Ninth Interim Report of the Subcommittee on Acute Exposure Guideline Levels (2003)

Ninth Interim Report of the Subcommittee on Acute Exposure Guideline Levels

In 1991, the U.S. Environmental Protection Agency (EPA) and the Agency for Toxic Substances and Disease Registry (ATSDR) asked the National Research Council (NRC) to provide technical guidance for establishing community emergency exposure levels (CEELs) for extremely hazardous substances (EHSs) pursuant to the Superfund Amendments and Reauthorization Act of 1986. In response to that request, a subcommittee of the NRC Committee on Toxicology (COT) prepared a report titled Guidelines for Developing Community Emergency Exposure Levels for Hazardous Substances (NRC 1993). That report provides step-by-step guidance for the derivation of CEELs for EHSs.

In 1995, EPA, several other federal and state agencies, and several private organizations convened an advisory committee—the National Advisory Committee on Acute Exposure Guideline Levels (AEGLs) for Hazardous Substances (referred to as the NAC)—to develop, review, and approve AEGLs (similar to CEELs) for up to 400 EHSs. AEGLs developed by the NAC have a broad array of potential applications for federal, state, and local governments and for the private sector. AEGLs are needed for prevention and emergency response planning for potential releases of EHSs either unintentionally from accidents or as a result of terrorist activities.

The NRC convened the Subcommittee on Acute Exposure Guideline Levels to review the AEGL documents approved by the NAC. The subcommittee members were selected for their expertise in toxicology, pharmacology, medicine, industrial hygiene, biostatistics, risk assessment, and risk communication.

The charge to the subcommittee is to (1) review AEGLs developed by the NAC for scientific validity, completeness, and conformance to the NRC (1993) guidelines report, (2) identify priorities for research to fill data gaps, and (3) identify guidance issues that may require modification or further development based on the toxicological database for the chemicals reviewed.

This interim report presents the subcommittee’s comments concerning the draft AEGL documents for 13 chemicals: Methanol, acrylic acid, crotonaldehyde, hydrogen sulfide, phenol, cyclohexylamine, ethylenediamine, HFE-7100, carbon monoxide, ethylenimine, propylenimine, allylamine, and chlorine dioxide.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on methanol. The document was presented by Peter Griem of F.O.B.I.G., Germany. The subcommittee recommends that the following revisions be made to the document. A revised draft should be reviewed by the subcommittee at its next meeting.

Page x, lines 2–13. The NAC should consider using the following three paragraphs, rather than the first two existing paragraphs, as an introduction to the Executive Summary:

Methanol (also known as wood alcohol) is a clear, colorless, volatile, flammable liquid with a sweet odor. It is used in paint removers, windshield washer fluid, automotive fuel, and antifreeze; as an industrial solvent; and as a raw material in the production of many commercially important organic compounds. Small amounts of methanol are produced over the course of normal body metabolism and are found in the exhaled air.

Methanol is rapidly absorbed after ingestion or inhalation. Percutaneous absorption is also considerable. Acute methanol toxicity varies greatly between species, primarily as a result of differential metabolism. At very high inhaled concentrations rodents exhibit much higher blood methanol concentrations than do primates. Primates accumulate greater amounts of the important toxic metabolite—formic acid (found in equilibrium in plasma with its anion, formate). Primates are more susceptible than rodents because of the greater accumulation of formates in primates. Clinical experience with those who ingested methanol (often under the mistaken assumption that they were consuming ethanol) demonstrates marked variations in individual susceptibility and delayed onset of severe, overt toxicity. The initial phase of inebriation is similar to that seen after ethanol but is usually mild and transient and is generally followed by an uneventful initial recovery. The most important clinical consequences develop between 6 and 30 hours after the initial exposure.

