6. Agents Combined Toxicity. With the availability of toxic data for both sarin and cyclosarin, a simple scheme was used in the 2000 modeling to account for the toxicity of the sarin-cyclosarin mixture so that contours for the combined mixture could be determined. Since cyclosarins GPL dosage is one-third sarins, we defined an effective composite agent dosage, Dcomposite, as the sarin dosage, Dsarin, plus three times the cyclosarin dosage, Dcyclosarin:
Then we measured Dcomposite against the GPL dosage threshold for sarin, 0.0432 mg-min/m3, to determine the possible exposure contours. The composite FNE dosage was similarly treated, except cyclosarins FNE dosage is half sarins, so the factor of three in the formula above was replaced by a factor of two.
Figures 6 through 9 show the unions of the GPL contours predicted by the four model combinations for March 10, 1991, through March 13, 1991, respectively, and the unions of the FNE contours for March 10, 1991. The upper panel for each day shows the 2000 results, assuming dry gaseous deposition and agent decay; the bottom panel shows the 1997 results (CIA and DoD, 1997). The dots represent unit locations.
For Day 1, the 1997 and 2000 results differ primarily due to using the new meteorological fields, as indicated by slightly different cloud trajectories. During Day 1 the material was airborne for only 10.75 hours, mostly at night. For Day 2, the general location of the hazard areas was roughly the same in the 1997 and 2000 simulations; however, the combination of 55% less agent mass and inclusion of agent removal mechanisms modified by higher toxicity values for cyclosarin caused the 2000 hazard areas to be smaller than the 1997 results. Agent removal mechanisms also apply to Day 3. The hazard areas are greatly reduced for Day 4 because of relatively small emissions. As in 1997, the 2000 results did not identify any forces inside the FNE contours.
6. The combined General Population Limit contours predicted by all models
for Day 1, March 10, 1991
Top: The 2000 results assuming dry gasesous deposition and agent decay;
Bottom: the 1997 results
7. The combined General Population Limit contours predicted by all models
for Day 2, March 11, 1991
Top: The 2000 results assuming dry gaseous deposition and decay;
Bottom: the 1997 results
8. The combined General Population Limit contours predicted by all models
for Day 3, March 12, 1991
Top: The 2000 results assuming dry gaseous deposition and agent decay;
Bottom: the 1997 results
9. The combined General Population Limit contours predicted by all models
for Day 4, March 13, 1991
Top: The 2000 results assuming dry gaseous deposition and agent decay;
Bottom: the 1997 results
Figure 10 shows a view of the FNE and M8A1 alarm areas inside the GPL potential hazard area for the 2000 model results for March 10 (only day applicable). The M8A1 area encloses the FNE area, meaning the alarm is designed to protect forces before the FNE level is reached.
Figure 10. The 2000 Khamisiyah modeling First Noticable Effects and M8 alarm detection areas for March 10, 1991
In summary, the 2000 modeling results show much smaller and somewhat different shaped hazard areas, due to changes in: (1) mesoscale meteorological modeling, (2) the source term, (3) agent removal mechanisms, (4) the exposure thresholds for sarin and cyclosarin, and (5) their combined toxicity. Even though the exposure areas were smaller, they must be interpreted in conjunction with the unit location data, which also improved significantly since 1997. Based on the 1997 modeling results and the unit location data, DHS estimated that roughly 99,000 servicemembers were within the GPL contours, with none within the FNE contours. Based on the 2000 results and unit location data, roughly 102,000 servicemembers were estimated to be within the GPL contours, and again none within the FNE contours. Of these two servicemember counts, about 66,000 servicemembers were within the GPL contours in both the 1997 and 2000 simulations. The 2000 count is slightly higher than 1997s, despite the smaller 2000 hazard areas, because the 2000 unit location and personnel data are more accurate.
A separate DoD methodology report (DHS, 2000) describes the refined methodology of modeling the demolition in the Khamisiyah Pit. Like the 1997 analysis, a peer review panel consisting of Dr. Richard Anthes of University Corporation for Atmospheric Research, Dr. Steven Hanna of George Mason University, and Mr. Bruce Hicks of National Oceanic and Atmospheric Administration carefully reviewed the report. In general, the panel determined the revised methodology as satisfactory, representing an improvement over the methodology used in 1997 (Anthes et al., 2000).
