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 86
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications 4 Implications for Clinical Practice and Public Health Policy Efforts to measure and understand the implications of the exposure of Americans to radioactive fallout from the atomic bomb tests at the Nevada Test Site in the 1950s raise a variety of clinical and public health issues including directions for further research, strategies for informing the public about risk, and policies for disease screening. This chapter focuses on thyroid cancer screening for people thought to be at higher-than-average risk for thyroid cancer from exposure to iodine-131 (I-131) fallout. It also briefly considers screening for other thyroid disorders. The final sections consider clinical and public health policies based on a review of the scientific literature on the accuracy, benefits, and harms of screening. In public health terms, screening for thyroid cancer is a form of secondary prevention.1 Like screening for other cancers, the goal is to detect disease in people without symptoms so that they can be treated early to reduce mortality and morbidity. Routine screening is directed at the population generally; targeted screening seeks people who have a higher-than-average risk of developing a disease. In either case, a screening program has value only when earlier detection of a disease results in earlier treatment that improves outcomes for the screened population. The appeal of screening as a means of reducing the burden of disease is powerful, and much has been claimed for a large array of tests that screen for 1 Primary prevention aims to eliminate or reduce health threats (e.g., by treating waste water) and to make people less susceptible to such threats (e.g., through vaccination). Tertiary prevention involves treatment and management of existing illness (e.g., through drug therapy for diagnosed heart disease) to reduce the threat of complications or progression.
OCR for page 87
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications diseases causing premature morbidity and mortality. The U.S. Preventive Services Task Force has, for example, examined and made recommendations about screening for 53 conditions, including heart disease, several kinds of cancer, infectious diseases, prenatal disorders, and sensory problems. As these and other recommendations and analyses make clear, clinical epidemiologic research confirms the value of some screening tests but does not support claims for others (USPSTF 1996; Russell 1994; Eddy 1991). Efforts to evaluate screening strategies and to develop evidence-based recommendations for screening can generate considerable controversy. Science-based conclusions (especially when the conclusion is that the evidence for screening is negative, inconclusive, or lacking) can conflict with the understandable public desire to believe that a particular screening test will save lives. A case in point is the controversy over a recommendation from an NCI consensus panel that screening for breast cancer in women ages 40 to 49 should be a matter for women to decide with their clinicians rather than a routinely advised practice (Begley 1997; Eddy 1997; Ransohoff and Harris 1997). Following protests and criticism from some advocacy groups and some members of Congress, a different NCI panel (the National Cancer Advisory Board) recommended routine breast cancer screening for this age group (Taubes 1997) even though many scientists think that evidence is still inadequate to support general screening in this age group. The discussion in this chapter builds on sections in other chapters of this report that have examined who is potentially at risk of thyroid cancer from I-131 exposure, how great the risk is, and how communication with the public should be structured. It also draws on the literature review and analyses presented in the background paper commissioned for this study (Appendix F). The discussion here also builds generally on the principles of evidence-based clinical practice and public health policy. It recapitulates some of the information on thyroid cancer presented earlier, so that this chapter can be read independently. This chapter reviews. The concepts and principles for screening recommendations. The burden of illness associated with thyroid cancer. The benefits and harms of screening. The tests used for screening. The evidence about test accuracy and the benefits of early detection. The screening recommendations of other groups. This study's conclusions and recommendations. PRINCIPLES FOR SCREENING RECOMMENDATIONS The specific conclusions about thyroid cancer screening for exposed persons were developed by the Institute of Medicine (IOM) committee that was described in Chapter 1. In developing its recommendations, the committee examined the
OCR for page 88
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications scientific literature (including the literature review by Eden, Helfand, and Mahon in Appendix F), considered position statements and guidelines developed by other groups and individuals, and benefited from the discussion of diverse experts at a March 17-18, 1998, workshop convened by the committee (see Appendix A for the agenda and participants). In addition, the committee reviewed information from the public meetings, analyses, and deliberations of the National Research Council committee that considered the I-131 thyroid doses from the Nevada tests, the cancer risk posed by I-131 exposure, and the approaches to communicating with the public about exposure and risk. Criteria for Clinical Recommendations In developing guidelines for thyroid cancer screening for people potentially exposed to I-131 fallout from the Nevada tests, the IOM committee began with several broad principles. Consistent with the established and recognized mission of the IOM and the National Research Council, the guidelines would be based on careful review and assessment of the scientific evidence. They would be intended to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances and to inform policy decisions about public health strategies. Recommendations set forth by the IOM and others (CDC 1996; USPSTF 1996; McCormick and others 1994; IOM 1992; Eddy 1991) have proposed a number of additional criteria for guidelines for clinical practice. First, the disease targeted for screening should be Important in terms of prevalence, incidence, and mortality or morbidity (disease burden). Amenable to treatment that produces better outcomes (benefits balanced against harms) than observation alone or no treatment at all. More successfully treatable (e.g., condition cured, progression slowed) if detected at an earlier stage than would be possible without screening (e.g., before symptoms are evident). Second, screening recommendations should be based on evidence related to the technical characteristics of a test used for screening. A screening test should be Reliable, accurate, and safe in detecting the disease earlier than would be possible under the conditions of usual care, as indicated by the test's sensitivity, specificity, and positive predictive value (see Addendum 4A for definitions and an illustration of the effect of disease prevalence on screening test performance). Projected to have benefits (e.g., improved life expectancy or quality of life) that outweigh harms (e.g., possible unnecessary diagnostic work-ups or surgery for false-positive results or nonprogressive cancers).
