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 66
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat 6 Radiologic Effects of Depleted Uranium Depleted uranium (DU) is naturally radioactive, so there is a potential for radiation-induced effects in addition to the chemical toxicity discussed in the preceding chapters. On a weight basis, U.S. Department of Defense DU is 99.8% 238U and 0.2% 235U, with trace amounts of 234U and 236U. Because of the long half-lives of the various uranium isotopes (see Table 6-1), DU has a low specific activity and hence is only weakly radioactive, with various uranium isotopes undergoing alpha decay and emitting x and gamma radiation. The weighted specific activity of DU is 14.9 kBq/g, or about 60% of the specific activity (25.4 kBq/g) of natural uranium, which contains 0.71 wt % 235U. The process by which DU is created from natural uranium not only reduces the high-specific-activity 235U but reduces even more the higher-specific-activity 234U. Much of the radioactivity in both natural and DU is attributable to trace amounts of the 234U isotope, which accounts for 49% of the activity in natural uranium and about 10% in DU. For a given deposition of uranium in tissues, the dose will be determined largely by and approximately proportional to the specific activity; 1 μg of 234U would deliver about 20,000 times the dose of 1 μg of 238U. However, in considering the dose from uranium deposited in tissues, the mass fraction also needs to be taken into account. Multiplying the mass fraction by the specific activity provides the relative activity or number of disintegrations for each uranium isotope, which can then be converted to the activity fraction, as shown in Table 6-1. For DU, the 238U isotope accounts for 88.8% of the activity and for more than 80% of the dose. Virtually all the rest of the dose from DU deposition in tissues will be from the 234U isotope; the contributions of the other two uranium isotopes are negligible. BIOLOGIC EFFECTS OF IONIZING RADIATION The hazard posed by ionizing radiation is derived from the energy it transfers to tissue that it traverses, which may cause ionization and other molecular
OCR for page 67
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat effects, including chromosomal breaks and additional effects in DNA that may result in genetic damage. All isotopes of uranium undergo decay by the emission of alpha particles from the nucleus with photons of x and gamma radiation. Most of the energy released by the radioactive decay of a uranium nucleus is in the form of kinetic energy imparted to the alpha particle, typically about 4.2 MeV. Despite the large amount of energy, alpha particles have a limited range in soft tissue—about 30 μm—and so are unable to penetrate the superficial dead layer of skin. Thus, alpha particles pose a hazard only if taken into the body. Photons, however, are able to penetrate the body, depositing relatively small amounts of energy as they traverse tissues, and may pose a hazard both internally and externally. Beta particles, which are emitted by some uranium decay products, have a variable range in tissue that depends on their kinetic energy, which is typically a fraction of that of an alpha particle. The most energetic beta particles have a range of only about 1 cm in soft tissue. Biologic effects of radiation are typically classified as deterministic or stochastic. A deterministic effect is one for which there is a clearly defined threshold and that increases in severity as the dose increases above the threshold. An example of radiation-induced deterministic effect (in this case, nonionizing-radiation-induced) is ordinary sunburn, which requires a minimal dose and increases in severity as the dose increases. A stochastic effect has a probability of occurrence that increases in proportion to the dose, but its severity is unrelated to the dose. An example of stochastic effects is the increased probability of skin cancer caused by exposure to sunlight. Stochastic risks posed by exposure to ionizing radiation include radiogenic cancers and genetic mutations. Deterministic effects are usually associated with high doses and typically occur relatively soon after exposure; thus, they are said to have a short latent period (time between exposure and manifestation of the effect). Stochastic effects, such as carcinogenesis, may not manifest themselves for many years and thus have a long latent period. Although compelling evidence to the contrary exists for some cancers, stochastic risks are generally assumed to follow a linear-no-threshold (LNT) dose-response curve, at least for the purposes of determining potential health effects and establishing radiation-protection standards. Thus, doubling the dose is assumed to double the incremental risk, tripling the dose triples the incremental risk, and so on. However, even if the risk coefficients (see discussion below) are accurately and precisely known and the LNT hypothesis is an accurate characterization of the dose-response relationship, the risk is most likely overstated because an exposed person could die from other causes before the radiogenic cancer would be manifested, particularly if the cancer has a long latent period. Furthermore, the simplistic LNT model of response does not consider possible other low-dose effects now under study, such as bystander effects and adaptive responses, that may or may not be significant. The primary radiologic concern related to chronic low-level exposure to DU, as might result from intake of DU and deposition in tissues or from external exposure due to living in a contaminated environment, is the development of a
OCR for page 68
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat fatal radiation-induced cancer. This is, as noted above, a stochastic effect, and its likelihood is characterized in terms of risk coefficients. A risk coefficient is a simplified mathematical statement of the probability that a specific type of malignancy will develop as a result of a specific dose.1 Although in their simplest form risk coefficients imply that the risk of developing a specific cancer is directly proportional to the dose received—that is, they imply a linear dose-response curve—the determination of risk is much more complex and is a function of other factors, including the specific type of cancer, dose rate, latent period, and age. Dose-response curves derived from epidemiologic studies typically serve as the cornerstone and primary basis of the development of risk coefficients. Data on stochastic effects in survivors of the atomic bombings in Japan, who received high radiation doses, are most widely used for that purpose. Epidemiologic studies of exposed worker populations are also extensively used. Because stochastic radiation effects in study populations are rare and often difficult to detect, they are typically observed at doses much higher than zero. In such studies, stochastic effects are not observed or cannot be determined at low doses, so there is a wide gap between a zero dose and the lowest dose at which they are observed. The theoretical low-dose response is therefore obtained by extrapolation, that is, extending the dose-response curve in a linear fashion down to zero—zero dose and zero effect. The risk coefficient is derived from the slope of this straight-line extrapolation. RADIATION DOSE As it decays, DU and its associated decay products emit alpha particles, beta particles, and photons of ionizing electromagnetic radiation (IER), that is, x rays and gamma rays.2 Unlike ultraviolet radiation, used in the sunburn example above, IER has the ability to displace electrons in atoms or to remove electrons, thereby producing ionization or charged particles as it traverses matter, giving up in the process some of or all its energy. The amount of energy deposited in a given mass of tissue is the dose, which determines the extent and severity of 1 Genericaly, the term risk coefficient defines the probability of an adverse event per unit of exposure (for example, the probability of a fatal automobile collision per 100,000 miles driven). As applied in radiation exposure, it usually refers to the probability of a fatal cancer or other stochastic effect per unit of exposure or dose (for example, a 0.04 risk of a fatal cancer per sievert). 2 X rays and gamma rays are forms of electromagnetic radiation that differ only in their mode of production. Gamma rays are electromagnetic radiation emitted from the nucleus of an excited atom; x rays are produced by excitation of the electron field surrounding the nucleus. Typically, x rays are of lower photon energy than gamma rays, but this is not necessarily the case, and the reader is cautioned about the scientific impropriety and inaccuracy of characterizing low-energy IER as x rays and higher-energy IER as gamma rays.