V. HEALTH RISK CHARACTERIZATION
This health risk characterization compares exposures that may have resulted from the Khamisiyah demolition and those associated with health effects in controlled studies. To determine the possible exposures, the DoD team modeled the demolition in the Pit based on information about unit movements and assumptions about the amounts and rate of agent released and the meteorological conditions at the time of the releases. Section III presented the sarin and cyclosarin exposure guidelines and derived limits that would have resulted in adverse health effects. The DoD team used that information as a basis for assessing the significance of the modeled exposure levels.
B. Analyzing Exposures
Due to the nature of the possible exposures and toxicants in question, no appropriate exposure guideline levels exist with which to compare modeled exposure concentrations. The existing guidelines, i.e., General Population Limit (GPL) and worker population limit (WPL), are intended for public and/or occupational health professionals to protect these populations from adverse effects due to long-term exposure. These guidelines have margins of safety, but scientists do not consider the guidelines appropriate for assessing whether health effects will occur at exposures near the guideline value. Therefore, it is necessary to use the information from animal and human toxicological studies to estimate exposure levels that could result in adverse health effects.
This risk characterization addresses whether adverse health effects are likely to have occurred from short-term, low-level exposures. What cannot readily be addressed is whether health effects could arise several months or years after a low-level, short-term exposure because scientists have not conducted critical studies examining this issue.
This characterization discusses the possible exposures in terms of levels and limits applicable to a short term exposure. Furthermore, because military forces generally are in good physical condition without sensitive sub-populations (e.g., infants, sick, and the elderly), this characterization considers comparisons with appropriate occupational levels.
McNamara and Leitnaker (1971) reported a dosage (i.e., Ct (concentration multiplied by exposure time in minutes)) of 0.5 mg-min/m3 as sarins no-effect level in humans, at which less than 1% of a working population would show miosis, red blood cell cholinesterase (RBC-ChE) depression, rhinorrhea, or other mild symptoms. Other scientists used this no-effect level in conjunction with pharmacokinetic data to derive the WPLs and GPLs. In deriving these limits, the basic premise was to prevent accumulating a larger dose than the reported no-effect level (0.5 mg-min/m3) from repeated or chronic exposures with an additional safety factor for the GPL. Therefore, an acute exposure below 0.5 mg-min/m3 should cause miosis or any other acute health effect in less than 1% of a healthy worker population. In its review of the WPL and GPL limits supporting this conclusion, the Armys Edgewood Research, Development, and Engineering Center (ERDEC) (Mioduszewski et al., 1998) selected two human studies as critical:
US Army ERDEC reported the 50% effective Ct (ECt50) for miosis and rhinorrhea (time and concentration for half the exposed subjects to exhibit an effect) as less than 2 mg-min/m3, with an exposure of 3 mg-min/m3 producing miosis in most of the population (Sidell, 1992). Other DoD and CIA assessments of possible agent exposures at Khamisiyah have used an estimated exposure of 1 mg-min/m3 as a first noticeable effects level (i.e., rhinorrhea, chest tightness, dimmed vision, miosis) (CIA and DoD, 1997).
C. Uncertainties of the Risk Analysis
Some degree of uncertainty is present in every human health risk evaluation or risk assessment. Common to the methodologies used, this risk evaluation contains various uncertainties, including:
While uncertainties are clearly associated with the toxicological effects of sarin and cyclosarin on the forces at Khamisiyah, they are likely to be far greater when associated with the level of exposure and the existence of numerous possible confounders.
1. Toxicological Uncertainties
Uncertainties exist in the toxicological assessment of any noxious chemical and may include, but are not limited to, study design, data quality, selection of endpoints, extrapolation from high to low dose, extrapolation from animal to human, extrapolation from one exposure route to another, selection of uncertainty factors, database weaknesses, and intra- and inter-species differences. Results from a short-term in vivo study may be different from results obtained in a longer term in vivo study, and thus require substantiation. These uncertainties do not discount the usefulness of toxicological studies in predicting the likelihood of the occurrence of a given effect in humans; they simply suggest caution in interpretation.
Several toxicological uncertainties are of specific interest to this evaluation. The most significant of these is the relative lack of data related to low-dose, no-effects exposure to sarin; none for cyclosarin. During vapor exposure studies and unintentional vapor exposures, the first signs and symptoms are usually miosis, rhinorrhea, and/or chest tightness. Early studies often defined an individual as exposed based on the appearance of at least one of these symptoms. Persons in the same area, without such symptoms or health complaints, were not considered to be exposed (Sidell, 1992; Perrotta, 1996). Thus, only those who had clinical signs or symptoms, reflecting a higher exposure level than those considered in the present evaluation, would have been studied and documented. It is understood why initial studies on military incapacitating/lethal agents would not have been performed at higher doses. Not using higher doses is also present, to a lesser degree, when evaluating the effects of human exposure to organophosphorus pesticides. However, some significant effects of exposure to these agents, such as OPIDN, have been shown to occur only with very high exposures and are eliminated from consideration for the purpose of this evaluation.
