The U.S. Department of Defense (DOD) developed a process to derive an occupational exposure level (OEL) for the solvent trichloroethylene (TCE) because the agency has concerns that the current occupational standards are inadequate to protect its workers from exposure to the solvent’s vapors. Occupational exposure to TCE occurs from its use as a degreasing agent, as a chemical intermediate in refrigerant manufacture, and as a component of spot removers and adhesives. TCE is also a common environmental contaminant at many industrial and government facilities, including those of DOD. Workers can be exposed to TCE at these sites when contaminated groundwater and soil migrate under buildings and TCE evaporates into the air. TCE is a human carcinogen and also has noncancer effects on the liver, kidneys, neurological system, immune system, reproduction, and development.
DOD’s approach, documented in a draft titled Trichloroethylene: Occupational Exposure Level for the Department of Defense, incorporated elements of narrative review and systematic review, and introduced evaluation techniques (e.g., study applicability tool, physiologically based pharmacokinetic [PBPK] model, Bayesian approaches to address uncertainty) to determining an OEL for the chemical. DOD asked the National Academies of Sciences, Engineering, and Medicine to convene a committee to review the scientific and technical basis of DOD’s proposed approach. The committee was to
- Provide an analysis of the overall approach and suggestions on individual components of the report that may lead to improvements in the accuracy of the proposed process. Individual components include the literature review, evidence synthesis based on weight of evidence [WOE], point-of-departure derivation, use of physiologically based pharmacokinetic modeling, use of extrapolation tools, and other elements of the process.
- Determine if the process in deriving an OEL for TCE, including the WOE approach to determine relevance of controlled laboratory studies and overall approach corroborating alternative lines of evidence, is scientifically sound.
- Determine if the derived OEL value is supported by the toxicity information and has followed the WOE approach outlined in the report and provide a summary opinion of the approach and the scientific support for the derivation of the OEL.
- Determine whether the development of a range of cancer risk levels was appropriately supported.
- Due to the controversial nature of the evidence on developmental effects, determine whether DOD’s report considered this evidence in an unbiased manner that was consistent with its use of other toxicological evidence and used sound professional judgment in its evaluation of the evidence.
ANALYSIS OF OVERALL APPROACH
The committee agrees with DOD that there is a need for an updated OEL for TCE. Because DOD is concerned that existing OELs for TCE may not be adequately protective of its workers, DOD could consider using the proposed OEL of 0.9 parts per million (ppm) as a temporary interim value, which is lower than current OELs for TCE. However, the committee could not endorse DOD’s process for deriving the OEL, and lays out several options to address the limitations it found in DOD’s report. The DOD report aims to present its approach as consistent with current best practices in systematic review, but this was not borne out in the hazard identification process. Established systematic review methods have set the bar for objectivity, rigor, and transparency. The committee questions DOD’s decision to deviate from these established best practices. This puts the agency in a position of having to develop, document, and defend a different approach, which is particularly difficult when applied to a chemical with a large and controversial database, such as TCE. In contrast, DOD’s dose-response techniques reflected best practices with respect to validating a PBPK model, evaluating dose-response relationships, and using Bayesian approaches to inform uncertainty adjustments. Below is an overview of the committee’s findings and recommendations.
Problem formulation is the process of defining the scope of a problem, formulating a question about it, and determining the assessment approach by which the question will be answered. Important elements of problem formulation for hazard assessment include specifying the agent, the relevant route of exposure, the health end point(s) to be evaluated, and the types of evidence that will be considered. Evidence could include animal, human, and mechanistic studies. Different assessment approaches can be used for conducting the hazard assessment, including a new narrative or systematic review, relying on an assessment by an authoritative body, or updating a high quality narrative or systematic review. If a systematic review is to be performed, problem formulation should lead to the development of a Population, Exposure, Comparator, and Outcome (PECO) statement that helps guide the review. Several PECO statements may be utilized to address multiple populations, exposures, or outcomes of interest.
DOD sought to derive an inhalation OEL for TCE using systematic review methods. DOD’s PECO statement indicated that animal and human evidence would be used, that all physiologically relevant exposure routes would be considered, and that all potential health end points would be included. DOD recognized
the need for evaluating a large number of studies and decided to use a 2011 assessment of TCE conducted by the U.S. Environmental Protection Agency (EPA) to aid in the assessment process. Given the large scientific database on TCE, the committee viewed the PECO statement as being too broad for performing a systematic review. A more practical approach would be to use the EPA assessment as a scoping review to identify which health outcomes to focus on, to determine what types of data to use, and to formulate specific questions. PECO statements could then be tailored to each question.
Other options DOD might consider are relying on the EPA assessment, updating EPA’s narrative review, or performing a new narrative review. Although these options can be less resource intensive and allow for more flexibility, they still need to meet minimum requirements with respect to how studies are selected, evaluated, presented, and summarized to support hazard identification. Decisions about which option to use will require DOD to consider the time frame and resources that it has available, as well as the overall goal of the assessment.
Hazard identification generally consists of an evaluation of the entire body of available scientific literature on a chemical, and different approaches have been used to perform this step to support development of exposure guidelines. Most occupational and environmental exposure guidelines are supported by the conventional approach of performing a narrative review, but systematic reviews have started to be adopted for performing hazard assessment for the purposes of public health protection. Animal, human, and mechanistic evidence are typically evaluated separately, and then integrated to make a final hazard determination. Systematic reviews have the advantage of providing a high degree of rigor and transparency in evidence synthesis, but they can be time- and labor-intensive, especially for chemicals with a robust database, and may require specialized expertise and data management to meet the expected standards of a systematic review.