OCR for page 89
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications Third, for screening to be implemented successfully, a screening strategy should be Acceptable to the public (e.g., involve reasonable burdens or harms as perceived by those to be screened). Acceptable to clinicians (e.g., not unreasonably burdensome to undertake). Effective and feasible to implement in normal practice. Cost-effective (e.g., have a cost per life saved or quality-adjusted life-year gained that is comparable to or less than that provided by other accepted interventions). The last of these common criteria for screening—cost-effectiveness—was not considered by the committee in its analyses nor was such analysis specified in the charge to the IOM from the National Cancer Institute. Clinical evidence and considerations alone determined the committee's conclusions. The committee, however, noted that if evidence does not support claims that an intervention is clinically effective or that its benefits outweigh its harms, then the further step of cost-effectiveness analysis makes no sense. The committee's criteria for making recommendations about thyroid cancer screening for those exposed to I-131 from the Nevada tests can be depicted as an evidence pyramid (Figure 4.1). The lowest tier involves evidence of a population health problem; next are the availability of effective treatment for the disease and of accurate and feasible screening tests; a yet higher tier involves evidence that early detection through screening improves outcomes; and at the top of the pyramid is evidence that benefits exceed harms. Recommendations for screening apparently health populations are generally held to a higher standard of effectiveness than recommendations for treatment in people with evident disease or injury. To make a positive recommendation for screening for thyroid cancer in people exposed to I-131 from the Nevada nuclear tests, the committee determined in advance 1) that a chain of evidence was required that exposure to I-131 from fallout increases the risk of thyroid cancer; 2) that effective treatment is FIGURE 4.1 Pyramid of evidence for national screening policy.
OCR for page 90
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications available for the disease; 3) that screening tests for thyroid cancer are reliable, accurate, and practical in detecting disease earlier than would occur with usual care; 4) that early detection of thyroid cancer by screening improves treatment outcomes; and 5) that the potential benefits of screening outweigh its potential harms. The strength of either a positive or a negative recommendation would depend on the strength of the evidence on these four points. For example, if evidence suggested that screening was efficacious but that good results depended on the skill or experience of the clinicians performing the screening tests, then a recommendation might describe when referral to a more experienced clinician should be considered. A committee recommendation about an intensive public health program to encourage or pay for screening would have to note (as described above) that the committee did not analyze the cost-effectiveness of thyroid cancer screening compared with other screening strategies of demonstrated value. BURDEN OF ILLNESS General Burden of Mortality and Morbidity Chapter 3 presented data for thyroid cancer in the United States. To recapitulate in the context of this discussion of screening policy, thyroid cancer is uncommon and rarely life-threatening. The overall age-adjusted incidence for all forms of thyroid cancer is 4.9 per 100,000 population (unless otherwise indicated, data are from NCI 1997b), and it accounts for about 1 percent of cancers diagnosed each year and about 0.2 percent of cancer deaths (an estimated 1,200 out of 564,800 in 1998) (ACS 1998a). The majority of these cancers are papillary carcinomas, which is also the form of thyroid cancer linked to I-131 exposure. Naturally occurring thyroid cancer is considerably more common in women than it is in men, 6.9 cases per 100,000 for women versus 2.8 for men (NCI 1997b), and it is more frequent in whites than in blacks.2 For men, the incidence 2 These data come from SEER (NCI 1997b). SEER stands for the Surveillance, Epidemiology, and End Results Program, which is part of the National Cancer Institute. The SEER program was established following 1971 legislation directing NCI to collect, analyze and disseminate data useful in preventing, diagnosing, and treating cancer. The SEER data reported here were collected from 9 or 11 designated cancer registries. For all years, the registry reporting areas include Connecticut, Iowa, New Mexico, Utah, Hawaii, Detroit, San Francisco-Oakland, Atlanta, and Seattle-Puget Sound with data from Los Angeles and San Jose-Monterey included for the most recent period. These registries cover an estimated 9.5 (9 areas) or 13.9 percent (11 areas) of the U.S. population. The registries abstract records for all resident cancer patients seen in hospitals inside and outside the geographic area; search other records (e.g., private laboratories) to identify additional patients; abstract death certificates for residents with a cancer diagnosis listed regardless of site of death; and follow most identified living patients who have a cancer diagnosis. Information abstracted includes patient demographics and primary disease site, morphology, diagnostic confirmation, extent, and initial therapy.
OCR for page 91
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications of thyroid cancer increases fairly steadily until age 70; for women, incidence peaks in middle age. The median age at diagnosis for all forms of thyroid cancer is 43 years for women and 50 years for men. For those who die of all forms of thyroid cancer (a small proportion of those diagnosed), the median age at death is 75 years for women and 69 years for men. Survival rates for this disease are high. The 10-year cancer-specific survival rate for persons with papillary carcinoma, the form linked to radiation exposure, is estimated at 95 percent and the 30-year survival rate is estimated at 90 percent (Wang and Crapo 1997; Mazzaferri and Jhiang 1994). Women's survival rates are somewhat higher than rates for men (96 percent for women versus 92 percent for men at 5 years for all thyroid cancers combined). Survival rates are somewhat lower for blacks than for whites (around 88 percent at 5 years for all thyroid cancers combined). Age-adjusted mortality rates for all forms of thyroid cancer are 0.4 per 100,000 for women and 0.3 per 100,000 for men. (By way of comparison, the mortality rate for lung cancer is 32.8 and 73.2 per 100,000 for women and men respectively; for breast cancer in women, the mortality rate is 26.4 per 100,000 and for men, the mortality rate for prostate cancer is 26.9 per 100,000.) Although relatively low at all ages, mortality rates rise after about age 45, an age already reached or being approached by the cohort potentially at risk of thyroid cancer from exposure to radioactive fallout from the Nevada tests. Chapter 3 noted that thyroid cancer is among the cancers that have increased in incidence but decreased in mortality rate over the past 30 years. The contrast between the incidence and mortality trends has been attributed to more sophisticated detection technologies (ultrasound for nodules and FNA biopsy for cancer) and more complete diagnostic reporting (Wang and Crapo 1997). The disease accounts for about 1 percent of all cancers and about 0.2 percent of cancer deaths (an estimated 1,200 of 564,800 in 1998, according to the American Cancer Society 1998a). Mortality does not represent the only burden of thyroid cancer. Symptoms of the disease itself, the processes of diagnosis and treatment, and the consequences of and follow up after treatment also must be considered. Symptoms for treatable disease are relatively uncommon but can include hoarseness and difficulty swallowing or breathing. Diagnosis by fine-needle aspiration (FNA) biopsy can cause anxiety and minor physical discomfort, as described later in this chapter and can lead to overtreatment for small lesions. Surgical treatment for clinically significant papillary thyroid cancer is generally considered to provide effective treatment and improve long-term survival (Mazzaferri and Jhiang 1994; Mazzaferri 1993b; Hay 1990). Some controversy, however, remains about specific surgical strategy.3 Surgical treatment involves 3 See, for example, Cady and Rossi (1988), Shaha and others (1995) on the case for subtotal thyroidectomy and Mazzaferri (1996), Schlumberger (1998), and Demure and Clark (1990) on the case for near-total thyroidectomy.
