6, 18.8, consisting of 1831 patients irradiated between 1937 and 1965 for the treatment of peptic ulcer and 1778 who were not 7SMR is the ratio of the mortality rate from a disease in the population (Griem and others 1994). An elevated risk of circulatory dis- being studied divided by the comparable rate in a standard population. Of- ease mortality was observed among those who received ra- ten the ratio is multiplied by 100.

OCR for page 155
164 BEIR VII compared to the general population) and among women over served among irradiated women. The SIR for cancer was 55 years at irradiation (SMR 5.8; 95% CI 2.5, 11.3). The 1.22 among irradiated women and 1.09 among nonirradiated. average ERR in this study was 0.19 Gy–1 (95% CI 0.08, 0.32) A significant increase in the SIR for cancers at heavily irra- for intrauterine 226Ra exposure, and the average absolute diated sites in the pelvic area was only observed 30 years or excess mortality from leukemia was 2.6 per 104 PY per gray. more after irradiation. A decreased risk for breast cancer was The risk of leukemia, lymphoma, and multiple myeloma also observed in this cohort, except for women treated at the was studied in an expanded cohort of 12,955 women treated age of 50 or more. No estimate of risk per unit dose was for benign gynecological disorders at one of 17 hospitals in presented. Massachusetts, Connecticut, Rhode Island, or New York State between 1925 and 1965 (Inskip and others 1993). Of Hormonal Infertility these women, 9770 were treated with radiation (either intra- cavitary 226Ra or external beam X-rays), while the rest were A U.S. cohort of 816 women who received X-ray therapy treated by other methods. The average age at treatment was to the ovaries and/or pituitary gland for refractory hormonal 46.5 years, and the average dose to active bone marrow in infertility and amenorrhea between 1925 and 1961 was fol- exposed women was 1.2 Gy. The RR for all cancers of he- lowed up until the end of 1990 (Ron and others 1994). The matopoietic and lymphatic tissue was 1.3 (95% CI 1.2, 1.5) average duration of follow-up was 35 years. Individual or- for irradiated women, compared to nonirradiated. The risk gan doses were estimated from radiotherapy records. Aver- of lymphomas, multiple myeloma, and nonacute lympho- age doses were 0.01 Gy to the breast, 0.9 Gy to the ovary, cytic leukemia was similar between irradiated and non- and 1.0 Gy to the sigmoid colon. Seventy-eight deaths from irradiated women. The RRs for acute lymphocytic leukemia cancer occurred in this cohort. No increase in mortality rates and for myeloid leukemia were elevated, however: RR 3.7 was found for leukemia or sites directly exposed to radia- (95% CI 1.3, 16) and 3.7 (95% CI 0.9, 36), respectively. For tion, such as the ovary or brain, based on a very small num- acute lymphocytic and nonlymphocytic leukemia, the SMRs ber of deaths (two leukemia, three ovary, and one brain can- were similar for women treated with radium only and with cer death). The SMRs were significantly elevated, however, both modalities, and were smallest for X-ray patients (differ- for cancer of the colon (15 deaths) and for NHL (6 deaths). ence not statistically significant). The ERR per Gy was 0.29 No estimate of risk per unit dose was presented. overall; 0.37 Gy–1 (95% CI <0, 1.5) for treatment with ra- dium only; 0.05 per Gy (95% CI –0.06, 0.33) for X-rays Ankylosing Spondylitis only; and ERR 0.21 per Gy (95% CI 0.05, 0.83) for the com- bination of both modalities. Average doses for the different A U.K. cohort consisting of 15,577 patients diagnosed treatment types were 0.6, 2.3, and 2.0 Gy, respectively, indi- with ankylosing spondylitis between 1935 and 1957 was cating a complex dose-response relationship. followed for mortality up to the end of 1991 (Weiss and A cohort of 2067 women who received radiotherapy for others 1994). The average duration of follow-up was metropathia hemorrhagica (uterine bleeding disorders) in 25 years. Of these subjects, 14,566 had received X-ray treat- Scotland between 1940 and 1960 was followed until the end ment for their disease. Radiation doses to various organs of 1990 (Darby and others 1994). The average follow-up were calculated for a sample of patients, and