Wide individual variations in response are most likely due to individual rates of formate production from methanol in the liver. People with pre-existing liver disease (e.g., cirrhosis) often appear resistant to methanol poisoning because of their relatively inefficient conversion of methanol to formic acid. Accumulation of formate in primates leads to depletion of the normal bicarbonate buffering capacity of the body, delayed-onset metabolic acidosis and death with acute cerebral edema, CNS depression, and coma. The severity of the poisoning and the patient’s prognosis are related directly to the extent of formate and lactate formation, which account largely for this metabolic acidosis. Among victims who survive the initial phase, vision can become severely impaired and permanent bilateral blindness can follow formate-induced retinal edema, demyelination of the temporal retina, hemorrhagic necrosis in the basal ganglia, and nerve head pallor. Pancreatitis has been associated with acute abdominal pain. Occupational methanol exposures in confined spaces or in workrooms with inadequate ventilation have been associated with recurrent giddiness (mild inebration), nausea, insomnia, blurred or dim vision, and conjunctivitis. The delayed onset of symptoms, the potent ocular degeneration, and the metabolic acidosis seen in primates poisoned with methanol are not observed in rodents. AEGL derivations should not be based on rodent data because of these pronounced interspecies differences in metabolism and toxicity.

Page x, lines 17–18. Undue attention is paid to the occupational exposure limit of 200 ppm, a level identified as an 8-hour time-weighted average (TWA) that was also selected as the TLV, REL, and PEL. The value was based on a recommendation by Warren Cook (Ind. Med. 14:936–946, 1945) that he made after reviewing a study (Sayers, R.R., et al. U.S. Bureau of Mines Report of Investigations No. 3617, USBM, Washington, DC, 1942) that found no adverse effects in dogs exposed 7 days weekly for 379 days to methanol vapor at 450–500 ppm. Thus, there is nothing unique about 200 ppm value except that it has been accepted and

in effect for more than 50 years. There appear to have been no reports of adverse health effects under these conditions.

Page x, line 27. Is there provision in the Standing Operating Procedure (SOP) document for use of an intraspecies uncertainty factor of 3, because a no-effect level (in the study of Batterman et al. 1998) was less than defined for AEGL-1? Protection of folate-deficient persons would be a better justification for an intraspecies uncertainty factor of 3.

Page 1, line 3. Increased emphasis should be placed on the primary hazard (i.e., fire) associated with methanol spills and releases. It would be helpful for the authors to search the flammability literature and, if possible, to include a brief description of accounts of wood-alcohol fires.

Page 2, lines 6–7. Conclusions are improperly referenced here. Frederick et al. (1984) did not report fatalities in a group of teachers/aids exposed to methanol when using duplicators. It was concluded by these investigators that the deaths of three former aids were not related to methanol exposure. The fatalities described by Becker (1983) were caused by ingestion, not inhalation, of methanol. IUCLID (1996) is a secondary literature source that cites only very old reports of fatal vapor exposures. Many of the occupational exposure cases likely involved percutaneous exposure as well as inhalation exposure. It is not clear whether a lethal or even a sublethal blood level can be achieved solely from inhalation. See subsequent comment on this point.

Page 3, lines 2–3. Although it is stated here that the fatal case of methanol poisoning involving concomitant ethanol exposure is not shown in Table 2, that case appears to be the third entry from the top of the table.

Pages 4–5, Table 2. It would be helpful to note in the subheading on blood methanol concentration that the time is “postexposure.”

Page 6, line 5. A more recent text reference that might be cited here is Bruckner, J.V., and D.A. Warren. 2001. Ch. 24, Toxic Effects of Solvents and Vapors. In: Casarette and Doull’s Toxicology: The Basic Science of Poisons, pp. 894–895, 6th Ed. New York: McGraw Hill.

Page 6, lines 6–8. This last sentence in the paragraph is out of place. For the most part, Table 3 pertains to effects seen in nonlethal experimental and occupational exposures (i.e., effects described in Sections 2.2.1 and 2.2.2).

Page 7, line 6. D’Alessandro et al. (1994) did not find a significant increase in serum or urinary formate levels.

Page 7, lines 21–23. It is not necessary to state that the subjects were right-handed.