OCR for page 69
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat biologic effects. Over the years, a complex and extensive system of dose quantities has evolved; although all dose quantities are based on energy absorption, one may be numerically different from another even though quantified in the same units. The set of radiologic quantities and units in current use internationally is based on the Systéme International d’Unités (commonly known as the SI), developed by the International Bureau of Weights and Measures (BIPM 1998). The evolution of the SI and its relation to an earlier and now obsolete system of quantities and units that is still in use in the United States are noted briefly below (for more detail, see Kathren 2001). The basic physical quantity of ionizing radiation is absorbed dose, the amount of energy deposited or absorbed by a material as a result of irradiation per unit of mass. It has the SI unit of joules of energy deposited per kilogram of absorbing medium and has been given the special name in the SI of gray (Gy); 1 Gy = 1 J/kg). The older system, still widely used in the United States and prevalent in the older literature, quantifies absorbed dose with the rad, defined as the deposition of 100 ergs/g of absorbing material. The gray and the rad characterize the same physical quantity (absorbed dose), and there is a simple correspondence: 100 ergs = 0.01 J, and 1 kg = 1,000 g; hence, 1 Gy = 100 rad. The dose from any ionizing radiation can be expressed as absorbed dose, but this is not a satisfactory method for relating dose to biologic effect or, in the case of stochastic effects, risk, because some kinds of radiation are more effective than other kinds in producing adverse biologic effects, and some tissues are more radiosensitive than others. Thus, two different kinds of radiation may result in the same absorbed or physical dose and have very different effects. To account for such differences, other dose quantities have been devised on the basis of radiation weighting factors (ICRP 1991). Multiplying the absorbed dose by the appropriate radiation weighting factor yields a single dose quantity that normalizes the risk posed by various kinds of radiation and tissues; this quantity is known as dose equivalent (DE) and is expressed in the SI in sieverts (Sv). A dose of 1 Sv carries the same stochastic risk irrespective of the type of radiation or the tissues being irradiated. For photons and beta radiation, the radiation weighting factor is unity, and the absorbed dose and DE will be numerically equal. For alpha particles, the weighing factor is 20, and the DE will be numerically 20 times greater than the absorbed dose. DE is sometimes known as biologic dose to differentiate from absorbed dose or physical dose. In the old system still widely used in the United States, DE was expressed in terms of rem. If the same weighting factors are used to calculate DE in both the old system and the SI, there is a direct correspondence: 1 Sv = 100 rem. DE is a useful quantity for normalizing or expressing the risks posed by exposure to different kinds of radiation. However, it does not account for the various sizes, radiosensitivities, and likelihoods of cancer induction in specific tissues and organs. To account for those differences, a tissue-weighting factor representing the proportion of stochastic risk posed by irradiation of a specific tissue is applied to the DE for each tissue and organ, and the results are summed.
OCR for page 70
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat The final result is a single value in sieverts equivalent to a total body irradiation that would result in the same overall risk as the sum of the exposures of the various tissues and organs, and this quantity is termed the effective dose (E). Although E is a defined dose quantity (ICRP 1977, 1991), it is basically a statement of risk. It is conceptually important in that it enables an exposure of one organ or a few organs, as might occur from an internal deposition, to be equated with a dose to the whole body by using the risk of fatal cancer induction as the basis. The underlying logic is relatively straightforward and can be understood by noting that uniform irradiation of the whole body will produce a risk of a fatal cancer in each tissue, and the sum of the stochastic risks to all the tissues is equal to the total stochastic risk. If only a portion of the body is irradiated—that is, one or two specific organs, as might be the case with a radionuclide incorporated into their tissues—the total radiogenic stochastic risk comes from those irradiated tissues. The radiation-induced carcinogenic risk to the unirradiated portions of the body is zero. Thus, the total stochastic risk to all the tissues posed by irradiation of only a portion of the body would be expected to be less than if all the tissues were irradiated at the same level. Through the use of tissue-weighting factors, the DE delivered to the irradiated tissues is adjusted to account for the fact that only a portion of the body was irradiated. There are some important caveats with respect to the use of DE and E quantities. DE should never be specified as a pure number of sieverts; it needs to be qualified by noting what organs or tissues were irradiated. E does not need such a qualification, because it is inherent in its definition that it refers to the whole body. And although both DE and E are expressed in sieverts, only in the case of a uniform whole-body irradiation will the numerical values of the two quantities be equal. If only a portion of the body was irradiated, the DE to that portion of the body will be numerically greater than E, and the converse will be true for the unirradiated portions of the body. The statement of the magnitude of E does not tell how it was obtained; there is no specification of the doses to the individual tissues and organs from which it was derived. Assessment of the tissue dose from DU that is taken into the body is a complex process that requires knowledge of the physical and chemical properties of the material and its biokinetics. Once uranium is incorporated into the body, it delivers a dose to the surrounding tissues primarily in alpha radiation. The energy of the typical alpha particle emitted by the uranium isotopes that constitute DU is about 4.2 MeV, which is perhaps 10 times or more greater than the average energy of a typical beta particle or photon emitted by the decay of uranium and its products. Alpha particles with that energy can travel about 30 μm through tissues—about the diameter of a single cell. Therefore, all the alpha-particle energy is deposited in a very small volume of tissue, and, inasmuch as dose is simply energy absorbed or deposited per unit of mass, the dose absorbed by this small volume will be much greater than the absorbed dose resulting from a single photon or beta-particle interaction. However, the general practice is to average the dose over the entire tissue or organ. With a weighting factor of 20, the DE from an alpha particle averaged over the entire tissue or organ into which
OCR for page 71