OCR for page 92
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications inpatient surgery that typically removes nearly all of the thyroid gland (leaving a small part of the gland near the entrance of the laryngeal nerve into the larynx) (Falk 1997). Surgical scars are generally visible on the neck above shirt or blouse collars, which may bother some people and not others. If a thyroid cancer, other than a very small one is found, surgery is often followed by radiation treatment with I-131 to destroy remnant thyroid tissue. After total or near-total thyroidectomy, patients require lifetime thyroid hormone replacement therapy, which if properly prescribed and monitored, does not present significant risk of adverse side effects. In addition to a very small risk of complications from anesthesia, identified surgical risks include permanent damage to the laryngeal nerve (with resulting severe hoarseness) and inadvertent removal of the parathyroid glands (with medication required to treat subsequent problems related to hypocalcemia, which can include muscle tremors, cramps, and seizures). A recent review of seven studies, involving a total of 1,754 patients, reported permanent hypoparathyroidism rates of 0.8 to 5.4 percent (weighted mean, 2.6 percent) for those undergoing total thyroidectomy; for subtotal thyroidectomy, the weighted mean was 0.2 percent (Udelsman 1996). The review also reported permanent recurrent laryngeal nerve injury in 0 to 5.5 percent (weighted mean 3.0 percent) for total thyroidectomy patients and in 1.1 to 3.2 percent (weighted mean, 1.9 percent) of those undergoing the less extensive procedure. Although the committee is not aware of specific research on the relationship between thyroidectomy complication rates and surgical skill or experience, the technical difficulty of thyroid surgery, especially for total or near-total thyroidectomy, is such that lower complication rates might reasonably be expected from surgeons and centers that perform higher volumes of procedures.4 Patients who undergo surgery for large invasive tumors or who have extensive lateral lymph node dissections (i.e., standard radical neck dissection) combined with total thyroidectomy are at higher risk for complications, as are older patients who are likely to have comorbid conditions (Mazzaferri and others 1977). Exposure to I-131 and Risk of Thyroid Problems The NCI report links higher exposure to I-131 to a combination of geographic location, age at exposure, and milk consumption, especially milk from ''backyard" cows and, particularly, goats. As discussed in the NCI report and in Chapter 2 of this IOM/NRC report, the NCI's estimates of the American public's exposure to I-131 from the 1951-1962 nuclear tests in Nevada are characterized by 4 Chen and others (1996) discuss outcomes for parathyroidectomy specifically. General discussions of the procedure appear in Hughes and others (1987), Banta and others (1992), Grumbach and others (1995), Imperato and others (1996), Tu and Naylor (1996), and Wennberg and others (1998).
OCR for page 93
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications substantial uncertainty. This uncertainty, which is inevitable given data limitations related to geographic measurements of fallout and reconstruction of milk consumption information, applies to general characterizations of total population exposure but it creates the greatest problems for making individual estimates. Accurate estimates of a specific individual's exposure are usually not possible. Chapter 3 concluded that evidence supports a link between exposure to radioactive iodine from fallout and papillary thyroid cancer. The major uncertainty of this relationship would be at the lower exposure levels that characterized most Americans' exposure from the Nevada weapons testing program. For the average American woman, the probability of being diagnosed with thyroid cancer by the age of 95 is roughly 1 chance in 150 without exposure and 1 in 90 with exposure to 0.1 Gy (10 rad). For the average man, the corresponding probabilities are 1 in 400 without exposure and 1 in 150 with exposure. As noted earlier, papillary thyroid cancer, the form related to radiation exposure, is rarely life threatening. Nonmalignant Thyroid Disease Associated with I-131 Exposure In addition to thyroid cancer, Chapter 3 investigates whether evidence links I-131 exposure to other thyroid disorders, which include thyroid nodules, hypothyroidism, hyperthyroidism, and autoimmune thyroiditis and goiter. The review in Chapter 3 cites considerable evidence of links at moderate and high exposures but little evidence suggesting a link between hypothyroidism or hyperthyroidism and I-131 doses in the range experienced by most of those exposed to fallout from the Nevada nuclear tests. For example, a study of Utah "downwinders" exposed to atomic weapons test fallout after birth (mean dose of 0.098 Gy [9.8 rad] with a maximum of 4.6 Gy [460 rad]) showed no excess of nonmalignant thyroid disease (Kerber and others 1993). Given the conclusions in Chapter 3, the IOM committee did not investigate the disease burden of nonmalignant disease or the evidence base for screening for such disease in people exposed to I-131 from the Nevada nuclear tests. The results of the Hanford Thyroid Disease Study (see www.fhcrc.org/science/phs/htds) may clarify the extent to which childhood exposure to low doses of I-131 is linked to thyroid problems in adulthood. This clarification should, in turn, clarify whether it is warranted for DHHS to examine the evidence on screening for nonmalignant disease.5 THYROID CANCER SCREENING AND DIAGNOSTIC OPTIONS Screening for thyroid cancer may involve two steps. For the first step, screening for thyroid nodules, the options reviewed by the committee are physical palpation 5 As this report was nearing public release and after the committee had concluded its deliberations, the American College of Physicians published new recommendations for screening for benign thyroid disease (Helfand and others 1998; Helfand and Redfern 1998).