Page 8, lines 31–33. Omit these lines and the citation of Ruth (1986). Such a vague value and reference is not necessary given the wealth of controlled odor studies.

Page 8, lines 39–41. The last sentence in this paragraph should be omitted. This pharmacokinetic study is not appropriate for establishing a threshold for causation of neurological effects.

Page 9, line 19. It would be best to delete the phrase “and other symptoms unrelated to methanol toxicity,” because it does not appear to be relevant to methanol toxicity.

Page 10, lines 35–36. This sentence is poorly written. It should be pointed out that case reports of mortalities exist only in the very old literature. Exposure levels and conditions are not known.

Page 12, lines 17–18. The phrase “no significant slight neurotoxic effects” is confusing.

Page 14, lines 17–18. The meaning of the following phrase is unclear: “number of animals exposed 1 respectively 4 hours unstated.”

Page 15, line 13. Part of this sentence is missing. Do the following words suffice: “Deaths occurred in monkeys gavaged with”?

Page 15, line 38. Scott et al. (1979) is listed as Scott et al. (1981) in the References section.

Page 15, line 39. Should LC_1o be LC_Lo or LC₁₀?

Page 18, lines 23–25. It is noted that a slight increase in glial cells in the optic nerve and a slight degeneration of peripheral nerves was “suspected” in animals exposed at 1,000 ppm after 6 months of recovery. What is meant by “suspected”?

Page 18–19, line 17. Although the primate studies of NEDO (1987) and Andrews et al. (1987) involve subacute exposure regimens, they should be used in support of the AEGL determinations.

Page 20, lines 2–7. The red discharge observed by Andrews et al. (1987) was very likely chromodacryorrhea, a red pigment secreted by stressed rats. As noted in lines 6 and 7, no irritation or histological changes were reported by other investigators in rats exposed to methanol at higher vapor concentrations.

Page 22, lines 31–32. What vapor concentration caused vaginal bleeding in two of six monkeys?

Page 23, line 24 and page 26, line 26. It is probably not necessary to provide such detailed information on the high, repeated-dose reproductive toxicity studies in mice and rats. It would be preferable to summarize these, because their relevance to acute exposures and effects in primates is very questionable.

Page 26, lines 30–35. The highest methanol concentrations that gave negative results should be stated.

Page 27, lines 28–29. What were the durations of the lethal 5,000- and 10,000-ppm exposures? What were the numbers of exposures in the experiments in which no monkeys died?

Page 28, lines 1–3. Is vaginal bleeding a sign of developmental toxicity? It is not stated what vapor concentration(s) produced vaginal bleeding or possibly resulted in a wasting syndrome. It should be pointed out here that the etiology of the wasting syndrome is unknown.

Page 28, line 8. Why is it stated here that there is no evidence of genotoxic effects when positive mutational findings are described in lines 37, 38, 41, and 42 of page 26?

Page 28, line 19. The range of values here is for percentage uptake/absorption, not absorption rate.

Page 28, line 20. It would be helpful to cite a more recent, more comprehensive study of the pharmacokinetics of inhaled methanol. Dorman et al. (1994) published the findings of such an investigation in rhesus monkeys. More detail on the characteristics of methanol absorption and disposition should be given in the text.

Page 28, line 25. Pollack and Brower (2000) conducted a study of disposition of methanol in pregnant vs nonpregnant rats. Uptake and elimination were unaffected by pregnancy. Fetal methanol levels approximated maternal levels.

Page 31, line 9. The PBPK models of Fisher et al. (2000) and Bouchard et al. (2001) for methanol inhalation by monkeys and/or humans should be mentioned.

Page 34, lines 9–10. Methanol, like other alcohols, causes CNS depression if present in sufficiently large amounts. A person who inhales ethanol vapor cannot attain blood levels sufficient to impair his/her ability to drive a motor vehicle. Methanol is less lipophilic and is therefore a less potent CNS depressant than ethanol. It follows that a person inhaling a high concentration of methanol will experience no more than mild inebriation.

