Toxicological and Performance Aspects of Oxygenated Motor Vehicle Fuels (1996)

This chapter contains a synopsis of the approaches to the assessment of human health risks adopted in the HEI and OSTP reports. It also contains an evaluation of the scopes of the two reports, the data and methods used in each, and recommendations for further development of the risk-assessment portion of the federal government's comprehensive evaluation of oxygenated fuels now being undertaken under the aegis of the NSTC's Committee on Environmental and Natural Resources. The final part of this chapter contains recommendations for research efforts that could improve the scientific basis of risk assessment.

The HEI report is broad in scope. It attempts to describe comprehensively

the human health risks associated with the use of gasoline containing MTBE or ethanol; other oxygenates are discussed, but, because health-effects and human-exposure data for these compounds are extremely limited, risk estimates for them could not be developed.

With respect to fuels containing MTBE or ethanol, the HEI report notes the absence of toxicity data on the actual fuels¹, and resorts to data available on selected chemicals, including the two oxygenates. Other chemicals from the fuels (both evaporative and combustion emissions) considered to be potentially important are formaldehyde, acetaldehyde, benzene, and butadiene. These later substances are, in effect, taken as at least partially representative of the potential carcinogenic risks associated with the use of gasoline. The HEI report qualitatively evaluates how the health risks of each of these selected compounds might change when the oxygenates are introduced. These risk comparisons are shown in Table 7.1, taken from the HEI report. Note that the HEI report also considered the potential health benefits associated with the use of oxygenates, represented by their effects on CO exposures.

The HEI report presents a comprehensive review of all data relevant to the risk questions being posed. As is appropriate for human health risk assessment, emphasis is placed on documented health effects, demonstrated in human and/or animal studies. Data gaps are noted, and judgments are made regarding risks about which some conclusions could be reached (i.e., those in Table 7.1).

The risk-assessment methodology adopted in the HEI report is largely qualitative. Although the authors present much of the relevant toxicology/emissions and exposures data in quantitative terms, they make no attempt to estimate the ranges of exposure and risk changes associated with the use of oxygenates in gasoline. Rather, they present the likely directions of change (oxygenated fuels compared with conventional fuels) for the substances listed in Table 7.1 and then provide the following overall assessment.

Based on its review of existing evidence on the exposure to and health effects of oxygenates used in gasoline, the HEI Oxygenates Evaluation Committee drew the following conclusions about the oxygenates themselves.

• I ntroducing oxygenates into gasoline to reduce CO emissions has increased exposure to MTBE for the general public during brief higher-level exposures while refueling and during more sustained but lower-level exposures while driving, and for service station employees during higher-level exposures over entire work shifts. These exposures can occur by both inhalation and skin contact. Workers who handle or transport neat MTBE can experience significantly higher average inhalation exposure levels than people in other situations. Few data on exposure to other oxygenates have been gathered.

• MTBE has been measured in some underground water; its presence in water may result in exposure by ingestion or skin contact should water supplies become contaminated.

• The potential health effects from exposure to gasoline containing MTBE include (1) headaches, nausea, and sensory irritation in some, possibly sensitive, individuals, based on reports after exposure to oxygenates; (2) acute, reversible neurotoxic effects, based on changes in motor activity in rats at high exposure levels; and (3) cancer, based on increases in the frequency of tumors at multiple organ sites in rats and mice at high exposure levels. Although questions persist about how to interpret each of these observed effects, they nevertheless point to a potential human health risk.

• The health effects from exposure to ethanol by ingesting moderate to large quantities have been extensively investigated. Under these conditions, ethanol can increase the risks of certain cancers, adversely affect the developing embryo, produce neurotoxic effects, and cause various other types of damage. However, it is unlikely that such effects would occur at the very low ambient levels to which most people are exposed by inhalation.

• Potential health effects from exposure to other oxygenates are not known and require investigation if their use in fuels is to be widespread.

  In addition to these conclusions about the oxygenates themselves, after qualitatively assessing the health effects of gasoline and motor vehicle emissions with and without oxygenates, the Oxygenates Evaluation Committee has come to the following conclusions about gasoline containing oxygenates.

• The potential health effects of exposure to components of conventional gasoline (without oxygenates) include short-term and cancer effects similar to those that could result from exposure to gasoline containing oxygenates.

• Adding oxygenates to gasoline can reduce the emission of CO and benzene from motor vehicles, and thereby potentially lower certain risks to members of the population. At the same time, using oxygenates increases exposure to aldehydes, which are carcinogenic in animals, and to the oxygenates themselves.