average esti- was 28 years. Absorbed doses to the active bone marrow and mated doses from all treatment courses occurring within to 20 solid organs or anatomical sites were estimated from 5 years of the initial treatment courses were attributed to all treatment records. Overall, 331 deaths from cancer were patients. The mean total body dose was estimated to be observed, and significantly elevated SMRs were observed 2.6 Gy. Irradiated patients had a significantly greater mor- for cancers at heavily irradiated sites (average local dose tality rate from cancer than expected from the national rates >1 Gy): cancer of pelvic sites, particularly urinary bladder for England and Wales, and significant increases were seen cancer (mean dose 5.2 Gy); colon cancer (mean dose for leukemia, NHL, multiple myeloma, and cancers of the 3.2 Gy); leukemia, and multiple myeloma (mean total active esophagus, colon, pancreas, lung, bones, connective and soft bone marrow dose 1.3 Gy). A deficit of breast cancer mor- tissue, prostate, bladder, and kidney. A linear dose-response tality was also observed in this cohort, due mainly to a large model for all cancers except leukemia gave an ERR of deficit in women who had received doses to the ovary of 0.18 Gy–1 (95% CI 0.10, 0.27) 5–24 years after treatment 5 Gy or more. No estimate of risk per unit dose was pre- (based on 741 deaths), decreasing significantly to 0.11 Gy–1 sented. 25 years or more (based on 845 deaths) after treatment. For A Swedish cohort study included 2007 women treated for lung cancer, the ERR was 0.09 Gy –1 (95% CI 0.03, 0.15) metropathia hemorrhagica between 1912 and 1977. Of these, 5–24 years after treatment, based on 282 deaths and an aver- 788 received radiotherapy for this condition. The population age dose to the bronchi of 8.88 Gy. For stomach cancer, the was followed up for cancer mortality and incidence from ERR was –0.004 Gy–1 (95% CI –0.05, 0.05) 5–24 years after 1958 to 1982, with a mean follow-up period of 28 years treatment, based on 127 deaths and an average stomach dose (Ryberg and others 1990). A total of 107 cancers were ob- of 3.21 Gy. There was no increased risk in breast cancer in

OCR for page 155
MEDICAL RADIATION STUDIES 165 this population, based on 84 deaths (ERR 0.08 Gy –1; with 131I. The follow-up was active, with an average dura- 95% CI – 0.30, 0.65); this may result from the fact that the tion of 8 years. No excess of leukemia or thyroid cancer was average dose the ovaries was high—5.5 Gy). observed among patients treated with 131I. The risk of leukemia mortality in this cohort was studied In a follow-up to this study, Hoffman (1984) studied can- further by Weiss and colleagues (1995), using a case- cer risk up to 1979 in the subgroup of 3696 women who had subcohort approach. A total of 60 leukemia deaths were ob- been treated at the Mayo Clinic, one of the original partici- served during the follow-up period. Radiotherapy records pating centers. Among these, 1005 had received 131I therapy were obtained for all but six of the deaths from leukemia alone and 2141 had been treated with surgery alone. A total excluding CLL, and individual doses to the red bone marrow of 527 cancer cases were identified in these two study were estimated as in the previous study; estimated doses groups; 175 were excluded because they occurred within a were also available for the subcohort, as described in Weiss year of treatment. The mean observation period was 15 years and colleagues (1994). The average dose to the total red bone for patients treated with 131I. The average whole-body dose marrow was estimated to be 4.44 Gy, but doses were non- is estimated to be of the order of 0.06–0.4 Gy in this cohort. uniform, with the heaviest dose to the lower spine. A linear- There was no increased cancer risk among those treated with exponential model (in which the exponential term allows for 131I and no indication of a relation with 131I activity deliv- cell sterilization in heavily exposed parts of the bone mar- ered. Nonsignificant increased risks were seen for cancers in row), varying with time since exposure, provided a good the two most exposed organs (thyroid and salivary glands, description of the risk for non-CLL. The estimated ERR per