Page 34, lines 10–12. It is often not possible to detect the initial, more subtle CNS-depressant effects of methanol in animals that are experienced and reported in humans. Thus, CNS effects observed in animals are not “similar,” but are usually more marked.

Page 35, Figure 1. It would be preferable to place Figure 1 adjacent to the section on Metabolism (4.1.2).

Page 36, lines 1 and 14–16. It would be worthwhile to compose a new paragraph addressing the apparent dual modes of action of methanol and low folate levels in reproductive toxicity. As already noted in lines 15 and 16, methanol seems to act in parallel with reduced folate levels to produce teratogenic effects in rodents. Andrews et al. (1998) concluded from their in vitro studies of mouse embryos that formate and methanol did not act by the same mechanism. Sakanashi et al. (1996) and Fu et al. (1996) observed that low dietary folate (alone) did not cause exencephaly in fetal mice, but did result in an increased incidence of cleft palate. Folate

levels were lower in fetal tissues than in maternal tissues. It would be helpful to mention here the role of folic acid in fetal development and the consequences of folate deficiency in development. It should also be stated in the new paragraph that methanol does not appear to alter folate levels (Sakanashi et al. 1996), nor does folate deficiency alter blood methanol concentrations (Lee et al. 1994b). Methanol alone causes exencephaly in fetal mice. Ethanol’s ability to cause fetal alcohol syndrome (FAS) is well known. Formate plays no apparent role in FAS. In summary, methanol and folate deficiency (each alone) can cause dissimilar fetal malformations in mice. The totality of their adverse effects (in combination) is greater than the sum of each’s actions alone in vivo. This potentiation does not appear to result from a metabolic interaction, but might involve pharmacodynamic interactions.

Page 36, line 18. The word “normal” should be included in this subheading.

Page 36, lines 41–43 and page 37, lines 1–5. The work ascribed to Medinsky et al. (1997) was actually published in the peer-reviewed literature in 1994 by Dorman et al. They, like Horton et al. (1992), saw no increase above background in blood formate levels in folate-deficient monkeys inhaling methanol at up to 900 ppm for 2 hours. Blood methanol concentrations in the 900-ppm folate-deficient group were higher than in the 900-ppm folate-sufficient animals, but the increase was not statistically significant.

Page 37, lines 32–33. Do the mean values of 5.1, 6.3, and 9.2 nmol/L represent maternal blood folate concentrations? Did Sakanashi et al. (1996) assess fetal folate levels?

Page 38, lines 20–24. Compare the causes of death by methanol and ethanol. The mode of action of methanol accounts for its greater lethal potency.

Page 39, line 8. It would be useful to add a short paragraph that summarizes the teratogenic effects and potential mechanisms of action of ethanol in rodents and humans. FAS has been a subject of study by many investigators. Bruckner and Warren (2001) provided an overview of the subject on page 893 of the 6th edition of Casarette and Doull’s Toxicology.

Page 39, lines 15–17. It is stated here that rodents are more prone to methanol-induced developmental toxicity. Is there experimental evidence of that?

Page 39, lines 42–43. This introductory sentence is confusing. Does it mean that a normal diet contains about half of the daily intake of folic acid needed by pregnant women?

Page 41, line 17. The 20-day study results of NEDO (1987) should be summarized here, because they should be more relevant to an acute exposure than the 7-month NEDO study. Neurotoxic effects were seen within 5 days of inhalation at 5,000 ppm. Were any adverse effects seen with shorter exposure durations?

Page 41, lines 20–22. It is stated here that Andrews et al. (1987) did not report any histopathological changes, yet changes are described in lines 8–11 on page 18.

Page 41, lines 29–42. The investigation of Batterman et al. (1998) should not be adopted as the key study for AEGL-1 calculations unless factual documentation was obtained to support the claim that inhalation of methanol at 800 ppm for 8 hours produced no adverse effects. Such documentation might include the protocol indicating that subjects were to report symptoms, the informed consent forms encouraging reporting of adverse symptoms, etc. Recollection by the second author does not suffice. The study was not single- or double-blinded, as the researchers and volunteers were aware of the vapor concentrations being utilized. Secondhand information from a study designed for other purposes cannot be considered reliable. It can be used as supportive, but not primary, data on which to base AEGL-1 values.