OCR for page 94
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications and ultrasound examinations of the neck (Charboneau and others 1998). If these tests detect a large nodule (1.5 cm or larger), the second step, FNA (fine needle aspirate) biopsy of the nodule, is usually performed. Observation (with periodic examinations) will be the approach for most of those who have small (less than <1.5 cm), otherwise unsuspicious lesions detected by palpation or ultrasound. Most such lesions, often discovered during ultrasound examinations for other conditions (e.g., carotid artery disease, parathyroid problems), are benign and even those that have cancer cells will usually not become life- or health-threatening. Some patients are referred directly for surgery following detection of a nodule, although this is not standard or recommended practice (AACE 1997a; Tan and Gharib 1997; Ezzat and others 1994; Mazzaferri 1993b). Physical palpation of the thyroid6 is performed in conjunction with a visual examination of the neck and a patient history that documents possible risk factors such as age, sex, family history of thyroid disease (especially papillary or medullary cancer), personal history of previous head or neck irradiation or thyroid problems, or symptoms such as hoarseness or difficulty swallowing (Gharib 1997; Belfiore and others 1995; Mazzaferri 1993b). Palpation seeks to detect nodules in the thyroid, and if they are found, to assess their size, number, firmness, and adherence to adjacent structures. The size and firmness of nearby lymph nodes are also checked by palpation. For asymptomatic, average-risk people, routine screening for thyroid cancer through palpation or ultrasound examination of the thyroid gland is not recommended (USPSTF 1996). Ultrasound examination of the thyroid involves noninvasive scanning of the thyroid gland with a machine that creates images by directing high-frequency sound—ultrasonic waves—to the gland and using computer algorithms to translate information about sound transmission or reflection (echoes) into two-dimensional images (Tan and others 1995; Ezzat and others 1994; Hsiao and Chang 1994). Ultrasound can detect many small nodules (less than 1 to 1.5 cm) that are not usually detectable by palpation and that will usually not progress to cause problems. Little is known about the natural history of these small nodules and the occasional cancers, including what causes a few of them to progress. Rather than further tests or therapy, such nodules generally warrant observation (Gharib and Mazzaferri 1998). Ultrasound examination also can provide additional information to further characterize nodules, including whether calcification is present and whether the nodule is solid or cystic (Watters and Ahuja 1992). Although such information—in combination with information from physical examination—can raise or lower the probability that a nodule is cancerous, ultrasound information is not by itself diagnostic. 6 This is a more focused examination of the thyroid than the physical examination of the neck that physicians may perform as part of a routine physical. The latter examination also includes palpating cervical nodes, listening for carotid bruits with the stethoscope, checking range of motion in the neck, feeling vertebral processes, and other elements.
OCR for page 95
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications To determine whether a detected thyroid nodule is benign or malignant, the current procedure of choice is FNA biopsy (Garcia-Mayor and others 1997; Gharib 1997; Lin and Huang 1997; Ashcraft and Van Herle 1981). Typically, a fine gauge needle is inserted into the nodule (or nodules) in several places and a sample drawn into the needle and an attached syringe. If the nodule is not readily palpated, the biopsy can be guided by ultrasound, although this technology is not available everywhere and can lead to the identification of additional, small nodules, few of which are likely to cause problems (Lin and Huang 1997; Taki and others 1997). Minor localized pain is common during and after the biopsy, and, rarely, a clinically apparent hematoma can occur at the FNA site (Gharib and Goellner 1993; Ashcraft and Van Herle 1981). Following the biopsy, the aspirated material (tissue and blood) is placed on slides or filtered through mesh and applied to slides, stained with various dyes, and then microscopically examined. Pathologists evaluate the sample on the slides and classify the lesion. In addition to positive or negative results, FNA samples can be categorized as indeterminate or unsatisfactory. For results categorized as indeterminate, the pathologic features of the cells are ambiguous and cannot be readily categorized as benign or malignant. Patients with indeterminate biopsies are usually referred to surgery. Of satisfactory samples, about 10 percent are likely to be classified as indeterminate. Unsatisfactory samples provide insufficient or inadequate material for evaluation and often prompt repeated biopsies aimed at securing an adequate sample for further analysis (LiVolsi 1997). Unsatisfactory FNA samples are obtained 10-15 percent of the time, even in the best-performing centers. The rate of unsatisfactory samples depends on the training and skill of the operator (e.g., endocrinologist, head-and-neck surgeon, radiologist, family practitioner), the number of sites sampled, preservation practices, the skill of the pathologist or cytopathologist, and the size of the nodule being biopsied (Merchant and Thomas 1995; Haas and Trujilla 1993; Piromalli and Martelli 1992). A screening approach that led to more FNA biopsies of smaller nodules (<1-1.5 cm) would likely generate more inadequate samples and, in turn, more repeat biopsies. Although no precise figures are available and recommendations vary, some patients, perhaps even a majority, with inadequate FNA samples will be referred for surgery. Addendum 4B considers the implications of indeterminate or unsatisfactory FNA biopsy results in more depth. ACCURACY OF SCREENING AND FOLLOW-UP TESTS Palpation and ultrasound examinations are widely accepted as safe, low-risk procedures (Gharib 1997; Mazzaferri 1993a; Ashcraft and Van Herle 1981). Concern about their use is not related to the direct risk of the procedures themselves but to their relative inaccuracy, particularly the probability of their producing false-positive results or identifying many very small nodules and cancers that are
OCR for page 96
