Comments on the Baltimore VA Study of Gulf War (1991) Veterans
by Glen D. Lawrence, Professor of Chemistry and Biochemistry
Long Island University, Brooklyn, NY

Hooper et al [Health Phys, 77(5), 512-519, 1999] describes the initial study population. DoD provided a list of 68 individuals who were wounded during Feb 1991 friendly fire incidents. Of these 68, 48 were contacted by mail and telephone and invited to participate. Of these, 33 agreed to participate, i.e., less than half of those on the friendly fire list. One must also question why only 48 of the 68 were contacted to participate, i.e., 20 or about 30% were omitted from the very start. Of the 33 participants, 23 reported they had been told they were wounded by shrapnel. Of these 23, 15 had shrapnel still present on x-ray, while it was not seen in the other 8.

Two others were found to have shrapnel but were not aware of it before the x-rays. From this information, one would deduce that there were a total of 17 soldiers with embedded shrapnel. Later it is seen that 2 soldiers with shrapnel were not excreting high levels of uranium in the urine, suggesting that the shrapnel was not depleted uranium. There was no indication whether these two with shrapnel who were not excreting high levels of uranium are the same two who were unaware of any shrapnel in their bodies. The nature of the shrapnel fragments was never established. However, clinical laboratory tests compare the group "with shrapnel" vs those "without shrapnel" without giving the number in each group. Consequently, the statistical comparisons are skewed (or skewered?) by including two individuals that apparently do not have depleted uranium shrapnel. The authors conclude: "These data show no evidence of adverse clinical outcomes associated with uranium exposure at this time in these individuals." One must keep in mind that they are comparing "groups of individuals" and not reporting whether there are, in fact, individuals with adverse clinical outcomes.

Blood and urine samples from these veterans in the study were run through standard clinical tests. The "Methods and Materials" section of the paper indicates alkaline phosphatase was measured in the blood, although alkaline phosphatase data is not reported anywhere in the paper, yet the other clinical test parameters are. Alkaline phosphatase levels are elevated when there is kidney damage from heavy metal poisoning. Is the missing data an oversight, or was it removed from the paper afterwards for some reason and the authors neglected to remove the fact that alkaline phosphatase was measured from the Methods?

Urine uranium is expressed per gram of creatinine, which is a generally accepted practice. The paper mentions there were two urine specimens excluded from analysis "because of nonphysiological creatinine and/or missed volume measurements." Since there were only 2 specimens it seems odd that the term "and/or" is used and that "missed volume measurements" was a factor for exclusion in this case, since uranium levels are reported on the basis of creatinine levels to avoid the necessity for volume measurements. Nonphysiological levels of creatinine are an indication of renal failure. Consequently, discarding these samples removes those participants with possible kidney damage from the study, or at least from the data for statistical comparisons used to draw their conclusion that there were no adverse clinical effects.

Urine specimens were collected at the Baltimore VA in 1993/94 and follow up urine specimens were collected in 1995 by sending empty bottles to the vets and having them return the specimens to the Baltimore VA by "overnight express" in June, July and August (hottest months of the year in the U.S.). Creatinine must be measured immediately after urine collection or urine must be refrigerated, otherwise creatinine degradation will occur within hours making the data unreliable. These samples were used to determine that spot collection is as reliable as 24 hr urine collection.

The "group" with shrapnel (2 with unknown shrapnel) had mean urine uranium excretion 150 times higher than the group without shrapnel (4.47 vs 0.03 microg U per g creatinine, or >10 microg U per liter of urine for shrapnel group). [Note: spot urine uranium concentration reported for individuals with shrapnel (n=8) was 0.04, 0.84, 3.95, 6.14, 11.08, 22.31, 31.39 and 35.26 microg/g creatinine, giving a mean of 13.87, 3 times higher than 4.47 as above, although 9 individuals with shrapnel are not included in this set (Table 4)]. By comparison, the limit for occupational exposure of workers in the uranium industry is 0.8 microg per liter, above which medical attention is given. This becomes an important point in later papers.

Another paper by McDiarmid et al [Environ. Res., Sec. A, 82, 168-180, 2000] compares 29 exposed soldiers (of the previous 33, of which 15 had DU shrapnel) with 38 nonexposed veterans of the Gulf War. The unexposed group was "identified through a number of sources including Department of Defense hospitalization data bank for Gulf War service people," i.e., vets who had been hospitalized for some reason. Data in this paper make it clear that these are not healthy vets, with more than 50% of those shown in Table 2 reported to have nervous system problems (data in Table 2 lists only 21 of the 38 in the nonexposed group, 11 with nervous system problems). In the paper, levels of neuroendocrine hormones are compared between "exposed" and "nonexposed" veterans, yet the comparison group has a high incidence of neurological problems. The paper shows that of the 4 hormones measured (prolactin, FSH, LH and testosterone), only prolactin shows a significant difference between high uranium and low uranium excreting veterans. However, there was a statistical slight-of-hand done, where a median split was done for hormone levels and then comparisons were done between high and low uranium excreters in the high and low hormone level groups. Consequently, even though mean levels of uranium excretion in the low FSH group was 3 times greater than uranium excretion in the high FSH group (for high uranium U excreters), there was not a statistical difference. Why wasn't there simply a comparison of hormone level with urinary uranium excretion (which would be the normal statistical comparison)?