DOD’s approach to hazard identification appeared to involve using EPA’s 2011 assessment as a starting point and then performing a systematic review to update it. Systematic review approaches were used to identify newer relevant studies, and then individual studies from the newer and older literature were evaluated using a study applicability tool to evaluate the quality, relevance, and risk of bias (i.e., internal study validity). The committee found several problems with this approach. First, in terms of a systematic review, DOD produced a critically low-quality review, as it lacked a protocol, had inadequate methods to assess risk of bias, and had incomplete descriptions of individual studies. Furthermore, DOD was inconsistent in the degree to which it evaluated different types of evidence. The study applicability tool was developed for animal studies and was only applied to studies of noncancer end points. A similar type of tool was not applied to the evaluation of either cancer or epidemiological studies, which is especially troublesome because human data are the basis of DOD’s estimates of cancer risk. The committee found critical flaws in the design of DOD’s study applicability
tool, because it combined criteria for evaluating individual study quality with criteria for evaluating a body of evidence (a collection of studies) and had some elements that are inappropriate for evaluating individual study quality. Most significantly, the quantitative scores are contrary to standard systematic review practices, as numerical scores falsely imply a relationship between scores and effect or association, along with several other critical limitations. The committee recommends that DOD abandon the use of this study applicability tool in favor of established tools to assess risk of bias of animal and human studies. For example, one option could be the approach developed by the National Toxicology Program’s Office of Health Assessment and Translation.
DOD gave special attention to evaluating the evidence on congenital heart defects, with particular scrutiny of a single study. The committee is aware that the data on this end point has been controversial, but found the emphasis on one study to be contrary to systematic review best practices. Importantly, DOD’s study applicability tool does not seem to have been applied to this study, as it did not receive an applicability score, which indicates that the study was excluded from consideration earlier in the process. The basis for singling out one study for exclusion appears arbitrary, is not transparent, and is inconsistent with the process of how other studies were evaluated.
In the DOD assessment, no separate synthesis and determination of the certainty of evidence was conducted for animal and human studies. It was unclear how or whether mechanistic evidence was identified or assessed. Additionally, it was unclear how evidence integration was conducted to conclude that the hazards that should be considered for TCE were neurological, kidney, liver, immunological, reproductive, and developmental effects.
DOSE-RESPONSE ASSESSMENT AND DERIVATION OF OELs
After hazard identification, a subset of studies on each health outcome is selected for dose-response assessment. The studies must contain adequate quantitative data on dose-response relationships that are either suitable for benchmark dose modeling or provide a lowest-observed-adverse-effect level (LOAEL), no-observed-adverse-effect level (NOAEL), or another appropriate point of departure (POD). It appeared to the committee that all studies that were used for hazard identification were also considered in the dose-response assessment. The strategy for selecting the POD from among the candidate values for each outcome was not explicitly discussed, but it appeared to be selected from among the lower end of the dose range. DOD preferred to rely on inhalation studies, and only considered oral studies if the inhalation data were found to be insufficient for dose-response assessment. PODs were presented graphically to illustrate the range of candidate values, and the associated applicability score was presented for each study. The committee found the presentation of the applicability scores in this step of the process to be inappropriate, because study evaluation techniques should already have been used to exclude studies that were irrelevant, poorly conducted, or had unsuitable data. Separation of study evaluation from the selection of PODs makes
the process more transparent. It is important that criteria be established and studies evaluated before determining PODs to reduce the possibility of having study selection be influenced by the POD values. When appropriate, the committee suggests DOD consider performing dose-response meta-analyses to derive a composite POD for an end point of interest. A composite POD derived from meta-analyses is based on data from multiple studies, which helps to reduce uncertainty associated with use of a POD from a single study and can increase the overall power to detect an association.
DOD refined and verified a PBPK model for TCE, and used it to derive human equivalent concentrations from animal data and to support extrapolation of oral doses to inhalation doses. The committee supports the use of these types of models and views its application as a strength of DOD’s process. Because the PBPK model does not have a component designed for pregnant or lactating animals, the justification and the potential impact on not accounting for anatomic, physiologic, and metabolic changes associated with these processes should be discussed, particularly in considering the need to protect pregnant or lactating workers. DOD should also consider whether it would be appropriate to also model physiological profiles that would be associated with exercise (work) physiology, if TCE exposures occur in workplace scenarios where ventilation rate and cardiac output would be elevated. DOD should also consider using the PBPK model to evaluate the oral data available on all target end points, rather than just the end points with insufficient inhalation data, to take advantage of the robust database on TCE.
Uncertainty factors can be used to address uncertainties associated with calculating toxicity values, such as extrapolation from a LOAEL to a NOAEL and extrapolation of animal data to humans. Ideally, data are used to determine the values of the uncertainty factors, and default values are used in the absence of data. DOD used a Bayesian method to determine the uncertainty factors for deriving an OEL for TCE. The committee endorses DOD’s approach to guide selection of uncertainty factors, and found the comparison with OELs derived using default uncertainty factors valuable for providing context.
DOD estimated the cancer risk at various inhaled concentrations of TCE, and judged that its proposed OEL of 0.9 ppm would be generally protective of unacceptable cancer risk. The committee had concerns with DOD’s estimates because cancer studies were not evaluated in the same manner as noncancer studies. As noted earlier, epidemiological studies, the basis of DOD’s cancer assessment, were not evaluated for study quality, relevance, or risk of bias. Thus, it is unclear if the most appropriate study was ultimately used as the basis for the cancer estimates.