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications not likely to cause harm. Because the likelihood of finding thyroid cancer (pretest prevalence) is low, even in most populations exposed to iodine-131, the great majority of positive test results for nodules will be false-positive test results for cancer when the nodule is biopsied. A detailed review of evidence about the accuracy of palpation and ultrasound in detecting thyroid nodules and of FNA in detecting thyroid cancer is presented in the "Screening for Thyroid Cancer" background paper (Appendix F). Palpation The standard for reporting sensitivity in the detection of thyroid nodules is ultrasound. Because ultrasound cannot definitively discriminate benign from malignant nodules, the test results in nodule—not cancer—detection. Research has not compared the accuracy of different approaches to palpation (e.g., whether it is best done from the patient's front or back) (Tan and Gharib 1997), but more experienced examiners are likely to produce more accurate results (Jarlov and others 1993; Jarlov and others 1991). Accuracy can be affected by the size, consistency, or location of nodules (e.g., on the front or back surface of the thyroid) and by a patient's physical characteristics (e.g., length and thickness of the neck). The sensitivity results from three studies of palpation ranged from 0.10 to 0.31 in detecting nodules identifiable with ultrasound (Table 1 in Appendix F). The study that reported 0.31 overall sensitivity found sensitivity of 0.45 for detection of nodules larger than 0.5 cm and 0.84 for nodules larger than 1 cm (Ezzat and others 1994). A study of palpation in a radiation-exposed population reported sensitivity of 0.89 for detection of nodules larger than 2 cm and 0.83 for those between 1 and 2 cm (Mettler and Williamson 1992). Other studies of the sensitivity of palpation for detecting larger nodules gave poorer results. For example, one study reported sensitivity of only 0.42 for nodules larger than 2 cm (Brander and others 1992). The U.S. Preventive Services Task Force (USPSTF 1996) reports specificity levels for nodule detection of 0.93 to 1.0 from three studies that compare palpation with ultrasound. Ultrasound Ultrasound examination is the reference standard for nodule detection, so its sensitivity and specificity for this purpose are not reported. It very commonly reveals thyroid nodules in people without symptoms who are being examined for other purposes. For example, four U.S. studies of ultrasound examinations report nodules in 13 percent (Carroll 1982), 41 percent (Horlocker and Jay 1985), 50 percent (Stark and others 1983), and 67 percent (Ezzat and others 1994) of those examined. Current ultrasound technology does not allow reliable and accurate discrimination between cancerous and noncancerous nodules. When ultrasound results are compared against FNA results for specificity in detecting cancer (rather than nodules), studies report a high rate of false positives,
OCR for page 114
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications and 1 in 240 for men. This is less than many other cancers in the general population. Thyroid cancer is rarely life-threatening; 30-year survival is over 90 percent for papillary thyroid cancer, the form linked to radiation exposure. Accurately identifying people's past exposure to I-131 is usually not possible because necessary data on the key risk factors from four decades ago are generally not available or are unreliable. Routine screening for thyroid cancer is not recommended because the available tests produce a large number of false-positive results, there is no evidence that early detection by screening improves outcomes, and the benefits of screening may not outweigh the harms. Why may my patients be concerned? News media have covered two recent reports on exposure to radioactive iodine in fallout from about 100 above-ground, nuclear weapons tested in Nevada in the 1950s and 1960s. These stories have reported that this exposure is linked to thyroid cancers. The first report, mandated by Congress and published in 1997 by the National Cancer Institute, provided county-level estimates of the American people's exposure to I-131. A related memo estimated that an 11,000 to 212,000 excess cases of thyroid cancer were likely caused by the I-131 exposure but epidemiological analyses suggest the number is probably in the lower part of this range. An Internet site provides the public a complicated method to estimate individual exposure (http://rex.nci.nih.gov). In 1999, the Institute of Medicine and the National Academy of Sciences, as requested by the U.S. Department of Health and Human Services, published a report assessing the health implications of I-131 exposure from the Nevada tests and advising the government on appropriate responses. The IOM/NRC report concluded that the available data from events four decades past were insufficient to permit either reliable or valid county-or individual-level assessments of exposure to I-131. It also concluded that there was no evidence to determine whether screening for thyroid cancer would improve survival or other health outcomes, taking possible harms of mass screening of the population into account. Does exposure to I-131 cause cancer? Although there is still some disagreement, it is now generally accepted that exposure to I-131 at young ages—especially under age 10—can cause thyroid cancer. The strongest evidence comes from the studies of thyroid cancer in children exposed from the nuclear accident at Chernobyl in 1986. Other suggestive evidence links thyroid cancer to upper body external radiation during childhood. The magnitude by which risk is increased is small, however—probably less than the normal variation in thyroid cancer in different populations around the world. In the general U.S. population, the chance of being diagnosed with thyroid cancer by the age of 95 is roughly 1 chance in 400 for men and 1 in 150 for women. For someone who received 0.1 Gy (10 rad) of I-131 as a child under age 1, the lifetime risk might be roughly 1 chance in 240 for men and 1 in 90 for women. As noted earlier, papillary thyroid cancer, the form related to radiation exposure, is rarely life threatening. By way of comparison, women have about a
OCR for page 115