based on three and two cases, respectively). Gy was 12.4 (95% CI 2.3, 52.1) 10 years after exposure; 1– Goldman and colleagues (1988) reported on an extended 25 years after exposure, the average ERR per Gy was esti- follow-up of 1762 women, included in the Cooperative Thy- mated to be 7.0, based on 35 cases. rotoxicosis Therapy Follow-up Study, who were treated at A Swedish cohort of 20,024 patients who received X-ray the Massachusetts General Hospital between 1946 and 1964. therapy between 1950 and 1964 for painful benign condi- A total of 1406 had been treated with 131I. No dose estima- tions of the locomotor system (including arthrosis and tion was conducted. The average follow-up duration was spondylosis) was followed for cancer incidence and mortal- 17.2 years. An elevated SMR was noted in this cohort (SMR ity until the end of 1988 (Damber and others 1995). The 1.3; 95% CI 1.2, 1.4) for all causes of death but not for all average length of follow-up was 25 years. Average conver- cancers (SMR 0.9; 95% CI 0.7, 1.1). A nonsignificantly in- sion factors between surface dose and mean absorbed dose creased SMR was noted for breast cancer (SMR 1.2; 95% CI in the red bone marrow were estimated by treatment site (for 0.9, 1.5); no association with 131I activity was found. six sites), based on the treatment records of random samples Ron and colleagues (1998a) reported on mortality to the of 30 subjects drawn from the cohort (Damber and others end of 1990 in the Cooperative Thyrotoxicosis Therapy Fol- 1995). The conversion factors were applied to the entire co- low-up Study. The cohort included 35,593 hyperthyroid pa- hort and used for stratification of subjects in different levels tients, 91% of whom had been diagnosed with Grave’s dis- of exposure. The average absorbed dose to the red bone ease. Fewer than 500 subjects were less than 15 years of age marrow was estimated to be 0.39 Gy. A total of 116 leuke- at the time of treatment. The mean length of follow-up was mia cases (115 deaths) were observed during the study pe- 21 years, and 51% of the subjects had died during the study riod. The SIR and SMR for subjects with mean absorbed period. Doses from 131I to 17 organs (other than the thyroid) doses of 0.5 Gy or more were 1.40 (95% CI 1.00, 1.92) and were estimated for each study subject by multiplying the 1.50 (95% CI 1.05, 2.04), respectively. No estimate of risk amount of administered activity by the age-specific dose fac- per unit dose was presented. tor and 24-h thyroid uptake provided for each organ by the International Commission on Radiological Protection (ICRP 1988). Treatment with 131I was not related to all cancer mor- Thyroid Diseases tality (SMR 1.02; 95% CI 0.98, 1.07) or to mortality from Iodine-131 is currently the treatment of choice for hyper- any specific cancer, with the exception of thyroid cancer thyroidism, largely because no serious side effects are (SMR 3.94; 95% CI 2.52, 5.86, based on 27 cases). A non- known. Concerns remain, however, about the subsequent significant increase in mortality from thyroid cancer was risk of cancer. Several studies of patients treated with 131I for seen with increasing 131I administered activity—when deaths hyperthyroidism have been carried out in the United States, occurring in the first 5 years after treatment were excluded, Sweden, and the United Kingdom. there was no evidence of a relationship with total activity; it The occurrence of leukemia and of thyroid neoplasms is therefore likely that the underlying thyroid disease played (both benign and malignant) was studied among 36,050 pa- a role in the observed cancer increase. tients treated for hyperthyroidism between 1946 and 1968 Cancer incidence was also studied in 4557 patients who and included in the Cooperative Thyrotoxicosis Therapy received 131I therapy for hyperthyroidism in Sweden between Follow-up Study (Saenger and others 1968; Dobyns and oth- 1950 and 1975 at Radiumhemmet, Sweden (Holm 1984). ers 1974). Approximately 20,000 subjects had been treated Information on thyroid disease and treatment was abstracted

OCR for page 155
178 BEIR VII FIGURE 7-2 Distributions of study-specific estimates of ERR and EAR for breast cancer according to level of average dose to the breast.