Page 42, lines 17–33. This paragraph contains excessive detail about adaptation of the ten Berge et al. (1986) approach to time scaling for AEGLs. In instances where this equation is utilized, the reader might be referred to an appendix or the SOP.

Page 44, lines 25–28. The inhaled concentration that produced vaginal bleeding should be stated. It should be noted that the wasting syndrome was of uncertain etiology. Sensorimotor development delay was evidenced by two of the behavioral tests, but not by the others.

Page 45, lines 13–15. This statement about the need for more research should be included in Section 8.3 on future research needs.

Page 45, line 19 and page 46, line 3: It appears (from information presented in the Executive Summary and that presented from in lines 4–26 on page 52) that the abstract of Rogers et al. (1995) and the personal communication of Rogers (1999) provided the basis for derivation of the AEGL-2s. It is very difficult, however, to tell from the discussion (that runs from line 19 of page 45 to line 3 of page 46) where the 7-hour 2,000-ppm NOAEL value and its associated blood level of 487 mg/L come from. For the sake of clarity, the NAC should limit the information to the relevant data from the key studies.

Pages 45–46. The mouse developmental toxicity studies of Rogers et al. should not be used to derive AEGL-2s. The toxicokinetics and metabolism of methanol are too different in mice and humans to extrapolate findings from one species to the other. Much higher blood methanol levels are reached in mice during inhalation exposures, due largely to their more rapid respiratory and cardiac output rates. Metabolic clearance rates are species-dependent—mice rely on the catalase pathway, while humans rely primarily on the zero-order alcohol dehydrogenase pathway. Species differences in folate levels/metabolism may also be

important, because folic acid deficiency accentuates the reproductive toxicity of methanol in rodents. Mice were utilized by Rogers et al. because of their low cost, ease of handling and husbandry, short gestation, and large numbers of offspring. Quite different conclusions and NOAELs would have been reached if the document’s authors had relied on rat developmental toxicity studies (e.g., Nelson et al. Fund. Appl Toxicol 5:727–736, 1985). Rats, however, have the same limitations as mice. In order to obtain scientifically rigorous AEGL-2s based on developmental toxicity, it would be necessary to use data from investigations with nonhuman primates. However, as noted in lines 14–16 on page 45, the monkey studies of Burbacher et al. (1999a,b) do not establish clear dose-response relationships.

Page 46, lines 7–14. Why is the uncertainty factor applied to internal concentrations? Even though a part of the UF can be used to account for pharmacodynamic differences, application of the total factor of 10 on the internal concentration is inappropriate.

Page 46, lines 16–33. The NAC is to be commended for its use of a pharmacokinetic model for species and time extrapolations. Valid models can be very powerful and useful tools in risk assessment. There are, however, some problems with the modeling that was performed.

Page 47, line 39. It would be helpful if Figure 2 is placed nearer to Section 7.1. Also, it would be useful if data points corresponding to some of the key studies are identified in the figure.

Page 48, lines 33–43 and page 49. The NAC has used a reasonable approach to calculate AEGL-3s. The only reports of fatalities from methanol inhalation are very old. In many of these cases there appear to have been concurrent inhalation and dermal exposure. Thus, inhalation alone was probably not responsible for these deaths. As noted above, Andrews et al. (1987) saw no adverse effects in monkeys that inhaled methanol at 5,000 ppm 6 hours per day, 5 days per week, for 4 weeks. NEDO (1987) observed lethargy in monkeys inhaling 10,000 ppm for 21 hours and mortality during the third such exposure. Formate levels likely accumulated under these exposure conditions.

Page 48, lines 41–43. The NAC should clarify that it used a modifying factor of 2 to conservatively estimate a peak blood NOAEL (555 mg/L) from a calculated peak blood level (1,109 mg/L) in a fatal case of methanol ingestion.