• Adding oxygenates is unlikely to substantially increase the health risks associated with fuel used in motor vehicles; hence, the potential health risks of oxygenates are not sufficient to warrant an immediate reduction in oxygenate use at this time. However, a number of important questions need to be answered if these substances are to continue in widespread use over the long term.

In addition to its conclusions about possible health effects, the Oxygenates Evaluation Committee noted a general lesson to be learned from introducing oxygenates to the general public. Although it is not possible to have complete information about a substance before it is used, the diverse experiences after introducing oxygenated fuels argue strongly that any future new use of a substance should (1) be preceded by a sufficiently comprehensive research

and testing program (including mechanistic and human studies), and (2) be accompanied by rigorous exposure assessment and epidemiologic studies.

The OSTP report is, by design, far more limited in scope than the HEI report. It focuses entirely on MTBE. It makes no attempt to evaluate the risks of MTBE-containing fuels in relationship to nonoxygenated fuels. In effect, the OSTP report is devoted entirely to the possible health risks associated with MTBE exposures resulting from its use in gasoline, in isolation from all other risks.

The OSTP report presents a review and analysis of available data concerning human exposures to MTBE and the compound's health effects as reported in the currently available human and animal database. Data on the compound's principal metabolites, tertiary-butyl alcohol and formaldehyde, are also reviewed.

With respect to acute health effects, The OSTP report contains a review of available data, but concludes that ''the available scientific evidence … was considered insufficient to develop estimates of effect at different exposure levels." The report does, however, present estimates of carcinogenic risks associated with MTBE exposures, using the quantitative methodology ordinarily used by EPA. Thus, upper-bound estimates of cancer risks per unit of lifetime average exposure were developed using the data from three animal studies (those described earlier by the committee) and the

linearized multistage model. These cancer-potency estimates are presented in Table 7.2, taken from the OSTP report.

These potency estimates are combined with estimates of potential population exposures to MTBE (presented in Chapter 2 of the OSTP report and discussed by our committee) to yield estimates of upper-bound, excess lifetime cancer risks from inhalation of MTBE. These estimates are presented in Table 7.3, taken from the OSTP report.

The principal assumptions underlying the results contained in Table 7.3 are presented in the report's text. Note that, in addition to upper-bound estimates, the authors included so-called maximum likelihood estimates (MLEs) of risk; the footnote to the table explains why the MLEs are not considered reliable.

According to the OSTP report, the greatest risks are those based on the assumption that data on lymphomas and leukemias, obtained in an oral-gavage study in rats (Belpoggi et al., 1995), are predictive of human risk. Under this assumption, service-station attendants may incur an extra lifetime risk from MTBE exposure as high as 1 in 2,000, and certain members of the general population may incur risks as high as 1 in 10,000; as is ordinarily the case, the report emphasizes that the actual risks are not likely to be greater than these upper-bound estimates, could be less, and may even be zero. As noted, no analysis of how these risks compare with risks from unoxygenated gasoline is presented.

Although the OSTP report presents all the available human and animal data concerning the toxicity of MTBE, only acute health effects and possible carcinogenic risks are presented in the final chapter, on risk characterization. The authors present EPA's inhalation reference concentration (RfC) for MTBE, derived with standard regulatory methods and defined by EPA as "an estimate (with uncertainty spanning about an order of magnitude) of a continuous inhalation exposure level for the human population (including sensitive subpopulations) that is likely to be without

appreciable risk for deleterious noncancer effects during a lifetime." EPA's RfC of 3 mg/m³ (0.83 ppm) was based primarily on findings of increased absolute and relative liver weights and several other effects in rats exposed chronically to MTBE by inhalation.

The OSTP report appears to endorse this RfC as a guide to risk assessments for the documented noncancer health effects of MTBE and concludes that the "reasonable worst-case annual average daily exposure estimate (0.019 ppm)" is well below the protective value. The OSTP report does not deal with the ingestion associated with MTBE, as might exist from consumption of contaminated water.

In addition to its characterization of MTBE's cancer risks, the OSTP report concludes that "it is not known whether the cancer risk of oxygenated gasoline containing MTBE is substantially different from the cancer risk of conventional gasolines." The report also notes that "data were generally inadequate to evaluate the health risks of oxygenates other than MTBE."

It is obvious that the two reports differ substantially regarding scope and methods for risk assessment. The OSTP report, in part because of its stated objective, is very much narrower in scope than the HEI report; but it also states that the type of comparative evaluation undertaken by HEI is not possible with currently available data. HEI's report attempts a comparative evaluation of risk but stops short of any attempt to deal quantitatively with risk.