OCR for page 155
MEDICAL RADIATION STUDIES 179

OCR for page 155
180 BEIR VII FIGURE 7-3 Distribution of study-specific estimates of ERR/Gy for breast cancer according to average age at exposure. Protraction of low-dose-rate exposure in the hemangioma pending on attained age in the ERR model and on both age at cohort may account for the reduced risk following exposures exposure and attained age in the EAR model. The excess in infancy in this cohort, although analyses within this study rates appeared to be similar in these cohorts, with a com- do not indicate a significant association. bined EAR estimate of 9.9 per 104 PY per gray (95% CI 7.1, Preston and colleagues (2002b) carried out a pooled 1.4) at age 50, suggesting similarity of risks following acute analysis of eight cohorts to estimate radiation-induced breast and fractionated low-dose-rate exposure. The ERR/Gy was cancer risk and evaluate the role of modifying factors. The greater among Japanese atomic bomb survivors; this differ- analyses included studies of the following populations: Japa- ence may be partly attributed to the lower background rates nese atomic bomb survivors (Thompson and others 1994), of breast cancer in Japan. the original and extended Massachusetts TB fluoroscopy The excess rates were higher for the mastitis and benign cohorts (Boice and others 1991b), the New York acute post breast disease cohorts with EAR estimates of 15 (95% CI partum mastitis cohort (Shore and others 1986), the Roches- 7.7, 24) and 32 (95% CI 21, 47) per 104 PY per gray, respec- ter infant thymic irradiation cohort (Hildreth and others tively, suggesting that women with some benign breast con- 1989), the Swedish benign breast disease cohort (Mattsson ditions may be at an elevated risk of radiation-induced breast and others 1993), and the Gothenburg and Stockholm skin cancer. hemangioma cohorts (Lindberg and others 1995). The analy- The hemangioma cohorts showed lower risks (EAR: 5.1 ses included 1502 breast cancer cases among 77,527 women, per 104 PY per gray; 95% CI 1.3, 11), suggesting a reduction about half of whom were exposed to radiation, with 1.8 mil- of risks following protracted low-dose-rate exposures. lion person-years of follow-up. No simple unified summary model adequately described the excess risk in all of these Thyroid Cancer studies. The excess risks for the thymus, tuberculosis, and atomic Thyroid cancer is one of the less common forms of can- bomb survivor cohorts showed similar temporal trends, de- cer. Its incidence is relatively high before age 40, it increases

OCR for page 155
MEDICAL RADIATION STUDIES 181 comparatively slowly with age, and it is about three times Ron and colleagues (1995a) conducted combined analy- higher in women than men. Ionizing radiation is a well-docu- ses of data from seven studies including five cohort stud- mented cause of thyroid cancer (UNSCEAR 2000b). ies—the atomic bomb survivors study, the Rochester thymic Of all the studies reviewed in the medical uses of radia- irradiation study (Shore and others 1993a), the Israeli tinea tion section above, only six provide dose-specific estimates capitis study (Ron and others 1989), and the Michael Reese of ERR and/or EAR. Table 7-4 and Figure 7-4 summarize and Boston enlarged tonsil studies (Pottern and others 1990; the results from these studies. In Figure 7-4, results are Schneider and others 1993)—as well as two case-control shown for all studies as well as restricted to studies in which studies of thyroid cancer nested respectively within the In- the average dose to the thyroid was less than 1 Gy. ternational Cervical Cancer Survivor Study (Boice and oth- All of the studies shown are studies of children who re- ers 1988) and the International Childhood Cancer Survivor ceived radiotherapy for benign conditions. All results relate Study (Tucker and others 1991). The analyses included a to thyroid cancer incidence. Because of the relatively good total of 707 cases, the majority of which (apart from the A- prognosis of most papillary thyroid cancers, studies of thy- bomb and cervical cancer survivors) were below age 15 at roid cancer mortality add little information about radiation time of exposure. risks. For subjects exposed below age 15, a linear dose-response In studies of external radiation exposure, the ERR/Gy was seen with a leveling or decrease in risk at the higher ranges from 3 Gy–1 in children exposed for enlarged tonsils doses used for cancer therapy. The pooled ERR was 7.7 Gy–1 and other benign head and neck disorders to 30 Gy–1 among (95% CI 2.1, 28.7), based on a random effects model that those exposed in Israel for the treatment of tinea capitis. took into account the heterogeneity of risk across studies. Similarly, the estimates of EAR vary from 0.9 per 104 PY The EAR was estimated to be 4.4 per 104 PY per gray (95% per gray in the hemangioma study to 13 in the tinea capitis CI 1.9, 10.1). Both of these estimates were significantly af- study. Although risk estimates from these studies vary con- fected by age at exposure, with a strong decrease in risk with siderably, the confidence intervals tend to be large; it is likely increasing age at exposure and little apparent risk for expo- that the estimates shown are statistically compatible. sures after age 20. The ERR appeared to decline over time TABLE 7-4 Risk Estimates for Cancer Incidence and Mortality from Studies of Radiation Exposure: Thyroid Cancer Average EAR/ Radiation Dose Dose Controls/ ERR/ 104 PY/ Reference Study Type (Gy) Range Cases Population Gy 95% CI Gy LB UB Comments Incidence Ron and Tinea capitis External 0.09 98 10,834 30.0 13.0 others (1989) X-ray Shore (1992) Meta-analysis of 131I 88 2 602 0.30 (0, 0.9)a 0.1 (0.0, 0.2)a hyperthyroidism studies Schneider and Benign head External 0.6 309 234 3.00 Overall others (1993) and neck X-ray 0.6 109 9.20 Before 1974 and screening program 0.6 200 1.80 After 1974 Shore and Enlarged External 1.36 0.03–10 37 2,657 9.00 (4, 24)a 2.9 (2.1, 3.9)a others (1993a) thymus X-ray Lundell and Hemangioma Mostly 1.07 <0.01–4.34 17 14,351 4.92 (1.26, 10.2) 0.9 others (1994) Ra Lindberg and Hemangioma 226Ra 0.12 15 11,807 7.50 (0.4, 18.1) 1.6 others (1995) Ron and Pooled analysis External 700 58,000 7.7 (2.1, 28.7) 4.4 (1.0, 10.1) others (1995a) of 7 studies X-ray NOTE: The number of cases and controls (or population size in cohort studies), as well as the mean dose and range, relate to exposed persons only. Empty cells indicate data not available from publication. LB = lower bound; UB = upper bound of CI. a90% CI.