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat the uranium is incorporated will typically be more than 100 times greater than the DE from a single photon or beta particle. Thus, the primary radiologic concern related to uranium incorporated into tissues is alpha irradiation. EXTERNAL EXPOSURE TO DEPLETED URANIUM: DIRECT RADIATION The uranium isotopes that constitute pure DU all undergo decay by emission of alpha particles accompanied by x and gamma radiation. In addition, there is considerable ingrowth of the uranium decay-series progeny after 50 y. The DU decay series will not reach equilibrium for over a million years. As discussed above, the alpha particles emitted by the uranium isotopes that constitute DU have low penetrating power and cannot penetrate the inert or nonliving outer layer of skin that covers most of the body. Hence, the alpha radiation from DU and its decay products is not of concern from the standpoint of external exposure. More important are the beta radiation emitted by the uranium decay products and the associated x and gamma radiation from the decay of the uranium isotopes and their decay products, and it is possible to achieve an external exposure from these kinds of radiation. The magnitude of such an exposure would be determined primarily by three factors: the quantity of DU, the distance and interposition of shielding materials between the body and the DU, and the duration of exposure. The maximal exposure would occur from direct contact of the skin with a chunk or slab of metallic DU or with simple compounds of DU. The dose rate at the surface of the metallic DU or DU compound (that is, the surface dose rate) is the maximal external dose rate. A sufficiently large slab of metallic DU provides what is termed an infinite-thickness slab under specified conditions, and increasing the thickness or size does not increase the dose rate. Fetter and von Hippel (1999) performed a comprehensive theoretical evaluation of the hazards posed by DU munitions and calculated the maximal dose rate to the skin from direct contact to be 2.5 mSv/h (250 mrem/h), mostly from beta radiation. Their calculated value is close to measured values that have been reported for various compounds and compositions of uranium over the years (Kinsman 1954; BRH 1970; Healy 1970; Kathren 1975). For a slab of natural uranium, the measured beta dose rate in air at the surface through a 7-mg/cm2 polystyrene filter, used to mimic the cornified (dead) layer of skin, is 2.33 mSv/h (233 mrem/h).3 Photons are reported to contribute another 10% to the surface dose rate, bringing the total to 2.55 mGy/h (255 mrad/h; Kathren 1975). The measured value of about 0.23 mSv/h (23 mrem/h) for x and gamma radiation is about 10 times greater than the 3 The measured values were originally reported in units of absorbed dose, which were converted to DE by multiplying by a radiation weighting factor of unity for beta and photon radiation to avoid confusion caused by the use of different radiation quantities and units and to provide a common basis for comparison.
OCR for page 72
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat theoretically determined 0.025 mSv/h (2.5 mrem/h) reported by Fetter and von Hippel (1999), but the difference could be accounted for by bremsstrahlung from the interactions of beta particles with the air or within the slab itself, which were included in the measured values but not fully in the calculations by Fetter and von Hippel. In either case, the external dose rates associated with uranium are relatively low and would require continuous direct contact of weeks to months to reach the threshold for deterministic effects (erythema) on the skin. The external dose rate from a large mass of DU decreases as a function of distance from the material, largely because of attenuation of the beta particles and to a lesser extent of x and gamma radiation by interactions with air. The dose rate near the surface of an infinite slab of uranium is remarkably constant for the first few millimeters but drops off steeply as the distance from the source increases: 10 cm from the source, the total air dose rate from beta, x, and gamma radiation combined is only 0.12 mGy/h, or 0.04 times the dose at the surface (Kathren 1975). Measurements have shown that the relative contribution of x and gamma radiation to the total dose also increases because of the large attenuation of beta radiation by the air. At or near the surface of a natural uranium slab, x and gamma radiation contributes about 10% of the dose; 10 cm from the surface, that fraction increases to about 20%; and 1 m from the source, the dose is almost exclusively from x and gamma radiation and about a few tenths of 1 mGy/h—somewhat greater than observed by Fetter and von Hippel but still small enough to preclude, for all practical purposes, deterministic effects. Although deterministic effects of external exposure to DU are extremely unlikely, external irradiation of the skin by DU and its decay products would pose a risk of stochastic effects. A contact time with the skin of 400 h, corresponding to a dose of about 1 Gy, has been calculated to increase the risk of skin cancer by about 40% (UNSCEAR 2000). If one assumes a linear-no-threshold response for skin cancer, that dose equates to an increased risk of 0.1%/h of exposure. Similarly, tissues lying below the skin would suffer an increased stochastic risk, but it would be smaller than the already low stochastic risk to the skin. Widespread contamination of the ground with DU would produce measurable external ionizing-radiation fields, but, as indicated above, the dose rates would be smaller than that from direct contact and thus insufficient to produce deterministic effects. However, assuming once again a linear-no-threshold response for cancer induction, exposure to such radiation fields would produce an additional stochastic risk, albeit small, of radiogenic cancer. Fetter and von Hippel (1999) estimated the effective dose rate for a person standing on ground uniformly contaminated with DU at 1 g/m2, which they considered an upper limit for a battlefield area, to be 0.01 mSv/y or about 0.1 times the dose rate from uranium naturally present in the soil and less than 2% of the typical dose rate associated with terrestrial and cosmic radiation. Even continuous exposure at that level for a period of several years would result in a negligible increase in the stochastic risk of radiogenic cancer or in the genetic effect relative to the natural incidence. A full decade of continuous exposure would provide a theoretical
OCR for page 73