Another uncertainty is the lack of long-term follow-up for exposed persons. Older doctrine had considered that if a victim recovered from the acute effects of nerve agent poisoning (or organophosphorus pesticide poisoning), he or she would have no adverse health effects. As a result of studies showing alterations in electroencephalographs (EEG)s and long term behavioral, psychological, and performance decrements resulting from organophosphate exposure, this doctrine has been called into question (Ecobichon and Joy, 1994). However, virtually all studies reporting these effects involved exposures that caused symptoms, except the one relating to long-term EEG changes in monkeys where the authors were unclear as to exactly what symptoms were present (Burchfiel et al., 1976). Two more recent studies on behavioral and performance effects associated with exposure to organophosphorus pesticides and soman in animals suggest that there is a threshold associated with percent reduction in ChE (Duffy et al., 1979; Duffy and Burchfiel, 1980; Burchfiel and Duffy, 1982; Sheets et al., 1997). Generally, a threshold is assumed for the dose-response curve for most neurotoxicants, based on the known capacity of the nervous system to compensate for or to repair a certain amount of damage at the cellular, tissue, or organ level (EPA, 1995). However, controlled studies of low-dose, asymptomatic agent exposures are not currently available.
Little toxicity data exist for cyclosarin. Toxicity studies on the effects of sarin and cyclosarin in combination suggest that they do not act synergistically (Clement, 1994). As described earlier (Section III), the available data suggest that cyclosarin is two to three times more toxic than sarin at the exposure levels of the FNE and the GPL used in this modeling. Thus, the DoD used a toxicity ratio of 1:3 (sarin:cyclosarin) for the GPL and 1:2 (sarin:cyclosarin) for the FNE.
The appropriateness of using cholinesterase inhibition as an endpoint for organophosphate toxicity has been debated. Small decreases in ChE and slight miosis have not been demonstrated as clinically adverse effects. ChE inhibition and miosis can reasonably be used as early markers of exposure. In the case of multiple toxic effects, preventing the most sensitive effect is generally considered protective based in part on the assumption that if the most sensitive effect is prevented, all other (toxic) effects are prevented (EPA, 1989). Recent studies in animals suggest that percent ChE inhibition, as a result of organophosphate exposure, may prove to be a useful marker even for such subtle effects as EEG changes and psychological/behavioral effects (Blick et al., 1987; Blick et al., 1988; Sheets et al., 1997); however sarin-specific data related to this point are not available. No post-exposure ChE levels were obtained for forces at Khamisiyah because no symptoms were reported.
2. Exposure Estimates
Determining whether an individual was actually exposed to a noxious chemical and, if so, to how much and for how long, is both difficult and associated with considerable uncertainty. Estimating an exposure in combat situations is no trivial problem. A large variety of factors can greatly impact the effective exposure that an individual receives. Temperature, humidity, skin moisture, exposed surfaces, type and condition of personal protective equipment, pretreatment (in the case of nerve agents), wind strength and direction, form of the agent (liquid or vapor), activity level of the soldier (at rest or running), and host susceptibility factors all make the estimation of field doses a complex problem (Perrotta, 1996).
Estimating the potential agent concentration in a vapor cloud from the demolition at Khamisiyah required extensive field testing to determine accurate source characterization of the demolitions and distribution of agents at Dugway Proving Ground, and then extensive modeling of the data. Modeling has inherent uncertainties itself. The estimated exposure levels are best estimates based on most reasonable assumptions that provide the possible hazard areas. Although both the laboratory tests at Dugway Proving Ground and US Army ERDEC and the field testing at Dugway Proving Ground have helped reduce some of these uncertainties, others remain.