OCR for page 155
182 BEIR VII FIGURE 7-4 Distribution of study-specific estimates of ERR/Gy for thyroid cancer according to level of average thyroid dose. NOTE: The estimate from populations exposed to 131I is excluded. about 30 years after exposure but was still elevated at is 0.3 Gy–1, lower than that from studies of external expo- 40 years. sures, but based on only two exposed cases. This study there- Three studies provided data on exposure protraction or fore provides little information about the risk of thyroid can- fractionation (thymus, tinea capitis, and Michael Reese); cer in relation to exposure to this nuclide. Studies of the analyses indicate that a small nonsignificant decrease in risk effects of 131I exposure later in life are reviewed in the pre- may be related to exposure fractionation in these studies. ceding section, although no dose-related estimate of risks A meta-analysis of hyperthyroidism studies provides a have been provided. These studies, taken together, provide risk estimate of thyroid cancer in relation to 131I exposure in little evidence of an increased risk of thyroid cancer follow- childhood (Shore 1992). The ERR estimate from that study ing 131I exposure after childhood.

OCR for page 155
MEDICAL RADIATION STUDIES 183 Leukemia Three studies have provided estimates of EAR per 104 PY per gray. Risk estimates between these studies are rela- Leukemia is one of the less common malignancies, but tively close, ranging from 1 to 2.6. substantial epidemiologic and experimental information ex- In most of the studies included here, the majority of sub- ists on the leukemogenic effects of ionizing radiation jects were adults at the time of exposure (with average ages (UNSCEAR 2000b). at exposure between 45 and 52 years in the uterine bleeding, Of all the studies reviewed in the “Medical Uses of Ra- benign breast disease, and cervical cancer survivor studies). diation” section, six provide dose-specific estimates of ERR Only the tinea capitis and hemangioma studies provide in- and/or EAR. Table 7-5 and Figure 7-5 summarize the results formation about exposures in childhood. In the hemangioma from these studies. In the figure, results are shown for all study—where all subjects were irradiated in infancy—the studies as well as restricted to studies in which the average overall ERR/Gy is similar to that seen in other studies; it is dose to the active bone marrow was less than 1 Gy. Results notable, however, that this is driven mainly by a higher ERR shown are for leukemia excluding CLL in all studies except for childhood leukemia; the ERR for adult leukemia in this the tinea capitis and uterine bleeding studies. study was very close to zero. In the tinea capitis study, in The ERRs/Gy shown in Table 7-5 range from 0.88 Gy–1 which all exposures were below age 15, no ERR is shown; in women who received an average dose to the active bone the EAR is similar to that seen in the other studies. marrow of 7 Gy from radiotherapy for cervical cancer to In one study (Inskip and others 1993), an effort was made 12.4 Gy–1 in subjects treated for ankylosing spondylitis (av- to estimate separately the effects of external exposures, erage dose 4.4 Gy). All other estimates, from studies with 226Ra, and the combination of the two. Estimates of risk from average doses ranging from 0.1 to 2 Gy, are relatively close, 226Ra alone or in combination with external radiation are in the range 1.9 to 5 Gy–1, and are statistically compatible. TABLE 7-5 Risk Estimates for Cancer Incidence and Mortality from Studies of Radiation Exposure: Leukemia Excluding CLL Average ERR EAR/ Radiation Dose Dose Controls/ per 104 PY/ Reference Study Type (Gy) Range Cases Population Gy 95% CI Gy LB UB Comments Incidence Boice and Cervix External 7 143 745 0.88 (SE: 0.69) others (1985) X-rays + intracavitary 226Ra Ron and Tinea capitis External 0.3 14a 10,834 0.9 others (1988b) X-ray Inskip and Uterine 226Ra 0.53 34a 4,483 1.90 (0.8, 3.2) 2.6 others (1990b) bleeding Inskip and Benign Overall 39 8,352 2.90 1.2 others (1993) gynecological Rad + ext 2.03 9 1,437 2.10 (0.5, 8.3) 1.0 (0.3, 1.9) disease 226Ra 2.31 26 5,508 3.70 (–1, 15) 1.5 (0.3, 2.9) External 0.59 4 1,407 0.50 (–0.6, 3.3) 0.1 (–0.2, 0.6) Lundell and Hemangioma Mostly Ra 0.13 <0.01–4.6 20 14,624 1.60 (–0.6, 5.5) Overall Holm (1996) 0.13 <0.01–4.6 9 14,624 5.01 (0.1, 15) Childhood leukemia only 0.13 <0.01–4.6 11 14,624 –0.02 (–0.8, 1.9) Adult leukemia only Mortality Weiss and Ankylosing External 4.38 1.27–6.99b 35 1,745c 12.4 (2.25, 52.1) others (1995) spondylitis X-ray NOTE: The number of cases and controls (or population size in cohort studies) as well as the mean dose and range relate to exposed persons only. Empty cells indicate data not available from publication. LB = lower bound; UB = upper bound of CI. aAllforms of leukemia combined. b10–90% range. cSubcohort with reconstructed doses.