Although neither report points to a substantial health risk associated with MTBE exposures, it is difficult to compare the two, because of their substantially different scopes, methods, and approaches to scientific uncertainty. It must be noted, however, that the OSTP report describes (albeit with appropriate qualifications)

substantial cancer risks associated with certain MTBE exposure scenarios (Table 7.3) but includes no serious attempt to explain their possible public-health importance. Such results, uninterpreted, could be considered at odds with those presented by HEI. (As previously discussed, this committee recommends that risk estimates based on the leukemia-lymphoma findings in the animal study be rejected until further investigation of the results is undertaken.)

Further comparisons of the two reports with respect to their approaches to the risk questions associated with fuel oxygenation would not seem useful here. Rather, we chose to emphasize how the risk-assessment portion of the government's comprehensive evaluation of oxygenated fuels might derive the greatest benefits by recommending ways in which their analyses might be improved.

Both reports appropriately emphasize the substantial uncertainties associated with the problem of developing a comprehensive, comparative assessment of the health risks associated with oxygenated fuels. Indeed, it is clear that the database for certain oxygenates is so meager that no useful analysis can be undertaken. Moreover, the full range of health effects and risks associated with the mixtures of compounds to which populations are exposed through the use of nonoxygenated fuels are not established, and there are, as yet, no data on the health effects of the corresponding mixtures that arise when oxygenated fuels are used (we refer to the mixtures that arise from evaporation, those from combustion, and those from migration

into water supplies, which differ from each other). It might be argued that until data are available on these mixtures, no attempt can or should be made to evaluate human health risks.

Such an argument fails on at least two grounds. First, although comprehensive data on the above mentioned mixtures are not available, a substantial amount of data do exist for selected components of those mixtures that are known to present potential health risks. Second, the argument misunderstands the purpose of risk assessment, as elucidated by the National Research Council (NRC 1983, 1994). Risk assessments are undertaken to describe the current state of knowledge regarding the subject being investigated and should include a through description of the uncertainties in that knowledge. Even a risk assessment performed with uncertain data is justified as long as the uncertainties are thoroughly characterized and stated. The risk manager, reviewing the assessor's work, may decided not to use it for decision-making if the uncertainties are judged to be too large, but the assessment itself should avoid such a policy judgment (NRC 1994).

The assessment contained in the HEI report comes close to the type of evaluation the committee thinks is most useful for decision-making regarding the risks associated with oxygenated fuels. It includes an assessment of the risks associated with such fuels, to the extent that those risks are represented by the specific chemicals selected for evaluation and relative to the risks associated with conventional fuels. It includes consideration of the benefits associated with the use of oxygenated fuels. Potential risks associated with water contamination are also described. All these risks are placed in perspective, and the major uncertainties associated with the assessment are well described. We suggest that the HEI report be used as the framework and database (with appropriate additions and reinterpretations, as described above by the committee and in previous chapters) for the comprehensive government risk assessment. In the subsections that follow, we describe the aspects in this report that we think need to be refined for this purpose.

The OSTP report seems far too limited in scope to form the basis for a comprehensive, comparative risk assessment. Although some features of the report are useful (see below), its failure to deal with any of the potential risk issues, on the argument that data are too limited, seems to this committee unsupportable. (The scope of the report, while stated to be narrowly focused on MTBE alone, apparently did not restrain the authors from reaching conclusions regarding the inability to deal with risk comparisons.) The OSTP report fails to present well-documented reasons to support its view regarding the inadequacy of the database for risk assessment (except for the obvious cases of oxygenates on which virtually no data exist).

The other advantage that derives from undertaking a risk assessment is that data and knowledge gaps that are most useful for improving the risk estimates are clearly revealed. The HEI report, because it attempts to be comprehensive, is quite effective in revealing critical research needs.

Although we recommend that the HEI report be taken as the framework and starting point for the comprehensive risk assessment, we propose that greater effort be made to provide some indication of the magnitudes of the health risks that are said to increase and decrease (relative to conventional fuels) with the use of oxygenates (Table 7.1). We recognize that, regarding the toxicities and dose-response characteristics of the compounds included in the risk analysis, the HEI report is correct with respect to the uncertainties associated with the selection of specific values for human risk assessment. Nevertheless, there are substantial quantitative toxicity data on each of the chemicals, and there is no reason why the federal government cannot adopt reference concentrations, reference doses, and cancer potency factors that have been prepared or developed for all the chemicals using federal guidelines for risk

assessment. It is recognized that such values are based in part on science-policy choices (NRC 1994), but, as long as their limitations are appropriately noted, there is no reason why they should not be used for risk assessment. Indeed, federal government agencies regularly issue assessments involving use of such toxicity values. Thus, while the scientists preparing the HEI report may have felt reluctance in describing such measures of toxic risk, we see no reason why the federal government cannot use its standard procedures for such toxicity assessments.

With respect to the exposures described by HEI and their changes following introduction of oxygenates, our comments are similar. It is not necessary to have perfect knowledge regarding exposure and exposure changes; in each case it should be possible to offer some view of at least the likely ranges of exposure to the selected chemicals. The OSTP report, for example, provides some information on the ranges of short-term and long-term MTBE exposures that may be experienced by certain segments of the population. Such an analysis was provided by HEI but was not used in any explicit way in the risk assessment.

Without some attempt to provide an indication of the magnitudes of the risks associated with the potential constituents of oxygenated and nonoxygenated fuels, it is difficult to accept the final conclusions of the HEI report. This is not to say that the assessment should be dominated by quantitative estimates, but only that some attempt to quantify risk ranges should be made. If the ranges of plausible risk increases and decreases overlap substantially, then this would be a reasonable basis for the conclusions reached in the HEI report.

We thus urge the adoption of the scope and framework for risk assessment contained in the HEI report, with greater attention to presentation of quantitative results. Uncertainties in the data should also be carefully described.

With respect to some of the specific approaches to risk characterization offered in the two reports, we have the following observations and recommendations.

With respect to noncancer health effects resulting from MTBE inhalation, the EPA's reference concentration (RfC) of 3 mg/m³ (0.8 ppm) seems appropriate. It is based on the application of standard uncertainty factors and interspecies adjustments to a well-defined NOAEL from a chronic-inhalation study.

Until the issues concerning the possible effects in humans following short-term exposures to MTBE-containing gasolines are clarified, it should be made clear that EPA's RfC may not be appropriate to assess risks of such effects in humans.

No discussion is presented in either report of the need for and value of a reference dose of MTBE—a daily oral intake that is expected to protect against noncancer effects of the compound when it is ingested. Because of the potential for ingestion of MTBE through groundwater that has become contaminated, a close examination of this issue should be undertaken. Several years ago, EPA issued a chronic health advisory for MTBE in drinking water of 20 µg/L. This value was based on data available in 1992, and several default assumptions were used because of large data deficiencies and uncertainties. The "acceptable" intake (dose) implied by the 20 µg/L advisory is substantially less than that implied by the current RfC. An attempt should be made to review currently available toxicity data to establish a reference dose for MTBE.

Because of the substantial doubts regarding the reliability of the leukemia and lymphoma findings reported by Belpoggi et al. (1995), these should not be used for human risk assessment at this time. The estimates of potential human risk reported in Table 7.3, based on these findings, should not be used until the data upon which they are based are verified. When the various questions we have

raised have been addressed, a decision can be made regarding the use of these results.

The data for MTBE-induced kidney tumors in male rats should not be used for human risk assessment until the recently reported (CIIT) data on the mechanism of action are reviewed and evaluated. If the new data support the view that _2u-globulin nephropathy is involved in the response, as this committee now believes, then this end point should be discounted for human risk assessment.

The data on MTBE-induced tumors in female mice can be used for risk assessment, but the combining of potency estimates from males and females (OSTP report, Appendix A) is not appropriate. The potency estimate only for females should be used. Although not recommended as the sole basis for assessing the risks, the risk assessment should also consider the NOAEL in the mouse study at 3,000 ppm (human exposure equivalent, 540 ppm) and the margin of exposure separating actual human exposures from this value.

The maximum-likelihood estimates of risk presented in Table 7.3 serve no useful purpose and should be discarded. For reasons stated in the footnote to the table, these estimates are not considered reliable. The committee sees no value in presenting such estimates.

• The committee has made a number of recommendations for refinements and improvements in the assessment of potential human health risks associated with prolonged exposures to gasoline containing MTBE and for assessment of the comparative risks associated with oxygenated and nonoxygenated fuels. Until these recommendations are acted upon, no definitive statement can be made regarding these health-risk issues. Based on the available analyses, however, it does not appear that MTBE exposures resulting from the use of oxygenated fuels are likely to pose a substantial human health risk.

• It appears that MTBE-containing fuels do not pose health risks substantially different from those associated with nonoxygenated fuels, but this conclusion is less well established and should become the centerpiece for the government's comprehensive assessment.