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications 1 in 8 lifetime risk to age 95 of being diagnosed with breast cancer, and men have about a 1 in 5 risk of being diagnosed with prostate cancer. For those diagnosed with papillary carcinoma, the risk of dying from thyroid cancer is very small. Ten-year survival rates are about 95 percent and 30-year survival is about 90 percent. For the population in general, the lifetime risk over 95 years of dying of thyroid cancer (all forms) is considerably less than 1 in 1,000. In comparison, women have about a 1 in 30 risk of dying of breast cancer and men have about the same risk of dying of prostate cancer. Is thyroid cancer linked to iodine-131 different from naturally occurring thyroid cancer? The cancer linked to I-131 exposure is papillary carcinoma, which has very high survival rates (95 percent), is a usually less aggressive type of thyroid cancer, and has a very good prognosis without screening (95 percent survival at 10 years, 90 percent at 30 years). For the population at risk, today's middle aged adults, there is no evidence that cancer related to childhood radiation exposure differs from naturally occurring cancer. Can I identify whether a patient is at high risk of thyroid cancer from iodine-131 exposure? Unfortunately, in most cases the answer is no—except that there is virtually no risk for someone who was an adult during the testing period. For those who were children at the time of the testing, one problem in estimating exposure is that fallout data were collected in very few places in the United States. This information is not sufficient to make good estimates about fallout in places where data were not collected. Also, because the primary way that I-131 gets to the thyroid is through drinking milk, you would need to know how much milk a person drank as a young child, whether it came from a cow or a goat (because goats' milk concentrates I-131 more effectively than cows' milk), and whether the milk came from a backyard cow or other source that left little time for natural decay of radioactivity. Research indicates that dietary recall data are highly unreliable, especially four decades after the fact. In addition to the uncertainties in estimating I-131 exposure, there are also uncertainties involved in estimating the probability of thyroid cancer related to particular levels of exposure, especially at the low levels generally linked to the Nevada weapons tests. Should people be screened for thyroid cancer? An Institute of Medicine committee made up of physicians (including specialists in endocrinology, radiology, pathology, surgery, and family practice) and others with expertise in public health and evidence-based medicine found that there is insufficient evidence to recommend routine screening for thyroid cancer or other thyroid disease in people who are asymptomatic, whether or not they have been exposed to I-131. For physicians who see patients concerned about thyroid cancer and interested in screening, a process of shared decisionmaking is appropriate. The U.S. Preventive Services Task Force reached similar conclusions about screening for thyroid cancer in the general population. The reasons for this policy are based on concerns about benefits and harms of screening, outlined below, and do not relate to the costs of a screening program. The problem with screening for thyroid cancer is that palpation is not a very reliable test for thyroid nodules or thyroid cancer. Ultrasound is more accurate in finding nodules, but because it can detect very small lesions, it will find nodules
OCR for page 116
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications in perhaps 30 to 50 percent of older adults. The great majority will be benign, but ultrasound can't distinguish them. Many people will then be referred for FNA biopsy, the results of which are also not perfect. If the physical examination is normal and no symptoms are reported, ultrasound is not recommended. One potential result of a screening program is that many people would have surgery when they did not have cancer or had cancers that would never cause problems. This extra surgery would be acceptable if the added vigilance from screening meant better survival and reduced morbidity from thyroid cancer. Survival rates for papillary cancer are already high, however, and no evidence exists to determine whether screening will produce better results. This combination of unproved benefits and clear possible harms is why a systematic screening for thyroid cancer is not recommended. What should I tell patients interested in testing for thyroid cancer? Given the tradeoff of benefits and harms associated with screening, the choice of whether to proceed with screening is a personal decision that should be made, whenever possible, with the input of the patient. If you have a patient who is interested in testing, it is important to advise the patient of the potential benefits and harms so that he or she makes an informed decision about whether screening is worthwhile. Some of the quantiative information presented above may help in your discussions with patients, but people often don't interpret quantiative information correctly. It is appropriate to check their understanding. You might start by explaining that thyroid cancer is uncommon and usually not life-threatening. Possibly 10 percent of the U.S. population may have tiny thyroid cancers, few of which will ever progress to cause problems. You can then explain that the primary screening test is palpation, which looks for thyroid nodules using a physical examination of the neck, combined with questions about possible symptoms (e.g., hoarseness) and risk factors (e.g., radiation therapy at a young age). Possibly one-third to one-half of American adults have thyroid nodules, most of which, however, are too small to be felt. If a lump is found, you may then recommend a biopsy, which uses a thin needle to draw tissue from the lump for examination under a microscope. Most of the time they will not find cancer. Patients should know that screening has the potential to produce harm because the tests used are not always accurate. The tests often identify questionable lumps or cells that lead people to have major surgery either when they don't have cancer or when they have a cancer that will never progress and cause problems. The surgery has very low risk of death but it does cause discomfort and scarring, will require lifelong thyroid hormone replacement therapy, and for a small proportion of cases can disrupt calcium metabolism or cause serious, permanent hoarseness. Patients should also know that there is no scientific evidence that they will enjoy better health as a result of screening or early treatment. If, after learning of these tradeoffs, the patient still wants to be screened, palpation of the neck combined with questions about possible symptoms (e.g., hoarseness) and risk factors (e.g., radiation therapy) may be reasonable and reassuring. Normally, if a nodule is palpated, FNA biopsy is recommended. If the physical examination is normal and no symptoms are reported, ultrasound is not recommended.