OCR for page 155
184 BEIR VII — FIGURE 7-5 Distribution of study-specific estimates of ERR/Gy for leukemia according to level of average dose to the active bone marrow.

OCR for page 155
MEDICAL RADIATION STUDIES 185 similar; the ERR/Gy for external exposures appears to be Radiation and Circulatory Diseases lower, but this result is based on only four cases exposed. Although radiation exposure is well established as a risk factor for cancer, a clear understanding of the relationship Stomach Cancer between radiation exposure and other diseases is lacking. It has been postulated that the cardiovascular system is resis- Incidence rates for stomach cancer vary considerably tant to radiation-induced injury (Stewart and others 1995). throughout the world, with particularly high rates in Japan. However, it appears that tissue damage may occur as a result Many countries have seen decreases in incidence and mor- of both therapeutic (Stewart and Fajardo 1984) and A-bomb tality over the past 50 years or so, believed in large part to be radiation exposure (Villeneuve and Morrison 1997; Shimizu due to healthier diets with increased fruits and vegetables and others 1999). Capillaries represent the most radiosensi- and less salt. tive component of the cardiovascular system, with charac- Of all the studies reviewed on medical uses of radiation, teristic changes including detachment of endothelial cells five provide dose-specific estimates of ERR and/or EAR. and thrombosis. Arterial changes resulting from radiation Table 7-6 and Figure 7-6 summarize the results from these exposure depend on vessel size, with small and medium- studies. In the figure, results are shown for all studies as well sized arteries undergoing changes in all vessel layers, and as restricted to studies in which the average dose to the ac- large arteries appearing to be relatively radioresistant, al- tive bone marrow was less than 1 Gy. though radiation exposure may predispose larger vessels to Among the studies of populations with external radiation the development of atherosclerosis (Louis and others 1974). exposure and/or 226Ra, the estimates of ERR/Gy range from Radiation exposure has also been implicated in the devel- negative (in the hemangioma study) to 1.3 Gy–1 in the study opment of cerebrovascular injury (O’Connor and Mayberg of benign breast disease. The confidence intervals are wide, 2000). Specific conditions postulated to arise from irradia- and they all overlap, indicating that these estimates are sta- tion include vasculopathy, intracranial aneurysm formation, tistically compatible. An ERR of 1.32 Gy–1 (not significantly cerebral radiation necrosis, intracranial atherosclerosis, and different from zero) was seen among patients treated for hy- stroke (Trivedi and Hannan 2004). perthyroidism with 131I. Both animal and human studies have identified intimal thickening, lipid deposition, and adventitial fibroses of the TABLE 7-6 Risk Estimates for Cancer Incidence and Mortality from Studies of Radiation Exposure: Stomach Cancer Average EAR/ Radiation Dose Dose Controls/ ERR/ 104 PY/ Reference Study Type (Gy) Range Cases Population Gy 95% CI Gy LB UB Comments Incidence Boice and Cervix External 2 0.5–3.5 348 658 0.54 (0.05, 1.5)a 3.2 (0.1, 10.4) others (1989) X-rays + intracavitary 226Ra Holm and Hyperthyroidism 131I 0.07 29 1.3 9.6 others (1991) Lundell and Hemangioma Mostly Ra 0.09 5 14,351 <0 <0 Holm (1995) Mattsson and Benign breast External 0.66 0–5.4 14 1,216 1.3 (0, 4.4) others (1997) disease Weiss and Ankylosing External 3.2 0.52–5.8 127 1,745b –0.004 (–0.05, 0.05) others (1994) spondylitis X-ray Carr and Peptic ulcer External 8.9 11 1,859 0.20 (0, 0.73) Among others (2002) subjects with stomach dose 10 Gy NOTE: The number of cases and controls (or population size in cohort studies) as well as the mean dose and range relate to exposed persons only. Empty cells indicate data not available from publication. LB = lower bound; UB = upper bound of CI. a90% confidence interval. bSubcohort with reconstructed doses.