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat increase in the risk of a stochastic effect—that is, radiogenic cancer or genetic effect—of less than 1% on the basis of the risk coefficients put forth by the International Commission on Radiological Protection (ICRP 1991); exposure for 70 y would increase the theoretical increase in risk to about 3.5%. INTERNAL EXPOSURE TO DEPLETED URANIUM Uranium may enter the body by inhalation, ingestion, or dermal exposure. Its uptake, distribution, and elimination from the tissues depend on the physicochemical characteristics of the uranium, the route of entry, and the biokinetics of uranium in the body. As discussed in Chapter 2, mathematical models have been developed to describe the biokinetic processes of DU. Irrespective of the biokinetic model used, it is clear that chemical toxicity is so overwhelming relative to radiotoxicity that radiation-induced deterministic effects from intake of uranium are extremely unlikely or impossible (ICRP 1988). However, stochastic effects—such as an increased likelihood of cancer for which no threshold is postulated—are possible. The organs at risk are those with the greatest concentration of DU, and dose and associated risk of stochastic effects can be calculated for each tissue and organ. Ingestion of DU does not pose substantial risk, and indeed the chemical effects would far outweigh the radiologic effect. Using the ICRP schema, one can calculate the effective-dose coefficients for DU. For ingested soluble DU, the effective dose coefficient is about 4.5 × 10−8 Sv/Bq. That corresponds to an effective dose of about 0.67 mSv for an intake of 1 g of DU, which would produce a total stochastic risk, including both carcinogenesis and genetic risk, of about 3.3 × 10−5. In other words, the total risk of a stochastic effect from ingestion of 1 g of DU is about 33 in a million. If linearity of the response is assumed, a person would have to ingest about 300 g—more than a half-pound—of soluble DU to be subjected to a 1% risk of a stochastic effect. Ingestion of such a large quantity of soluble DU is virtually certain to produce chemotoxic effects on the kidneys. If insoluble DU is ingested, the fraction absorbed is one-tenth that of the fraction of soluble DU absorbed, and the stochastic risk per unit intake is concomitantly lower. A much larger stochastic risk is posed by inhaled DU than by ingested DU. The stochastic risk posed by inhaled DU, which is almost exclusively a risk of lung carcinogenesis, is greater when insoluble DU is involved, in contrast with the risk posed by ingested DU. That is particularly true because insoluble particles, once deposited in the lungs, may reside there and irradiate lung tissue for years. For example, for an inhaled insoluble (class S4) DU aerosol with 1-μm active median aerodynamic diameter (AMAD), the effective dose coefficient, calculated according to ICRP (1994b), is 7.5 × 10−6 Sv/Bq. Thus, the risk posed 4 As discussed in Chapter 2, ICRP broadly classifies inhaled aerosols in terms of their absorption rate in the body as fast (F), moderate (M), and slow (S).
OCR for page 74
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat by inhalation of 1 mg of DU is about 5.6 × 10−6, or about 5 in a million. For class M and class F aerosols, the risk to the lungs is lower because of the more rapid clearance, although the risk to other tissues of the body is somewhat higher because of increased absorption. Effective dose coefficients calculated according to ICRP (1994b) for class M and class F aerosols of DU with the same AMAD are 2.7 × 10−6 and 5.0 × 10−7 Sv/Bq, respectively, which are stochastic risks smaller than associated with class S material. More detailed radiation dose and risk factors can be obtained from ICRP (1996), which considers such factors as age at the time of intake, and from the Environmental Protection Agency (EPA 1999). The summary of radiation dose and risk factors presented in Tables 6-2 and 6-3 was developed from the data in those documents; the values are normalized to both 1-Bq and 1-mg intakes for inhaled and ingested DU and for morbidity and mortality. The data presented below are by no means complete but merely representative of the risk factors associated with common routes of exposure and based on currently available data from the scientific literature. TABLE 6-2 Radiation Dose (Sv) and Risk per Becquerel (Bq) Intake of Depleted Uranium Summary of Radiation Dose and Risk Factors Route of Intake Organ Effective Kidney Lung Bone Liver Radiation dose (Sv) per Bq DU intake Inhalation Class M 1.29 × 10−6 2.27 × 10−5 3.58 × 10−5 4.81 × 10−7 2.92 × 10−6 Class S 1.72 × 10−7 6.79 × 10−5 4.61 × 10−7 6.29 × 10−8 8.18 × 10−6 Ingestion 2.57 × 10−7 2.48 × 10−8 7.19 × 10−7 9.69 × 10−8 4.50 × 10−8 Injection 1.29 × 10−5 1.24 × 10−6 3.59 × 10−5 4.84 × 10−6 2.25 × 10−6 Radiation-morbidity risk per Bq DU intake Inhalation Class M 9.45 × 10−10 1.47 × 10−7 3.55 × 10−10 5.51 × 10−10 1.51 × 10−7 Class S 1.04 × 10−10 4.08 × 10−7 3.93 × 10−11 5.96 × 10−11 4.08 × 10−7 Ingestion 2.04 × 10−10 1.11 × 10−10 9.92 × 10−11 1.2 × 10−10 1.75 × 10−9 Radiation-mortality risk per Bq DU intake Inhalation Class M 6.14 × 10−10 1.4 × 10−7 2.49 × 10−10 5.24 × 10−10 1.42 × 10−7 Class S 6.81 × 10−11 3.88 × 10−7 2.75 × 10−11 5.67 × 10−11 3.88 × 10−7 Ingestion 1.23 × 10−10 7.18 × 10−11 4.85 × 10−11 2.06 × 10−11 6.11 × 10−10
OCR for page 75
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat TABLE 6-3 Radiation Dose (Sv) and Risk per Milligram Intake of Depleted Uranium Summary of Radiation Dose (Sv) and Risk Factors Route of Intake Organ Effective Kidney Lung Bone Liver Radiation dose (Sv) per milligram DU intake Inhalation Class M 1.80 × 10−5 3.17 × 10−4 5.00 × 10−4 6.72 × 10−6 4.09 × 10−5 Class S 2.41 × 10−6 9.48 × 10−4 6.44 × 10−6 8.78 × 10−7 1.14 × 10−4 Ingestion 3.60 × 10−6 3.47 × 10−7 1.00 × 10−5 1.35 × 10−6 6.29 × 10−7 Injection 1.80 × 10−4 1.73 × 10−5 5.02 × 10−4 6.77 × 10−5 3.15 × 10−5 Radiation-morbidity risk per milligram DU intake Inhalation Class M 1.32 × 10−8 2.06 × 10−6 4.96 × 10−9 7.70 × 10−9 2.12 × 10−6 Class S 1.46 × 10−9 5.70 × 10−6 5.50 × 10−10 8.33 × 10−10 5.70 × 10−6 Ingestion 2.84 × 10−9 1.56 × 10−9 1.39 × 10−9 1.68 × 10−9 2.45 × 10−8 Radiation-mortality risk per milligram DU intake Inhalation Class M 8.58 × 10−9 1.96 × 10−6 3.48 × 10−9 7.31 × 10−9 1.99 × 10−6 Class S 9.51 × 10−10 5.42 × 10−6 3.84 × 10−10 7.92 × 10−10 5.42 × 10−6 Ingestion 1.71 × 10−9 1.00 × 10−9 6.77 × 10−10 2.88 × 10−10 8.53 × 10−9 There is some suggestion that the biokinetic models put forth by ICRP and others may not completely characterize absorption and distribution in the tissues and thus may lead to inaccurate assessments of risk. Recent analysis of the tissues of a former uranium worker who died at the age of 83 years from an acute cerebellar infarct expectedly showed that the greatest depositions of uranium were in the respiratory tract and the skeleton (Russell and Kathren 2004). The quantity of uranium in the kidneys was unexpectedly small, and the spleen and urinary bladder contained 4.27 and 1.76 times the quantity in the liver, respectively. More important than total organ content is concentration (grams of uranium per gram of tissue) because dose is directly proportional to concentration in the tissue. Other than the respiratory tract, the highest concentrations were in the bladder and the spleen, and this raises the question of whether the models adequately address stochastic risk. In this case, the concentration in the spleen was more than 40 times that in the liver; therefore, the dose to the spleen might have been 40 times that to the liver. Similarly, the concentration in the bladder was nearly 10 times greater than that in the liver. A high concentration was also found in the thyroid.