OCR for page 117
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications Consistent with the recommendations for patient and public information, information for physicians should be tested before distribution. DHHS should enlist physician organizations in the process of developing information, testing it with representative generalists and specialists, and disseminating it. The committee recognizes that the availability of information and recommendations does not translate automatically or easily into clinical practice. Physicians facing questions from their patients will, however, have some motivation to seek guidance on iodine-131 exposure and clinical practice. Following publication of this report, DHHS could organize briefings for relevant professional organizations in an effort to enlist their support and assistance in reaching their members, for example, through journal articles, stories in professional newsletters, and presentations at national and regional meetings. If DHHS pursues the regional information development and distribution strategy described in Chapter 5, physician leaders could be involved in the planning process. Research and Surveillance Given the lack of research on how people understand and perceive the risks of different cancers and the benefits and harms of cancer screening, the committee suggests that DHHS consider studies to develop more knowledge about perceptions of cancer risk and screening benefit and harms and about the ways perceptions may differ depending on disease characteristics (e.g., prevalence, risk factors, mortality and morbidity, screening test accuracy, treatment consequences). Such research would be helpful in developing strategies for informing and counseling people about risks and options, for example, in deciding whether and how strategies might need to vary depending on disease characteristics. Also helpful would be research on how perceptions are affected by different ways of presenting quantitative information and different ways of structuring clinician-patient communication. With respect to thyroid cancer or radiation-exposed populations specifically, the committee suggests that the evaluation component of the ATSDR medical monitoring program might consider the feasibility of a controlled study to compare the responses to different information formats or different counseling strategies for eligible patients who come in for screening. Although the program might contribute to knowledge in this area, the ATSDR medical-monitoring program is not likely to produce useful information about mortality effects of screening for thyroid cancer because the population that requests screening and meets eligibility criteria will be self-selected and because high long-term survival rates can be expected without screening. The committee already has noted that the results of the Hanford Thyroid Disease Study (which is a research effort rather than a screening program) may help clarify the extent to which childhood exposure to low doses of I-131 is linked to thyroid problems in adults. This clarification should, in turn, indicate whether
OCR for page 118
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications systematic examination of the benefits and harms of screening for nonmalignant thyroid disease in radiation-exposed persons is warranted. In addition to supporting some additional research to inform clinical practice and public health policy, the committee also suggests that DHHS strengthen its surveillance capacity by considering ways to work closely with states to improve the quality and scope of SEER data collection and reporting. It should consider whether collection of additional patient information on residence and personal history is warranted and should review the information collected and the way information is coded to be sure that FNA and open biopsy results are being adequately captured, and that operative results and treatment of confirmed thyroid cancers are being documented. CONCLUSIONS This chapter has examined the potential benefits and harms of screening for people who could be at higher-than-general risk for thyroid cancer as a result of exposure to I-131 fallout. It briefly discussed screening for other thyroid disorders. The IOM committee that prepared this chapter recognized the significant uncertainties that surround the issues of I-131 exposure and cancer risk as illustrated in Figure 4.2, which summarizes the causal pathway from I-131 release to diagnosis of cancer and gives examples of the sources of variation and uncertainty associated with each step in the pathway. The IOM committee accepted the analysis presented elsewhere in this report indicating that those most exposed to I-131 from fallout were at increased risk of thyroid cancer, although much uncertainty surrounds estimates of exposure for specific individuals; concluded that the evidence did not support a positive recommendation for a program to promote systematic thyroid cancer screening for those potentially exposed to I-131 from the Nevada atomic bomb testing program; described a simplified process of shared decisionmaking about screening by palpation as a reasonable approach for people who come to clinicians with requests for screening and with concerns about their risk of thyroid cancer due to I-131 exposure; recommended against screening by ultrasound examination for those who choose screening; suggested that DHHS develop materials for physicians and for physicians to give to concerned patients as part of its program of information and education for the concerned public and clinicians; and proposed directions for research, including the development of more information about how people perceive the benefits and harms of screening and about how different ways of presenting risk information and structuring decisionmaking affect patient perceptions and understanding.
OCR for page 119
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications FIGURE 4.2 Causal pathway: I-131 exposure and consequences with examples of sources of variation and uncertainty. The committee was not charged with doing a cost-effectiveness analysis, and it based its conclusions solely on clinical and epidemiologic grounds. It would not, in any case, have proceeded from analysis of effectiveness to analysis of cost-effectiveness because it found no evidence showing that screening for thyroid cancer is effective.
OCR for page 120
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications Although screening for those at increased risk of thyroid cancer might seem an obvious strategy, screening asymptomatic persons has value only when earlier detection of disease results in earlier treatment that improves outcomes for the screened population and when the benefits of screening exceed the harms. For thyroid cancer screening, there is no evidence of improved outcomes or of benefits that exceed harms. The committee recognizes that it will be challenging to communicate these conclusions in ways that respond to understandable public concerns, beliefs, and values and that accommodate limits on popular comprehension of quantitative information and risk analyses. The next chapter discusses how DHHS might confront this challenge. Figure 4.2 traces the steps involved in the causal pathway connecting atmospheric nuclear weapons tests to the detection and treatment of radiation-induced thyroid cancers. For each step of the pathway, the enclosed boxes contain information about factors likely to introduce uncertainty and variation. The cumulative effect is to dramatically reduce the ability to identify those individuals at highest risk for radiation-induced thyroid cancer and to assure an individual entering the screening portion of the pathway that benefits will exceed harms. ADDENDUM 4A: INTERPRETING SENSITIVITY AND SPECIFICITY Screening tests are accurate when they correctly identify disease in people who have the disease (true positive) or