OCR for page 155
186 BEIR VII FIGURE 7-6 Distribution of study-specific estimates of ERR/Gy for stomach cancer according to level of average dose to the stomach. vascular system following irradiation. These changes are cardiovascular diseases, with radiation being only one of associated with atherosclerosis and the normal aging pro- many possible risk factors that may act directly or indirectly cess, although irradiation may accelerate the development of on the vasculature. To clarify the role of radiation in the these conditions (Trivedi and Hannan 2004). etiology of cardiovascular diseases, further studies involv- Although the dose required to produce specific conditions ing long-term, low-level exposures are needed, taking into or vascular effects is uncertain, it appears that over extended account all of the known risk factors for cardiovascular periods, the nature of the changes induced are similar for outcomes. low doses (on the order of 5 Gy) and for high doses (in the Excess heart disease mortality has been observed among region of 40 Gy). There is a broad spectrum and severity of women with breast cancer who were irradiated with cobalt-

OCR for page 155
MEDICAL RADIATION STUDIES 187 60 (Host and Loeb 1986, not reviewed here) and among per- as leukemia risk. Information that has become available since sons with HD who received mediastinal irradiation (Boivin 1989 has contributed to the examination of risks for these and Hutchison 1982; Hancock and others 1993b, 1993c). malignancies. Most affected patients had received at least 30 Gy to the A large number of studies involving radiation exposure mediastinum, although some had received less (Trivedi and for medical reasons have been described and discussed. Al- Hannan 2004). though these studies of medically exposed cohorts have in- creased our general knowledge of radiation risks, not all of them contribute substantially to quantitative risk assessment. DISCUSSION Many studies lack the sample size and high-quality dosim- Since the publication of BEIR V (NRC 1990), new infor- etry that are necessary for precise estimation of risk as a mation concerning health effects of radiation exposures has function of dose, a point that is illustrated by the large confi- become available from epidemiologic studies of populations dence intervals for many of the risk estimates shown in exposed to medical uses of ionizing radiation. The longer Tables 7-2 to 7-6 and by the limited number of studies for follow-up periods in recent reports have increased the statis- which risk estimates per gray are available. tical power in examining dose-response relationships at the Nevertheless, studies of populations exposed to therapeu- doses used for medical purposes. tic and diagnostic radiation provide information on issues Available studies of the effects of radiotherapy for malig- that cannot be addressed with atomic bomb survivor data nant or benign diseases confirm the presence of a heightened alone. Some examples are the evaluation of risk in Cauca- risk of development of a number of primary or second pri- sian populations where baseline cancer and other disease mary cancers on follow-up. Because the doses in most series risks may be very different from those in a Japanese cohort. far exceed 100 mGy to the site of interest, they provide lim- Also, studies of medically exposed cohorts allow for the ited direct quantitative information on the risk of low-level evaluation of risk from protracted exposures. In addition, radiation, particularly when they involve large doses where studies of medical uses of radiation have been important in cell killing may lead to underestimation of the risk per unit establishing the lack of radiation risk for CLL, since this dose. These studies provided valuable information for the cancer is very rare in Japan. study of risk modifiers, including age at exposure, attained Often there is interest in comparing results from different age, and possible differences in patterns of risk across studies to gain information on the modifying effects of fac- countries. tors such as baseline risks and protraction of exposure that Analyses that are restricted to populations with low doses may differ among the studies. It should be kept in mind that are complicated by the limitations of statistical variability as such comparisons can be difficult to interpret since there are well as by limitations of sample size and study design, in- nearly always several differences among the cohorts being cluding dose reconstruction. Limitations also include chance, compared. As an illustration, the ERR/Gy for breast cancer small undetected biases, and the consequences of doing mul- in the Life Span Study cohort has been found to be higher tiple tests of statistical significance. Indeed, among diagnos- than the ERR/Gy in tuberculosis fluoroscopy patients (Howe tic radiation studies, only studies of repeated chest fluoros- and McLaughlin 1996; Little and Boice 1999; Preston and copy and scoliosis examinations are informative concerning others 2002a). However, it is not clear whether this differ- the