Page 54, line 15. The data points that are included in Figure 3 should be limited to primate/human data.

Page 55, Figure 3. This figure is difficult to interpret. The title should be changed to “Consistency of Data With Derived AEGL Values.” The figure’s legend should be inverted so that the more severe effects appear at the top, as do the data points. The diamonds, representing censored data, should be omitted for the sake of clarity. The AEGL values might be easier to locate if they were designated with asterisks.

Page 56. It would be useful to address differences between recommended AEGLs and current occupational exposure limits. As previously pointed out, undue attention is paid to the TLV of 200 ppm.

Page 58, line 39. Is there a basis for stating that “details are...intentionally reported wrongly by the victims”?

Page 75, lines 3–4. Why are so many case reports included under the heading of “key study”? The report used as the key study was that from which the lowest fatal peak blood methanol concentration was determined.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on acrylic acid. The document was presented by Peter Griem of Germany. The subcommittee recommends that the following revisions be made to the document. A revised draft should be reviewed by the subcommittee at its next meeting.

The concentration applied as vapor versus aerosol has a significant impact, especially in the case of a highly soluble compound such as acrylic acid. In the vapor state, one would expect local effects in the olfactory epithelium and general upper respiratory tract. On the basis of the data cited, this in fact is what appears to occur after exposure to the vapor. However, the aerosol could be delivered to the deep lung and, therefore, could be more toxic than the vapor, further explaining the apparent discrepancies discussed in Section 7.3. Compare the 1-hour LC₅₀ (rats) for the aerosol at 3,850 ppm (page 32, line 3) against the 3,996-ppm vapor concentration for 4 hours with no deaths after 14 days (rats)(page 8, lines 1–2). (Note: Aerosol concentration is always given in mg/m³, not ppm, which is on a volume/volume basis)

What is the likely form of an acute airborne acrylic acid exposure to the general public? It would seem that even if an aerosol was formed, it would quickly convert to vapor due to the relatively high vapor pressure of acrylic acid. If that is the case, an AEGL based on the vapor is more relevant.

The saturated vapor pressure of acrylic acid is 3.1 mm Hg at 68 °F per the MSDS. That is approximately 4,100 ppm (12,000 mg/m³). In one section on page 8, line 16, it is stated as 5,000 ppm. But on page 32, line 21, it is stated that 2,142 ppm is the highest concentration that could be attained. On page 8, line 2, a concentration of 3,396 ppm is cited. Above the saturated vapor pressure for the test chamber temperature and pressure, the concentration must exist as an aerosol. It is difficult to compare exposure data if the form of the concentration is unknown and significant differences apparently exist on the basis of the cited data.

Page 4, line 33. Why would whole-body exposure have a greater effect than nose-only exposure? The reader might assume it is because of eye irritation, but there is no explanation given.

Table 2. Regarding the personnel, the reader might assume that these are different groups of people working in different areas. Is there any information on the frequency of exposures for these different groups? Of special interest is the notation on the last exposure group. It is well known by experience with ammonia and formaldehyde that personnel become acclimatized, accustomed, or oblivious to, or acquire other adaptations to, common irritants and appear unaffected at concentrations the “nonacclimated” person finds objectionable or intolerable.

Page 22, lines 31–32. Is maternal toxicity different than other types of toxicity? The effects noted seem to be common to all species regardless of gender or pregnancy status. If body-weight gain is the issue, emphasize that, and state the nature of the eye and nose irritation that also occurred.

Page 29, line 16. “Between species” should be “within species,” because the intraspecies UF is to account for variability within the human population.

Page 31, line 8. The text states that the lethal effects after inhalation are also caused by local destruction of respiratory tract tissues. Is this statement based on red foci in the lung of rats at gross necropsy (page 5, line 3)? Where are the experimental data to support this statement?

Page 31, line 19. “Between species” should be “within species,” because the intraspecies UF is to account for variability within the human population.