OCR for page 76
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat What weight should be given to the tissue distributions in the above case is problematic because high uranium concentrations were not found in the spleen and bladder in postmortem analysis of the tissues of two other people with no history of exposure to uranium although a high uranium concentration was found in the thyroid of one (Kathren 1997). The high spleen concentrations are consistent with the observations of Hedaya et al. (1997), who noted high concentrations of uranium in the spleens of rats after intraperitoneal injection of uranium. Because high concentrations of uranium in the spleen have not been reported in other animal studies, the observations of Hedaya et al. may simply be an anomaly or may reflect a very short residence time in the spleen. Given that the function of the spleen is macrophagic removal of abnormal erythrocytes and that blood has been found to contain relatively high concentrations of uranium, the evidence may suggest a mechanism of brief retention in the spleen (Russell and Kathren 2004). Of perhaps greater importance are the observations of low uranium concentrations in the kidneys in the above cases and others (Wrenn et al. 1985b; Fisenne and Welford 1986), which suggest that the risk of radiation-induced stochastic effects on the kidney may be lower than indicated by the various biokinetic models. In any case, for a given concentration in the kidneys, the risk of chemotoxic effects of DU far outweighs the risk of radiologic effects. Despite the above indications of potential shortcomings, the biokinetic models and dose-calculation methods are adequate to provide a good indication of the radiologic risks posed by DU. Even if the biokinetic models need to be revised along the lines indicated above, such revision would result in only a small change in calculated risks. EPIDEMIOLOGIC STUDIES Since the late 1970s, the Committee on the Biological Effects of Ionizing Radiations of the National Research Council has carried out exhaustive reviews of the scientific literature pertaining to the effects of low-level radiation on human populations (NRC 1980, 1988, 1990, 1999, 2006). Reviews of DU have also been performed by the Royal Society (2002) and the Institute of Medicine (IOM 2000). Epidemiologic studies of natural uranium and cancer must consider total risk of both chemical and radiologic carcinogenicity of uranium because it is impossible to separate them in such studies (see discussion of chemical carcinogenicity in Chapter 7). The large body of radioepidemiologic literature (discussed below) includes studies of Gulf War veterans, uranium miners, other occupationally exposed groups, and the general population. Uranium miners have long been known to have a higher risk of lung-cancer mortality than the general population, associated primarily with the inhalation of radon progeny and other factors, such as smoking, that potentiate the effects of the exposure (NRC 1999). More relevant are the numerous epidemiologic studies of populations occupationally exposed to radiation, especially studies of uranium workers,
OCR for page 77
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat studies of military personnel and civilians exposed to DU from munitions used in Kosovo and Iraq, and the recent large study of 13,960 British uranium workers (McGeoghegan and Binks 2000a). Although the studies provide some support for the risk coefficients, the epidemiologic studies of human populations have revealed no definitive evidence linking uranium exposure to human deaths (ATSDR 1999). Studies of Gulf War Veterans Information on possible radiologic health effects of DU in humans has come from extensive followup studies over a 12-y period of military personnel exposed to DU, notably by Hooper et al. (1999), McDiarmid et al. (2000, 2001a,b, 2004a,b), and Squibb et al. (2005). Those studies and others (Mitchel et al. 1999; Leggett and Pellmar 2003; Mitchel and Sunder 2004) generally confirm the applicability of the current biokinetic models and should point to additional refinements, particularly with respect to the biokinetics of embedded DU fragments in combat wounds. The studies have demonstrated none of the classic chemotoxic effects associated with uranium exposure or deterministic effects of irradiation. No cases of leukemia, bone cancer, or lung cancer have occurred in the 10 y of followup of 15 U.S. Gulf War veterans with high uranium excretion and documented DU fragment wounds (McDiarmid 2001; McDiarmid et al. 2004b). However, little meaning can be attached to that result because of the small number of exposed veterans and the short followup. An 11-y followup has been conducted of UK Gulf War veterans by Macfarlane et al. (2003). Among the 2,092 veterans who reported being exposed to DU, seven incident cancers were found compared with 139 in 26,426 Gulf War veterans who reported no DU exposure. The rate ratio (rate of disease in an exposed group divided by the rate in an unexposed group) was 0.63 (95% confidence interval, 0.3-1.36). Details of the types of cancer in the seven cases were not reported, but it was stated that there were no excesses of specific cancer types. The exposure to DU was by self-report only and was not verified. Studies of Uranium Workers Studies of uranium workers are summarized in Table 6-4 (detailed descriptions are provided in Appendix B). The evidence on specific cancers is discussed below. Lung Cancer The evidence on uranium exposure and lung cancer is mixed. The two largest U.S. studies of uranium-processing workers had increased standardized
OCR for page 78
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat TABLE 6-4 Standardized Mortality Ratios with (95% Confidence Intervals) and [Observed Number of Deaths] for Selected Cancers in Uranium Workers Study (Reference) Total Cancers Lung Cancer Renal Cancer Hepatic Cancera Brain and CNS Cancer Testicular Cancer Bone Cancer Leukemia and Aleukemia Lymphomab Colorado Plateau uranium-mill workers (with no history of uranium mining) (Waxweiler et al. 1983; Pinkerton et al. 2004) 0.90 (0.78-1.04)  1.13 (0.89-1.41)  0.81 (0.22-2.06)  0.79 (0.22-2.03) c — — — 0.66 (0.21-1.53)  2.29 (1.06-4.34)  TEC/Y12 (1943-1947): Oak Ridge uranium conversion and enrichment, all workers (Polednak and Frome 1981) 0.85 (0.80-0.91)  1.09 (0.98-1.22)  0.75 (0.47-1.14)  0.57 (0.32-0.93  0.95 (0.66-1.32)  0.55 (0.17-1.33)  0.90 (0.36-1.87)  0.92 (0.66-1.24)  0.62 (0.42-0.90)  TEC/Y12 (1943-1947): Oak Ridge uranium conversion and enrichment, alpha and beta chemistry departments (Polednak and Frome 1981) 0.85 (0.76-0.95)  0.99 (0.82-1.19)  — — — — 0.78 (0.13-2.58)  0.65 (0.34-1.13)  — Y12 (1947-1974): Oak Ridge uranium-metal production and recycling (Checkoway et al. 1988; Loomis and Wolf 1996)d 1.02 (0.93-1.12)  1.20 (1.04-1.38)  1.39 (0.80-2.26)  0.92 (0.39-1.81) c 1.28 (0.76-2.02)  0 (0-1.63) [0/2.26] 0 (0-2.53) [0/1.2]e 0.60 (0.29-1.10)  0.60 (0.26-1.19)  Mallinckrodt uranium-processing workers (Dupree-Ellis et al. 2000) 1.05 (0.93-1.17)  1.02 (0.83-1.24)  1.17 (0.54-2.18)  0.42 (0.07-1.30) c 1.57 (0.84-2.64)  0.93 (0.05-4.08)  — 1.11 (0.57-1.89)  0.52 (0.13-1.42) 
OCR for page 79
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat Fernald: fabrication of uranium products (Ritz 1999) 1.09 (0.98-1.22)  1.01 (0.83-1.21)  0.63 (0.20-1.46)  1.62 (0.70-3.20) c 1.24 (0.64-2.17)  0.67 (0.01-3.74)  0 (0-3.7) [0/0.99]e 1.16 (0.62-1.98)  1.81 (1.03-2.96)  Linde uranium-processing facility (1943-1949) (Dupree et al. 1987; Teta and Ott 1988) 1.06 (0.83-1.32)  0.97 (0.60-1.48)  — 0 (0-1.84) [0/2.0] 0.43 (0.12-1.09)  — 0 (0-2.63) [0/1.4] 1.00 (0.50-1.79)  0.89 (0.32-1.93) f Portsmouth gaseous diffusion (Brown and Bloom 1987)g 0.87 (0.71-1.05)  0.93 (0.67-1.25)  — — — — 25 (0.03-7.0) h 1.18 (0.43-2.55)  1.72 (0.91-2.99)  Savannah River nuclear-fuel production (Cragle et al. 1988) 0.74 (0.65-0.85)  0.83 (0.66-1.02)  0.38 (0.10-1.02)  0.85 (0.27-2.05)  0.55 (0.24-1.08)  — — (0.89-2.26)  0.53 (0.17-1.29) i United Nuclear Corp. nuclear-fuel fabrication (Hadjimichael et al. 1983)j 0.87 (0.68-1.10)  1.06 (0.63-1.69)  1.10 (0.22-3.20)  — 2.70 (0.99-5.88)  — 0.95 (0.01-5.30)  1.78 (0.30-5.88)  2.13 (0.24-7.71) i Florida phosphate workers (Checkoway et al. 1996)k 0.94 (0.88-1.00)  1.18 (1.07-1.29)  0.83 (0.54-1.24)  0.56 (0.29-0.97)  0.85 (0.57-1.23)  — — 0.99 (0.72-1.33)  0.61 (0.35-0.98)  Atomic Weapons Establishment, UK (Beral et al. 1988) 0.88 (0.63-1.20)  0.65 (0.34-1.13)  4.30 (0.89-12.6)  — 0.85 (0.02-4.74)  — — 0 (—)  1.44 (0.04-8.02)  Capenhurst, UK, 235U-enrichment plant, mortality (McGeoghegan and Binks 2000b)l 0.88 (0.75-1.02)  0.89 (0.70-1.13)  0.49 (0.08-1.62)  0.60 (0.10-1.98) c 1.39 (0.61-2.75)  0 (0-5.55) [0/0.54] 0 (0-6.51) [0/0.46] 0.69 (0.22-1.68)  1.23 (0.54-2.42) 
OCR for page 80
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat Study (Reference) Total Cancers Lung Cancer Renal Cancer Hepatic Cancera Brain and CNS Cancer Testicular Cancer Bone Cancer Leukemia and Aleukemia Lymphomab Capenhurst, UK: 235U-enrichment plant, incidence (McGeoghegan and Binks 2000b)l 0.82 (0.70-0.95)  0.84 (0.63-1.11)  0.45 (0.08-1.48)  0.45 (0.08-1.50) c 1.03 (0.33-2.48)  0.96 (0.16-3.18)  0 (0-7.68) [0/0.39] 0.74 (0.24-1.78)  0.59 (0.19-1.43)  Springfields, UK: mortality (McGeoghegan and Binks 2000a)l 0.86 (0.81-0.91)  0.85 (0.77-0.95)  0.60 (0.33-1.00)  1.18 (0.76-1.75) c 0.67 (0.41-1.03)  0.61 (0.10-2.01)  0.67 (0.11-2.22)  1.00 (0.69-1.39)  0.77 (0.51-1.13)  Springfields, UK: cancer incidence (McGeoghegan and Binks 2000a)l 0.81 (0.76-0.86)  0.75 (0.65-0.85)  0.63 (0.36-1.03)  0.53 (0.29-0.90) c 0.64 (0.35-1.09)  0.92 (0.43-1.75)  0 (0-1.56) [0/1.92] 0.79 (0.51-1.18)  0.92 (0.63-1.30)  Total Observed/Expected Casesm 4,859/5,339 1,868/1,824 99/121 75/96 144/154 8/16 11/17 192/197 128/154 aSome studies reported only “liver, biliary tract, and gall bladder” combined. bUnless otherwise noted, lymphoma was defined as International Classification of Diseases-8 (ICD-8) codes of 200 (“Lymphosarcoma and reticulosarcoma” and 201 (“Hodgkin’s disease”) or their equivalent in other ICD versions. cIncludes both liver and gall bladder. dWhite men only. There were few nonwhite workers, and there was underascertainment of mortality in women. eWhen there were no observed cases, both the observed and the expected values are given. fIncludes all lymphohematopoietic cancers (ICD-8 200-209). gIncludes only “Subcohort I,” which consists of those who at some time worked in one of the departments considered to have uranium exposure. hData not given for “Subcohort I,” so entire cohort included. iIncludes only lymphosarcoma and reticulosarcoma (ICD-8 200). jCancer-incidence data on “industrial” male employees, which excluded office workers. kStandardized mortality ratios similar for white and nonwhite men, so results for combined groups are presented. lMortality or cancer incidence in radiation workers only. For incidence data, standardized incidence ratios are given. mSums do not include row labled TEC/Y12 (1943-47): “Oak Ridge uranium conversion and enrichment, alpha and beta chemistry departments,” because those workers were already included in the TEC/Y12 row for all workers.
OCR for page 81
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat mortality ratios (SMRs) that were statistically significant (Loomis and Wolf 1996) or nearly so (Polednak and Frome 1981), as did a study of phosphate workers (Checkoway et al. 1996). But other studies did not show lung cancer-excesses, and a study that included the four largest U.S. cohorts of uranium workers (Dupree et al. 1995) did not show an exposure-response trend with respect to internal uranium exposure. One possibility is that high smoking rates in the workers, which was confirmed by the few data available, may have led to increased overall rates of lung cancer. Nevertheless, it cannot be ruled out that inhaled uranium particles may lead to an increased incidence of lung cancer, especially given that alpha particles are emitted by uranium. Lymphoma Lymphoma is a biologically plausible outcome of inhalation exposure to uranium, given that uranium deposited in the lungs tends to migrate to the thoracic lymph nodes (Singh et al. 1987). An early study of Colorado Plateau uranium millers suggested an increase in nonleukemic lymphopoietic cancers (four observed and 1.02 expected; Archer et al. 1973). Additional followup of the cohort confirmed the finding with a significantly increased risk (Pinkerton et al. 2004). A statistically significant SMR of 1.81 was found in Fernald workers, and a suggestively increased rate in the Portsmouth gaseous-diffusion workers (Brown and Bloom 1987). In contrast, the two largest studies (Polednak and Frome 1981; McGeoghegan and Binks 2000a) and several others found no indication of an increased risk, so uncertainty remains. Another source of uncertainty regarding lymphoma is that lymphoma’s International Classification of Diseases coding has changed appreciably, so death-certificate codes are somewhat inconsistent, and there were variations among studies in the reporting of lymphomas. Leukemia and Bone Cancer Uranium tends to deposit in the bone, so bone cancer and leukemia are diseases of interest. The collective uranium-worker studies had too few bone cancers for a useful assessment. None of the individual studies showed a statistically significant increase in the rate of leukemia, and the pooled observed number of leukemias was less than expected on the basis of general population rates. If the “true” SMR for leukemia were about 1.5, the TEC/Y12 study or the phosphate-workers study would probably have been able to detect it. Most other studies would have adequate statistical power to detect only much higher risks. Renal Cancer The kidneys are suspected target organs because of the nephrotoxicity of uranium. However, the degree of nephrotoxicity depends on the route of expo-
OCR for page 82
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat sure, the solubility of the uranium compound, and the concentration of exposure. The proportion of exposure to various uranium compounds was not documented in most of the studies, so only rough assessments can be made. It is noteworthy that the TEC/Y12 chemistry workers were exposed to uranium tetrachloride and a number at Y12 workers to uranium hexafluoride, and both compounds are relatively soluble. There were nonsignificant suggestions of excess renal cancer in Y12 workers (but not TEC/Y12 workers) and UK Atomic Weapons Establishment workers and a suggestion of a deficit among the UK Springfields workers. Overall, about 18% fewer renal cancers were observed than would be expected on the basis of general population rates. There were nonsignificant suggestions of excess chronic nephritis in Mallinckrodt and Colorado Plateau uranium workers but a 15% deficit in chronic nephritis across all the studies. An increased mortality rate (SMR, 2.6) from chronic renal failure among U.S. uranium miners has been reported, but details were not given (Thun et al. 1982). Testicular Cancer One study reported an early excess of testicular cancer in Gulf War veterans (Gray et al. 1996), but followup failed to support the finding (Knoke et al. 1998); neither study had data with reference on DU. None of the uranium-worker studies showed an excess of testicular cancer, but a number of them did not report on testicular cancer. Other Cancers Only the Fernald study showed a nominal (nonsignificant) increase in hepatic cancer; the others had uniformly negative results. Several studies showed suggestive increases in brain and central nervous system (CNS) cancer, but only one small study (United Nuclear Corporation) approached statistical significance. Overall, the numbers of observed and expected brain and CNS cancers were almost identical. Strengths and Weaknesses Perhaps the greatest weakness of the uranium-worker studies is the lack of information on individual (or work-location-based) uranium exposure concentrations, so exposure-response analyses could not be performed in most studies. Similarly lacking were uranium urinary bioassay data, which would have yielded an estimate of exposure concentrations (once solubility of the uranium compounds was taken into account). The data on lung cancer are difficult to interpret because most studies had little or no data on smoking. Even a nested case-control study of four uranium facilities that made a concerted effort to gather smoking information from medical records obtained it on fewer than half the study subjects (Dupree et al. 1995).
OCR for page 83
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat Most of the studies had to rely on death-certificate information with its questionable accuracy as to cause of death. Some pertinent types of disease, such as thyroid cancer, cannot be studied adequately with mortality data, because their case-fatality rates are low. Nearly all the studies relied on comparisons with external general populations to evaluate the mortality experience of study workers (that is, SMRs). It is well known that SMRs tend to be less than unity in working groups because such groups do not include people who are too ill to work. Similarly, those who work for longer periods and hence have the potential to accrue more exposure are also those who tend to maintain good health over a sustained period. Hence, SMRs tend to be biased downward and so to be are less likely to demonstrate excesses, and SMR statistics tend to be on the conservative side. SMRs in an entire worker cohort may also fail to detect excesses that are largely confined to the subgroup of workers who received high exposures because such excesses are “diluted” out by the typically much greater number of workers who had smaller exposures. However, given the weaknesses, the studies cited above included nearly 110,000 workers with potential uranium exposure, many of whom had substantial and prolonged exposures and long followups. That little excess risk of cancer or renal disease was seen suggests that uranium compounds are not highly carcinogenic or nephrotoxic in humans. The final row of Table 6-4 shows the overall observed and expected numbers of various cancers. Only for lung cancer and leukemia are the numbers of deaths about as large as the expected numbers, and in no case is there an important excess. Community Studies Uranium-mill tailings have been used since 1951 as construction fill material in some counties in Colorado. Mason et al. (1972) investigated whether the counties that used uranium-mill tailings extensively had higher cancer rates. They evaluated lung cancer, leukemia, and all other cancers combined and found no correlation in either males or females in the rates of these cancers with mill-tailings use. Boice et al. (2003a) evaluated cancer mortality in counties adjoining the Apollo and Parks uranium-plutonium-processing facilities in western Pennsylvania in comparison with six control counties. They found no excess total cancers, childhood leukemia, or cancers of the lungs, kidneys, bone, or liver. Those results agreed with the findings on cancer incidence in those counties (Boice et al. 2003b). A study of cancer mortality during 52 y was conducted in Karnes County, Texas, which for 40 y had over 40 uranium mines and three uranium mills (Boice et al. 2003c). Cancer mortality in that county—including deaths in a number of uranium workers and those exposed to uranium dust, mill tailings, and so forth—was compared with that in four other counties in the region that
OCR for page 84
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat were matched on sociodemographic characteristics. There were no differences in cancer mortality between the uranium-industry county (1,223 cancers) and the other counties, nor were there differences between periods before and after uranium operations began. Relative risks comparing Karnes County with the other counties for cancer sites of interest include total cancers (RR, 1.00), lung (RR, 1.08), kidney and renal pelvis (RR, 0.58), liver (RR, 0.81), brain and CNS (RR, 0.92), non-Hodgkin lymphoma (RR, 1.00), Hodgkin disease (RR, 1.79; 95% confidence interval, 0.9-3.6), and leukemia (RR, 1.15). Lopez-Abente et al. (1999, 2001) conducted a study of cancer mortality in Spanish towns that were within 30 km of a uranium-processing plant. On the basis of the closer-in region of 0-15 km from any of the four sites, the SMRs were 0.89 for all cancers, 0.98 for leukemia, 0.55 for non-Hodgkin lymphoma, 1.00 for Hodgkin disease, 0.92 for lung cancer, 0.74 for brain cancer, and 0.93 for renal cancer. When trend tests were examined according to distance from a processing plant, there were no statistically significant trends if only exposed areas were included, but there was a significant trend for renal cancer if control areas (more than 50 km from a site) were included. That appears to have occurred because the SMR for the control areas was very low. The exposed areas more than 20 km from a uranium-processing facility had higher SMRs than those closer in, so it is unlikely that that was a uranium effect. RADIOACTIVE CONTAMINANTS IN DEPLETED URANIUM Because of the nature of its production, DU may contain trace amounts of fission products and transuranic elements, which have biokinetic and dosimetry characteristics different from those of DU and may be taken up by different organs and irradiate these organs preferentially. However, if they are present at all, the quantities of those contaminants are so small that the dose and risk associated with them are trivial and for practical purposes can be ignored. SUMMARY In summary, the following can be concluded with respect to the radiologic effects of DU: DU is only weakly radioactive and does not pose a reasonable risk of acute deterministic effects. For intakes of soluble DU, the chemical toxicity is of much greater concern than the potential radiologic effects. The external radiation hazard posed by DU is small, and long-term direct contact with bare skin is required to produce important effects. Inhalation of insoluble DU, if great enough, may provide a small but significant risk of lung cancer and possibly lymphoma due to irradiation of pulmonary-associated lymph nodes.
OCR for page 85
Review of the Toxicologic and Radiologic Risks to Military Personnel from Exposures to Depleted Uranium During and After Combat Epidemiologic studies have yielded inadequate evidence of a risk of cancer or other chronic diseases after exposure of Gulf War soldiers to DU. The epidemiologic data on workers exposed to uranium compounds are substantial (nearly 110,000 exposed, or potentially exposed, workers followed for long periods). Those data have weaknesses—such as little exposure-response information, inability to adjust for smoking habits, and no evidence on exposure of children or other presumptively susceptible populations—but the preponderance of the evidence indicates that there is not an appreciable risk of cancer in humans exposed to uranium. RECOMMENDATIONS Additional followup studies of exposed populations have the potential to improve knowledge of the health effects of DU. To permit an adequate assessment of the risks of cancer, renal toxicity, and other possible health effects faced by DU-exposed soldiers, a careful followup of the exposed groups should be continued, including the cohort of DU-exposed soldiers now being followed by the Department of Veterans Affairs. Furthermore, continued followup of the largest groups of workers and those who had the greatest highest exposure is recommended. If sufficiently detailed records are still available, it would be valuable to reconstruct individual exposures in a few of the largest studies with the greatest range of exposures. That would permit the evaluation of exposure-response analyses that would help to solve the problems in evaluating health end points caused by the “healthy-worker effect.” More information is needed on other exposures sustained in the uranium-cohort workplaces, such as exposures to solvents, other metals, and asbestos. A program of examination of subgroups of workers with high, medium, and low exposure to uranium, with appropriate matching on other risk factors, should be implemented for selected health-related end points and biomarkers, including renal function, genotoxicity, and bone enzymes and activity.