magnitude of ERR and EAR as a function of dose. It ence occurs because of the higher baseline risks in the must be noted that although the dose rates in these studies Caucasian fluoroscopy cohorts, the lower dose rate in these are low, the cumulative doses received by tuberculosis pa- patients, the lower energy of the X-ray exposure used in tients are high, and even scoliosis patients followed radio- fluoroscopy (Brenner 1999), or some combination of these logically for spine curvature received average cumulative factors. doses of the order of 100 mGy or more. For lung cancer, the ERR/Gy from the studies of acute, Most of the information on radiation risks therefore still high-dose-rate exposures are statistically compatible and in comes from studies of populations with medium to high the range 0.1–0.4 Gy–1. It is difficult to evaluate the effects doses, with the notable exceptions of childhood cancer risk of age at exposure or of exposure protraction based on these following in utero exposures and thyroid cancer risk follow- studies because only one study (the hemangioma cohort) is ing childhood exposures, for which significant increases available in which exposure occurred at very young ages and have been shown consistently in the low- to medium-dose in which protracted low-dose-rate exposures were received. range. The study of tuberculosis patients appears to indicate that substantial fractionation of exposure leads to a reduction of risk. SUMMARY For breast cancer, EARs appears to be similar (of the or- In this chapter on medical radiation, particular attention der of 9.9 per 104 PY per gray at age 50) following acute and has been paid to estimating the risk of cancer at specific fractionated exposures to moderate- to high-dose-rate radia- sites—namely, the lung, breast, thyroid, and stomach, as well tion. Effects of attained age and age at exposure are impor-

OCR for page 155
188 BEIR VII tant modifiers of risk. The excess risks appear to be higher in For leukemia, ERR estimates from studies with average populations of women treated for benign breast conditions, doses ranging from 0.1 to 2 Gy are relatively close, in the suggesting that these women may be at an elevated risk of range 1.9 to 5 Gy–1, and are statistically compatible. Esti- radiation-induced breast cancer. The hemangioma cohorts mates of EAR are also similar across studies, ranging from 1 showed lower risks, suggesting a possible reduction of risks to 2.6 per 104 PY per gray. Little information is available on following protracted low-dose-rate exposures. the effects of age at exposure or of exposure protraction. For thyroid cancer, all of the studies providing quantita- For stomach cancer, the estimates of ERR/Gy range from tive information about risks are studies of children who re- negative to 1.3 Gy–1. The confidence intervals are wide and ceived radiotherapy for benign conditions. A combined they all overlap, indicating that these estimates are statisti- analysis of data from some of these cohorts with data from cally compatible. the atomic bomb survivors and from two case-control stud- Finally, results of two studies of patients having under- ies of thyroid cancer nested within the International Cervical gone radiotherapy for HD or breast cancer suggest that there Cancer Survivor Study and the International Childhood Can- may be some risk of cardiovascular morbidity and mortality cer Survivor Study provides the most comprehensive infor- for very high doses and dose-rate exposures. The magnitude mation about thyroid cancer risks. For subjects exposed be- of the radiation risk and the shape of the dose-response curve low the age of 15, a linear dose-response was seen, with a for these outcomes, if an effect exists, are uncertain. leveling or decrease in risk at the higher doses used for can- In conclusion, studies of medically irradiated populations cer therapy. The pooled ERR was 7.7 Gy–1 and the EAR 4.4 provide information on the magnitude of risk estimates per 104 PY per gray. Both estimates were significantly af- (mainly in the medium- to high-dose range) and on the ef- fected by age at exposure, with a strong decrease in risk with fects of factors, such as exposure pattern and age at expo- increasing age at exposure and little apparent risk for expo- sure, that may modify risk. Further studies of medically sures after age 20. The ERR appeared to decline over time exposed populations are needed to study possible gene-ra- about 30 years after exposure but was still elevated at diation interactions that may render parts of the population 40 years. more sensitive to radiation-induced health effects. Studies of Little information on thyroid cancer risk in relation to 131I populations (particularly children and infants) with lower- exposure in childhood was available. Studies of the effects to medium-dose diagnostic exposures also are needed be- of 131I exposure later in life provide little evidence of an cause of the increasing use of procedures such as CT and increased risk of thyroid cancer following 131I exposure after radiological monitoring of infants. childhood.