3. Risk Assessment Process
The risk assessment process, including the use of equations for developing risk estimates or conclusions, involves numerous general assumptions. For example, the GPL and other limits used were developed from data points from the best available studies; and uncertainty factors, which are typically multiples of 10, were included in their development. These uncertainty factors have been used successfully in regulatory programs for decades, but concerns exist that the resulting exposure guidelines may be overstated in some situations and possibly understated in others. In addition, some generalizations in this and any assessment involve the weight of the exposed personnel, their breathing rates, exposure routes, and duration of exposure. For example, many exposure estimates assume that an exposed individual weighs 70 kg and breathes 15 l/min of air (EPA, 1989). Although EPA risk assessment methodologies have general utility in establishing protective limits, they are not appropriate for assessing whether effects may have occurred. An exposure level designed to protect the most susceptible individuals does not necessarily predict health outcomes for healthy workers or servicemembers possibly exposed to nerve agent at Khamisiyah. Finally, confounders exist and the number of permutations can be numerous.
D. Summary of Risk Characterization for US Forces at Khamisiyah
A comprehensive toxicity assessment for sarin and cyclosarin was performed, regulatory guidance and relevant data reviewed, and a risk characterization completed. A review of existing toxicological data suggests that the development of long-term adverse health effects associated with exposure to sarin (or cyclosarin) at concentrations below those needed to cause short-term signs, symptoms, or inhibition of ChE is unlikely. However, human studies specifically designed to address long-term effects from low-level exposure to sarin are not available. Studies, such as those for OPIDN, that are clearly negative for a particular health effect at high levels, provide confidence that no increased health risk exists for these effects at the lower doses. The first noticeable effects of exposure to sarin and cyclosarin, assuming the exposure is not high enough to cause significant hypoxia, are readily reversible. We have found no reports of symptoms among the forces at Khamisiyah. OPIDN, which is of concern with some organophosphorus pesticides, has been reported only in animals after exposure to very high levels of nerve agents (e.g., multiple lethal doses with antidotal treatment).
While evidence suggests that psychological, behavioral, and EEG changes are fairly common with exposure to high doses of organophosphate chemicals, including nerve agents, data at lower doses are sparse or nonexistent. None of the animal or human studies regarding sarin-induced EEG changes fully address the question of chronic effects from short-term asymptomatic exposure. Whether the EEG changes occur in humans exposed to levels below those required to produce mild signs or symptoms or significant ChE depression is not known. The CDC stated that "the EEG changes reported after intoxication with sarin were considered to be of questionable significance, given the difficulty of demonstrating such changes and the absence of clinically significant effects even when EEG changes are present" (CDC, 1988). The organophosphate data that are available at low doses, including monkey studies with soman and the recent well-controlled study of six organophosphates in rats, suggest that these effects have a threshold that correlates with ChE inhibition (Sheets et al., 1997; Blick et al., 1987; Blick et al., 1988). Collectively, these data suggest that there is a threshold for OPIDN effects and that low-level exposures to sarin or cyclosarin resulting in no symptoms or significant inhibition of ChE are unlikely to produce any long-term neurobehavioral effects. No post-exposure ChEs were obtained from service members at Khamisiyah because there were no reported symptoms of nerve agent exposure.
Evidence indicates that sarin does not have carcinogenic, mutagenic, or teratogenic properties. Therefore, no increases in birth defects or cancer would be expected from low-dose, asymptomatic exposure to sarin. Available data suggest that exposure levels below doses that produce convulsions do not cause cardiomyopathy. Other than sarins well-known anticholinergic effects, no general adverse health effects from low-level exposure to sarin have been reported, nor have any reports of immunotoxicity associated with exposure to sarin been identified. An examination of several follow-up measurements conducted on the individuals exposed to anticholinesterase chemicals (including sarin) during experiments at Aberdeen Proving Ground in Edgewood, Maryland, did not find significant increases in hospital admissions, self-reported medical problems, impairments, malignancies, or other adverse health outcomes (NRC, 1985). Review of current toxicological and medical data indicate those long-term health effects from brief, low-level, asymptomatic exposures to sarin are unlikely to result in significant ChE depression.
Uncertainties in this risk assessment include the lack of human data addressing the cholinergic and noncholinergic effects of very low dose exposure to sarin and uncertainties inherent in modeling and retrospective exposure assessment. Some uncertainties and confounders will remain even though the best available data and the best estimates and conservative modeling and risk assessment methods were used in this analysis.
Based on the available data about the apparent health status of the service members at Khamisiyah, modeling and exposure data, and toxicological data, DoD concludes the exposures the service members possibly received were almost certainly below what would be expected to cause acute health effects, such as miosis, or long-term effects, such as OPIDN. Although exposure to sarin itself at the estimated concentrations may not result in any adverse health effects, this finding does not preclude the possibility of adverse health effects resulting from any number of combinations of noxious chemicals and/or other stressors.