This paper also compares renal function parameters (Table 4) for a low uranium excretion group to a high uranium excretion group, although a cutoff value of 0.10 microg U/g creatinine is used to distinguish between low and high uranium, even though: "All values over 0.8 microg/g creatinine were in veterans with retained metal fragments." This places 3 individuals in the group of high uranium excreters, even though they have no shrapnel and are excreting levels of uranium that are within the range for people exposed to natural levels of uranium. In other words, they are adding people who are excreting normal levels of uranium to the group excreting high levels of uranium, which skews the statistical comparisons to result in no statistical significance between groups. There is a large gap between levels of uranium excretion by those with DU shrapnel and those without DU shrapnel (there were still 2 individuals with shrapnel excreting less than 0.05 microg U/g creatinine, as described for the Hooper et al paper above, indicating the shrapnel is not DU). Anyone with scientific integrity would have drawn the line between high uranium and low uranium excretion in the gap between individuals with DU shrapnel and those without shrapnel. There was no basis for drawing the cutoff line at 0.1 microg U/g creatinine.

Another paper [McDiarmid et al, J Occup. Environ. Med, 43(12), 991-1000, 2001] reports data for an enlarged cohort of friendly fire vets. This paper reports on 50 male vets exposed to DU from friendly fire who were examined in 1999; 21 returning (mentioned above) and 29 new. The paper indicates that only about 1/3 of the estimated 120 vets involved in several friendly fire incidents had been located (somehow the original list of 68 (see Hooper et al paper above) got expanded to 120. Figure 1 shows data for 23 returning and 29 new participants [note: this is a total of 52, not 50 as indicated in Materials and Methods. Again a couple of participants seem to be shuffled around (included or not included in data sets) without explanation]. A urinary U excretion level of 0.8 microg/L is shown as the DoE value for the limit for exposed workers. A value of 0.36 microg/L is the maximum observed for dietary/drinking water contribution for the general population; there are no subjects between these two lines. The researchers chose to use a value of about 0.1 microg/g creatine in this project to separate low vs high U excretion, which adds 3 subjects to the high group from the low group if the 0.36 or 0.8 cut off points were used; this results in a 25% increase in the high U group. Furthermore, one of the subjects added to the high U excretion group in the new group of participants does not have shrapnel. There are 17 returning with "current or past history of shrapnel (metal type unknown)." i.e., this implies 100% of the original shrapnel inflicted vets returned. It is also noteworthy that 2 of the 11 returning individuals with high U had variations in the 3 urinary U measurements (1994, 1997 and 1999) of about 100 fold. This is a huge variation. The graphs expressing urinary U excretion values are plotted on a log scale, making a 10 fold difference look relatively small. Uranium levels differ by more than a factor of 1000 for all participants.

Another paper [McDiarmid et al, Health Phys., 80(3), 270-273, 2001] reports urinary U concentrations in an enlarged cohort of 169 vets who submitted 24 hr urine samples for analysis. From this cohort, 12 exhibited high urine U (>0.05 microg/g creatinine). A repeat test of 6 of these 12 resulted in 3 with uranium in the low range (specimen not submitted by the other 6). It should also be pointed out that the cutoff between high vs low U in this paper is half that in another paper (see above) and well below the level observed in the general population with high U in drinking water (stated as 0.36 microg/L (or ca. 0.36 microg/g creatinine). If the 0.36 microg/L were used, only one of these subjects would be in the high U category, and that person did not submit a second urine specimen for analysis. There were also two individuals in this cohort with reported (not confirmed) shrapnel in the body, one in the high U group and one in the low U group. There is no indication which subject in the high U group had retained shrapnel. They conclude that only embedded DU shrapnel is a positive predictor of high urinary U levels, yet they have placed at least 9 individuals in their high urinary U group, when only one of them has retained shrapnel.

This paper does not shed any light on health of this cohort, nor does it give any coherent discussion of the data or facts. In the conclusion they state that: "elevated urinary uranium results are unlikely to be observed [9 years after the war], except for those veterans with retained DU metal fragments." Yet they have categorized 12 with high urinary uranium by lowering the cutoff, decreasing that number to 9 with a repeat analysis of 6 out of 12. Why weren't the other 6 analyzed? However, only one of these would truly be considered high U, and at best 2 others have urinary U excretion above 0.1 microg/g creatinine. The authors also offer assurances that, "as a group", they have not developed kidney abnormalities, lung cancer or any classical uranium-related adverse outcome. The reader is cautioned that "as a group" means there was not a statistical difference between groups, but the high uranium group included many low uranium excreters, which would confound the statistics. They don't mention bone tumors, and admit that there were neuropsychological and neuroendocrine differences in the DU exposed individuals, although they say "these findings are of unclear clinical significance."

Conclusions

The data are presented in a way that makes it difficult to determine wether high levels of DU in the body are causing significant health effects. The data compare "exposed" soldiers with high vs low levels of urinary U. One paper [Environ Res] reports that 90% of those exposed to DU (n=29, from the original cohort of 33) have active medical problems, albeit injuries leads the list, but there are large numbers of other medical problems, including musculoskeletal, cardiovascular, psychiatric and nervous system problems. The non-exposed vets (n=21, although Materials and Methods indicates 22 or 38) also had a high level of active medical problems (71%). Consequently, these papers are comparing the physiological and neuroendocrine parameters of two groups of unhealthy individuals, some with high levels of U in the urine and some with low levels of U in the urine, although many in the high uranium group should be in the low uranium group. It is important to keep in mind that those individuals with low excretion of U in the urine at this time may have been exposed to high levels several years before their health effects were being monitored.

One vet from this study appeared in the documentary, "The Invisible War: The politics of radiation." He described the shrapnel that he had in him, that he was in the VA program, and that when he complained about his left kidney bothering him, they would look at the right kidney. He also described a bone tumor he had in his shoulder that was removed and he requested a sample of the tumor be sent to a university research center, but the VA refused to send the material to an independent investigator. The vet was a little vague about details in the video, just as the researchers conducting the Baltimore VA study are a little vague about details in their papers.