when they correctly identify no disease in people who have no disease (true negative). Tests are inaccurate when people without the disease have a positive test result (false positive) or when people with the disease have a negative result (false negative). Measures of sensitivity, specificity, and positive predictive value are used to assess the accuracy and efficiency of screening tests in identifying people with and without disease (Table 4.3). These measures are defined for the straightforward screening situation when the alternatives are, first, that the disease is either present or absent and, second, that the test results are either positive or negative. When indeterminate test results and conditions are factored in, the computations of sensitivity and specificity are not defined, but grouping can be done. Treating indeterminate results as positive lowers the positive predictive value of the test. Table 4.4 highlights the importance of disease prevalence in assessing the value of screening and shows a worked example using the concepts defined above. Given the same test accuracy, the positive predictive values of the test—or the probability of disease given a positive test result—goes from 8.3 percent to 0.9 percent when the prevalence of a disease drops from 1 percent to 0.1 percent. Even for a test that is reasonably sensitive and specific, the lower the prevalence of the disease, the more false-positive results will be generated relative to true-positive results. Table 4.5 uses the same example but presents it as a Bayesian analysis. In the column headed "revised probability," the first number (8.3 percent) is the probability
OCR for page 121
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications TABLE 4.3 Definition of Terms Term Definition Formulaa Sensitivity Proportion of persons with condition who test positive a/ (a + c) Specificity Proportion of persons without condition who test negative d/ (b + d) Positive predictive value Proportion of persons with positive test who have a condition a/ (a + b) Negative predictive value Proportion of persons with negative test who do not have a condition d/ (c + d) a Explanation of symbols: Condition present Condition absent Legend: Positive test a b a+b a = true positive Negative test c d c+d b = false positive a+c b+d c = false negative d = true negative SOURCE: USPSTF, 1996. TABLE 4.4 Importance of Disease Prevalence Testing Conditions: Size of Population = 100,000 Sensitivity of test = 90% Specificity of test = 90% If disease prevalence = 1% Disease Present Disease Absent Positive Test 900 (a) 9,900 (b) 10,800 (a+b) Negative Test 100 (c) 89,100 (d) 89,200 (c+d) Probability of disease given a positive test result = 8.3% (a / a+b) × 100 1000 (a+c) 99,000 (b+d) 100,000 (a+b+c+d) If disease prevalence = 0.1% Disease Present Disease Absent Positive Test 90 (a) 9,990 (b) 10,080 (a+b) Negative Test 10 (c) 89,910 (d) 89,920 (c+d) Probability of disease given a positive test result = 0.9% 100 (a+c) 99,900 (b+d) 100,000 (a+b+c+d)
OCR for page 122
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications TABLE 4.5 Importance of Disease Prevalence: Bayesian Probability Analysis Test conditions: Sensitivity of test, 90% Specificity of test, 90% Patient Status Prior probability Probability of positive exam Product of probabilities (col1 × col2) Revised probability col1/∑col1 If prevalence is 1% Disease 1% 90% 90 8.3% No Disease 99% 10% 990 91.7% Σ = 1,080 If prevalence is 0.1% Disease 0.1% 90% 0.009 0.9% No Disease 99.9% 10% 0.999 99.1% Σ = 1.008 of disease given a positive test and is the same as in the first part of Table 4.4. The first sum at the bottom of the column headed "product of probabilities" (10.80 percent) multiplied by 100,000 would give the number of positive tests as shown at the right side of the first part of Table 4.4 above. ADDENDUM 4B: INTERPRETATION OF INDETERMINATE AND UNSATISFACTORY FNA SAMPLES The terms sensitivity and specificity are clearly defined only when the test result is either positive or negative. In the case of FNA, the test results can also be indeterminate and unsatisfactory. As shown in Table 2a and 2b in Appendix F, the test performance of FNA will depend upon whether one treats indeterminate and unsatisfactory test results as positive or negative. The following analysis illustrates how treatment of these results affects the calculation of the probability of cancer. For these illustrative purposes, the analysis relies on experience from a major academic center as reported in the table below, which presents the approximate probability of positive, negative, indeterminate, and unsatisfactory results conditioned on the actual findings for the nodule aspirated. Cancer No Cancer Positive 85% 3% Negative 7% 65% Indeterminate 4% 20% Unsatisfactory 4% 2% 100% 100%
OCR for page 123
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications If it is assumed that the prevalence of cancer among nodules biopsied is 3 percent, then (based on the tabular algorithm for Bayes rule presented in Table 4.4) one can revise the probability of cancer based on a positive FNA result: Positive Diagnosis Prior Conditional Product Revised Cancer 3% 85% 255 46.7% No cancer 97% 3% 291 53.3% Σ = 546 The probability of a positive result increases from 3 percent to 46.7 percent. Similarly, after a negative FNA result: Negative Diagnosis Prior Conditional Product Revised Cancer 3% 7% 21 0.3% No cancer 97% 65% 6,305 99.7% Σ = 6,326 The probability of cancer given a negative test decreases from 3 percent to 0.3 percent. Now consider the information provided by an indeterminate or an unsatisfactory FNA result. Indeterminate Diagnosis Prior Conditional Product Revised Cancer 3% 4% 12 0.6% No cancer 97% 20% 1,940 99.4% Σ = 1,952 Unsatisfactory Diagnosis Prior Conditional Product Revised Cancer 3% 4% 12 1.0% No cancer 97% 12% 1,164 99.0% Σ = 1,176 In both cases, the probability of cancer decreases (from 3 percent to 0.6 percent and 1 percent, respectively). This change occurs because both of these results carry information: The chance of either an indeterminate or an unsatisfactory result is only 4 percent if the nodule contains cancer but is higher (20 percent and 12 percent, respectively) if the nodule is free of cancer, because cancers tend to be more cellular and because criteria for the diagnosis of cancer are often more explicit than criteria for diagnosing benign disease. How, then, should an indeterminate result be treated? The physician might treat the interdeterminate result in the same way he or she might treat a positive result, and proceed to surgery. In that case:
OCR for page 124
Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications Positive (including indeterminate results) Diagnosis Prior Conditional Product Revised Cancer 3% 89% 267 10.7% No cancer 97% 23% 2,231 89.3% Σ = 2,498 The revised probability after a positive or indeterminate result would be only 10.7 percent, but 24.98 percent of patients would proceed to surgery (compared to 5.46 percent, as calculated above for positive FNA results) whereas 75.02 percent would avoid surgery. In contrast, if an indeterminate result were treated as negative (i.e., the work-up stops), then the revised probability of cancer would be 0.4 percent because of false negatives. In this case, 82.78 percent of patients would avoid surgery (compared to 75.02 percent above). Negative (including indeterminate results) Diagnosis Prior Conditional Product Revised Cancer 3% 11% 33 0.4% No cancer 97% 85% 8,245 99.6% Σ = 8,278 In a similar vein, consider the effect of an unsatisfactory FNA. Because cancers are more likely to provide satisfactory samples, an unsatisfactory result lowers the probability of cancer, as shown above, from 3 percent to 1 percent. Recall that a negative FNA only lowers the probability of cancer to 0.3 percent (because of the possibility of false-negative results).
Representative terms from entire chapter: