While the duration of actual combat engagement during the first Gulf War in February 1991 was relatively short, measured in days, the legacy of adverse health effects presumed related to it has been disproportionately lengthy. The constellation of symptom complaints in returned troops termed ‘Gulf War Syndrome’ or more generally, ‘unexplained illness’, has received the bulk of both scientific and public attention. However, a small collection of ‘explained’ adverse health outcomes have also been reported over the 15 years since the War's end. One of the best-characterized examples in this category involves the cluster of DU ‘friendly fire’ incidents and the DU-related health effects accrued to those soldiers who were its victims.

During one 48h period of the Gulf War in February 1991, approximately 115 US tank crew members in 20 US vehicles (six Abrams tanks and 14 Bradley fighting vehicles) were mistakenly fired upon by other US forces using DU penetrators (OSAGWI 2000). DU is a high-density metal, lending it armour-piercing capability. At the time a DU penetrator strikes a target, its ‘pyrophoric’ character and the high temperatures generated by the impact promote small particles of DU to ignite and form a DU oxide dust that deposits on and within the targeted vehicle. Small shards of DU metal are also produced when the penetrator pierces the target's armour (AEPI 1995; Toohey 2003; Parkhurst et al. 2005).

Eleven fatalities resulted from these friendly fire incidents and approximately 50 casualties required medical attention (OSAGWI 2000). Traumatic injuries were the primary consequence, although the potential for significant wound contamination, inhalation and ingestion exposures to DU dust also occurred to both tank crew members and rescuers (Parkhurst et al. 2005), and some soldiers were hit by DU metal fragments that remain embedded in soft tissues. Each soldier's level of exposure to DU dust and/or shrapnel by these potential exposure routes would have differed depending on his location and movements subsequent to the impact of the munitions. Thus, exposure assessment has been an important component of the Baltimore VA DU Follow-Up Surveillance Program.

In 1993, the US Department of VA in conjunction with the Department of Defense initiated a medical surveillance programme to assess the health of the Gulf War veterans involved in the friendly fire incidents described above. The cohort defined as ‘DU-exposed’ in this programme is composed of soldiers with the highest potential for significant DU exposure, i.e. those who were in or on the Army vehicles when they were hit. In addition to assessing health, the programme was charged with evaluating and developing methods to measure uranium body burdens under different exposure scenarios, including the novel exposure mode of metal released from fragments present in soft tissue. As well, the surgical management of embedded fragments and inherent medical consequences were to be evaluated.

The surgical management of shrapnel in place in 1991 for these affected servicemen was based on a standardized protocol that considered metal fragment size and location in the body as the primary determinants for removal. Additionally the surgical morbidity expected from extensive fragment removal was also calculated. Many soldiers possessed multiple fragments, from 10 to more than 30, comprised of dimensions ranging from 1mm to much larger, approximately 20mm in diameter (Toohey 2003). Easily accessible fragments were removed, but because of the extensive traumatic injuries sustained by many of these soldiers, the smaller and more difficult to reach surgically were left in place by the surgeons who first cared for these soldiers in Iraq, Germany and during their initial care at Army Medical Centers in the US.

A biennial schedule of surveillance commenced in late 1993 for approximately 100 survivors of the Gulf War friendly fire events. To date, 74 of the soldiers in the ‘DU-exposed’ cohort have been evaluated, out of the approximate 90 for whom contact information is available. Although all expenses for the evaluation are paid for, including a reimbursement for lost wages during the 3-day, in-patient assessment, a number of victims of the original friendly fire incidents have not felt the need to participate in health evaluations. The number of participants involved in each of the biennial visits to date is shown in table 1. Many of the ‘DU-exposed’ cohort have participated in all of the visits, providing the ability to examine changes in clinical parameters over time. In 1997, using an enhanced protocol, the health status of the soldiers in the ‘DU-exposed’ cohort was reassessed and compared with Gulf War deployed soldiers with no known potential for DU exposure as judged by their answers to questions on a DU exposure questionnaire which provided information on their location and activities during the Gulf War (McDiarmid et al. 2000). Subsequent analysis of DU-exposure status of this ‘Non-DU-exposed’ cohort was confirmed by whole body radiation counting (McDiarmid et al. 2000) and urine U isotopic analysis (Gwiazda et al. 2004).

Table 1 Summary of surveillance visits for Gulf War soldiers (74 unique DU-exposed cases have been evaluated).

The enhanced health surveillance protocol conducted in 1997 was based on the known and presumed toxicology of uranium and the unique nature of the on-going absorption of U from embedded shrapnel via dissolution of the metal fragments. With minor modifications, this enhanced protocol continues as the basis for the DU Program biennial surveillance visits.

DU poses both radiological and chemical hazards to human health (The Royal Society 2001, 2002). Created as a by-product of the uranium enrichment process, which involves removal of natural U's more radioactive isotopes: U²³⁵ and U²³⁴, DU is weakly radioactive with a radioactivity approximately 40% lower than that of natural U (AEPI 1995; table 2). Assessments of radiation doses from DU conducted by The Royal Society (2002) indicate that high energy, poorly penetrating alpha emissions from DU contribute the most to tissue dose rates, with effects primarily on tissue in immediate contact with the metal. Beta particles released from the daughter product ^234mPa which is in secular equilibrium with ²³⁸U will penetrate further into tissues, but dose rates are over tenfold lower than those from the alpha particle emissions (The Royal Society 2002).

Table 2 Comparison of the radioactivity of natural and depleted uranium.

Radiation dose estimates for Gulf War soldiers in or on vehicles hit by munitions containing DU penetrators have also been calculated using DU oxide air concentration data generated by the Capstone DU Aerosols: Generation and Characterization Study conducted by the US Army Heavy Metals Office (Parkhurst et al. 2005). Based on estimated inhalation exposure doses to mixtures of DU oxides of different solubilities, the most likely and upper bound exposure scenarios yielded maximum calculated 50 year committed effective doses [E(50)] of 6.0 and 8.7rem, respectively, which exceed the US annual occupational radiation exposure dose limit of 5rem total effective dose equivalent, but are less than the Nuclear Regulatory Commission (NRC) planned special exposure limit of 10rem (10 CFR 20). Estimated 50 year committed effective doses to tissues [HT(50)] were highest for the lung by a factor of at least 10 compared to other tissues, including the bone surface, kidney, red marrow and liver, which is consistent with risk assessments developed for likely Gulf War exposures by The Royal Society (2002). The most likely and upper bound HT(50)s for lung reported using data from the Capstone report were 44 and 61rem, respectively (Parkhurst et al. 2005).

The health effects of exposure to depleted, natural and enriched forms of U have been extensively investigated using animals studies (Leggett 1989; ATSDR 1999; The Royal Society 2001). Work that started in the 1940s as part of the Manhattan Project provided a strong database for establishing ‘safe’ exposure levels for workers in uranium industries (Voegtlin & Hodge 1949, 1953; Tannenbaum 1951; Hodge et al. 1973). As a whole, these studies have shown that health risks from exposure to natural U are highly dependent on the solubility of the U compounds involved and the timing and route of exposure. Acute and chronic inhalation exposure to insoluble U compounds can cause radiation-induced lung tissue damage; for acute inhalation and oral exposures to soluble U compounds, the kidneys are the primary target organ due to uranium's chemical toxicity (ATSDR 1999; The Royal Society Report 2001). Recent studies involving chronic exposure scenarios have also demonstrated the sensitivity of the kidney to U. Both animal experiments (Gilman et al. 1998a,–c) and studies of occupationally (Saccomanno 1982; Thun et al. 1985) and environmentally (Zamora et al. 1998, 2002; Kurttio et al. 2002) exposed populations have reported renal dysfunction following chronic inhalation and oral U exposures.

Although in vitro genotoxicity (Lin et al. 1993; Miller et al. 1998a,b, 2002a,b, 2003; Yazzie et al. 2003; Stearns et al. in press) and in vivo animal carcinogenicity studies (Hueper et al. 1952; Leach et al. 1973; Cross et al. 1981; Mitchel et al. 1999; Hahn et al. 2002) using natural U and DU raise some concern regarding cancer risks in DU exposed veterans, studies of occupationally exposed cohorts do not support the extrapolation of the results of these studies to humans. There is poor evidence for an increased risk of lung cancer in uranium-exposed cohorts, except for uranium miners whose risk is attributed to their co-exposure to radon, a more intensely radioactive constituent than natural uranium by a factor of 10000 (Kathren & Moore 1986; Samet 1989; Kathren et al. 1989; Samet et al. 1989). In addition, although bone is known to be an important long-term storage depot for uranium (Kathren et al. 1989), excesses in bone cancer or leukaemia have not been reported in exposed cohorts (IOM 2000). Although questions remain regarding the similarity between occupational exposures and those experienced by Gulf War soldiers, these results are consistent with the low risk of increased cancer predicted by The Royal Society (2001, 2002) for Gulf War exposures.

Based on the extensive literature available on U and DU toxicity, health concerns in DU-exposed veterans have focused on the chemical effects of this metal as well as its radiological toxicity. In particular, effects of DU on the kidney and other target organs known to be affected by other heavy metals have been examined. The expanded surveillance protocol developed for the Gulf War DU cohort for the 1997 visit included neuroendocrine, immunologic and reproductive parameters, in addition to renal markers and measures of genotoxicity (McDiarmid et al. 2000, 2001, 2004a,b, 2006).

Quantitative exposure assessment is a critical component of surveillance programmes designed to identify health effects related to chemical exposures. In the case of DU, sensitive methods needed to be developed to accurately determine systemic exposure to DU. In addition, although the radioactivity of DU is lower than that of natural uranium, the concern that biological effects of low-level radiation could lead to health consequences established the need to quantify radiation as well as chemical exposure.

Despite the low background level of this whole body counting chamber, only nine of the 27 ‘DU-exposed’ Gulf War veterans in the Baltimore VA friendly fire cohort exceeded the detection limit for this technique (figure 1). All of the veterans with whole body radiation scores above background were known to have embedded fragments. None of the ‘Non-DU-exposed’ veterans seen in 1997 gave results above the detection limit. For soldiers with embedded fragments, body scan data also provided information on the distribution of DU fragments in the body and direct radiation counting of small areas known to have fragments provided a means of verifying whether the fragments were composed of DU or non-DU metal for many of the veterans.

Figure 1 Whole body radiation counting results for ‘DU-exposed’ and ‘Non-DU-exposed’ Gulf War veterans participating in the Baltimore VA DU Follow-Up Program. Participants are ranked from low to high along the x-axis based on their (more ...)

Radiation dose estimates have also been determined using urine U excretion data from veterans with shrapnel and the ICRP (1995) biokinetic model for U. Shown by Leggett & Pelmar (2003) to be applicable for estimating the time-dependent rates of release of U from embedded fragments from urinary excretion rates, the ICRP biokinetic model was used with urine U excretion data collected from veterans for up to 10 years post-exposure to predict the rate of accumulation of U in tissues from back-calculated rates of release of DU from shrapnel (Squibb et al. 2005). The maximum upper estimate of the lifetime (50 years) effective dose determined using this method was 0.06Sv, which is slightly above the NCRP (1993) public lifetime limit of 0.05Sv but below the occupational limit of 1Sv. The tissue with the highest upper estimated lifetime committed effective dose was the bone surface (0.9Sv), followed by the kidney (0.3Sv), liver (0.1Sv) and red marrow (0.1Sv). These are all well below the NCRP (1993) occupational lifetime tissue committed effective dose of 25Sv. It is important to note that these radiation dose estimates are only for DU released from embedded shrapnel and do not include radiation doses resulting from inhalation, ingestion or wound contamination by DU oxides generated during the friendly fire incidents.

To more accurately assess low-level exposures to DU and specifically quantify DU versus natural U exposure, a more sensitive analytical method than KPA was developed using an inductively coupled plasma dynamic reaction cell mass spectrometer (ICP-DRC-MS) (Ejnik et al. 2005). The detection limit for quantifying total U in urine using the ICP-DRC-MS technique is 0.1ngl⁻¹, an improvement in sensitivity of 200-fold compared to KPA analysis. Accurate isotopic analysis for calculating the U²³⁵/U²³⁸ ratio in urine specimens is also possible using this quadrapole ICP-DRC-MS technique for samples with U concentrations as low as 10ngl⁻¹ (Ejnik et al. 2005). With this capability, it is possible to calculate the amount of DU in a urine sample that also contains natural U at total U concentrations of 10ngl⁻¹ and above, thus providing direct evidence of DU exposure. The use of the dynamic reaction cell in this methodology provided a simple means by which to remove a polyatomic interference at m/z 234.8 present in some urine samples (Gwiazda et al. 2004), which artificially elevates the U²³⁵/U²³⁸ ratio and can falsely indicate the presence of enriched U. Other techniques developed for solving this problem involve magnetic sector high-resolution ICP-MS instruments or more extensive sample preparation to first remove the uranium from matrix constituents (Gwiazda et al. 2004). A QA/QC comparative study involving multiple laboratories using either the ICP-DRC-MS or high-resolution magnetic sector instruments (Gwiazda et al. 2004; Ejnik et al. 2005) showed excellent correlations between these techniques for total U uranium and U²³⁵/U²³⁸ ratio results at urine U concentrations as low as 10ngl⁻¹.

Since whole body counting was not a sensitive enough method for quantifying DU exposure in the Gulf War friendly fire cohort, urinary excretion of U was established as a measure of systemic U exposure arising from the in situ release of DU from embedded fragments and/or tissue stores.

The results of 24h total urine uranium analysis for the sub-cohort of Gulf War ‘DU-exposed’ veterans seen in the 2005 surveillance visit are shown in figure 2. Uranium concentrations for this group ranged from 0.003 to 44.1μgg⁻¹ creatinine. All values equal to or over the upper limit for dietary exposure to U for the general population from drinking water (0.365μgl⁻¹, ICRP 1974) were from participants with known retained shrapnel fragments and U isotopic analysis results indicated the presence of DU in their urine samples. Five individuals with a history of shrapnel had urine U concentrations below the 0.365μgl⁻¹ value, but still had detectable DU in their urine. The urine U concentration of the one veteran with no evidence of shrapnel but with measurable DU in his urine was low (less than 0.1μgg⁻¹ creatinine), which is consistent with a wound contamination, ingestion and/or inhalation only exposure to DU. One individual with known shrapnel injuries had no evidence of DU based on isotopic ratio results, suggesting that either his shrapnel is not composed of DU or that release of DU from the shrapnel is so low that it is not detectable. Others in this cohort have total urine U concentrations below the limit of detection for DU analysis.

Figure 2 Uranium concentrations in 24h urine specimens distributed from low to high for Gulf War veterans participating in the DU Follow-Up Program surveillance visit in 2005. The top two lines (65.1and 9.1μgl−1) represent (more ...)

The concentration of uranium in a 24h collection of urine, expressed as micrograms per gram of creatinine, has been used in this surveillance programme as a measure by which to establish two groups (high versus low urine U) for comparing differences in clinical parameters for their relationship to urine U excretion. ‘High’ has been defined as urine uranium concentrations greater than 0.10μgg⁻¹ creatinine. While there is no generally accepted standard normal urine uranium value, we chose a value intermediate between, at the low end, several estimates of mean urine uranium concentration in non-exposed populations in the literature (6–22ngl⁻¹, Dang et al. 1992; Medley et al. 1994; Ting et al. 1999; NHANES 2003); and at the high end, upper dietary limits due to natural uranium in soil and ground water (up to 0.365μgl⁻¹) (ICRP 1974). Over the past 12 years, mean urine U concentrations in the high and low-exposure groups of Gulf War veterans participating in each of the six surveillance visits have been consistent (figure 3). Urine U concentrations have not dropped significantly in the veterans with embedded DU fragments, demonstrating that these individuals are experiencing chronic U exposure due to release of U from their fragments.

Figure 3 Mean, maximum and minimum urine uranium concentrations for 27 Gulf War veterans with at least four urine samples submitted between 1993 and 2005. Results for individual veterans are distributed from low to high based on their mean urine U concentration. (more ...)

Table 3 Summary of significant observations in clinical results for five surveillance visits from 1994 to 2003 (L, Low urine uranium group (U<0.1μgg−1 creatinine); H, High urine uranium group (U≥0.1μgg (more ...)

The mean retinol-binding protein (RBP) urine concentration has been higher in the high U group in the two most recent evaluations, though the difference between the groups has not reached statistical significance at the p<0.05 level. This observation of a higher excretion of this low molecular weight protein is biologically plausible, as U can exert an inhibitory effect on protein reabsorption by renal proximal tubule cells. We are following this RBP observation carefully, as a potential ‘sentinel’ marker of proximal tubular effects from U (McDiarmid et al. 2006; Squibb et al. 2005).

While there is good biologic plausibility for U to be genotoxic, based on animal data (Hueper et al. 1952; Leach et al. 1973; Cross et al. 1981; Mitchell et al. 1999; Hahn et al. 2002), some epidemiologic data in exposed workers (Martin et al. 1991) and, more recently, the evidence reported from in vitro studies (Lin et al. 1993; Miller et al. 1998b, 2002a,b, 2003; Yazzie et al. 2003; Stearns et al. 2005), the results of our studies are mixed. The HPRT frequency results, when analysed using the U low/high stratified grouping, are not significantly different, although mean mutation frequency (MF) was higher in the high U group in both 2001 and 2003 (McDiarmid et al. 2004a,b, 2006). There appears to be an inflection point in the mean MF distribution when MF data are plotted against the natural log of urine U concentrations. This inflection is driven by a small number of veterans with higher urine U concentrations, suggesting that there may be an exposure threshold for the increase in MF. We have not, however, observed consistent changes in sister chromatid exchange (SCE) or chromosomal aberrations (CA) (table 4). A more descriptive level of analysis of the types of mutations observed in the HPRT assay are underway to assist in interpreting our findings. In the 2005 assessment, we added fluorescent in situ hybridization (FISH) assessment of peripheral blood lymphocyte DNA to examine specific chromosomal ‘hot spots’ for genotoxic injury resulting from U exposure. These results will be helpful in assessing the in vivo genotoxicity of DU in humans.

Table 4 Summary of genotoxicity parameters (L, Low urine uranium group (U<0.1μgg−1 creatinine); H, High urine uranium group (U≥0.1μgg−1 creatinine); ns, no significant differences (more ...)

The Baltimore VA DU Follow-up Program continues to conduct health surveillance in a cohort of Gulf War soldiers exposed to DU when they were in or on vehicles hit by friendly fire involving munitions with DU penetrators. Urine U concentrations remain elevated in veterans with embedded DU fragments, demonstrating a chronic systemic exposure to U in this group of soldiers. Not all soldiers in the ‘DU-exposed’ Gulf War friendly fire cohort have embedded DU fragments; thus potential health effects from DU exposure by inhalation, ingestion and wound contamination alone are also being monitored through this programme. With the exception of the elevated urine U excretion, no clinically significant, expected U-related health effects have yet been identified in veterans with or without embedded fragments, though subtle changes in renal function and genotoxicity markers in soldiers with urine U concentrations greater than 0.1μg⁻¹ creatinine have been observed.