Section 4.2. Is this the common mechanism of a simple irritant? Restated, is this a common irritant response or is there another mechanism occurring?

Section 8.3. Perhaps some additional research could be proposed to decrease some of the uncertainties in toxicokinetics and confirm or refute differences in toxicodynamics mentioned in the UF sections.

It does not seem appropriate to use a personal communication (Renshaw 1988) as a key reference. One cannot tell the nature or extent of the communication from the citation (e.g., verbal, internal memo, or report with data). The authors acknowledge this limitation on page 37, line 8, but

there are apparently some data referenced (Appendix C, page 51). The reviewers need to evaluate this documentation.

The uncertainty factor (UF) section on page 29 (lines 10–17) needs some clarification and rewording. The rationale needs to be better articulated. Apparently, a total factor of 1 is applied for the intraspecies UF rather than the default 10-fold factor (generally broken down into two factors of 3 to account for toxicodynamic and toxicokinetic differences). However, because the toxicity is local, occurs at the portal of entry (i.e., irritation), and is not systemic, the authors discount using the toxicokinetic component of the default UF. In addition, the authors dispense with the toxicodynamic component of the UF because the effects were likely seen in a sensitive subpopulation of workers, accounting for potential variability in tissue response amongst humans. The argument that the workers were a sensitive subpopulation is also made on page 52 in the appendix. A stronger argument needs to be made for there to be a credible basis for discounting a UF for sensitive subpopulations. What about other sensitive subpopulations (e.g., the elderly and children)?

The subcommittee is not convinced that histological changes in the olfactory epithelium is the most appropriate end point for an AEGL-2, the level above which irreversible or other serious, long-lasting adverse health effects, or an impaired ability to escape, could be experienced. The AEGL seems conservative given that these are relatively subtle changes in rats (monkeys had similar olfactory but not respiratory tissue changes as rats), which receive a 3-fold higher target tissue dose than do humans (Frederick et al. 1998). Also, can’t this target tissue repair/regenerate?

Again, the UF section needs some additional explanation; it needs to be explicit and transparent as to what the toxicokinetic and toxicodynamic components for each UF are and why they are or are not included. The UF of 1 for interspecies is appropriate, although conservative, because the tissue dose producing adverse effects in rats is 3-fold higher than that for humans. Toxicodynamics across species—at least rats and monkeys—appear the same. The local nature of the effect at the portal of entry and the work of Frederick and colleagues argues against the need for a toxicokinetic component in extrapolating from rats to humans. It is not clear why the 3-fold UF for the toxicokinetic component of the intraspecies UF is retained. Wouldn’t the same argument of local vs systemic effects for the intraspecies UF made for the AEGL-1 apply here as well? Or are the authors thinking of the elderly and children? Or of various activity patterns that may affect dose to the olfactory tissue? The authors are apparently assuming that there is no uncertainty or variability in target-tissue response in the human population (UF for toxicodynamics=1).

Clarify the rationale for each UF of 3 leading to a final UF of 10. The UF for intraspecies needs to be better explained.

Why was an MLE for LC01 used (page 32, line 23; Appendix B, page 49) when SOPs (page 39) state that the NAC/AEGL committee will generally use BMCL05 as the preferred approach? Although the BMCL05 and MLE01 do not differ much at shorter time intervals, the BMC05 values are lower at the longer time intervals than are the MLE01. That needs to be explained and a stronger argument made.

There are a number of inconsistencies among data presented in tables and in the text (e.g., Table 5, Majka et al 1974) and whether exposure was by vapor or aerosol. The NAC needs to carefully edit its report.

There are many spelling and grammar errors.

Page 2, line 6. Is “inhalative” a word, and if so, is it used in the proper context? Perhaps the word “inhalation” is correct.

Page 29, line 21. Is olfaction a word and is it properly used here?

At its January 27–29, 2003, meeting, the subcommittee reviewed the revised AEGL document on crotonaldehyde. The document was presented by Sylvia Milanez of Oak Ridge National Laboratory. The subcommittee recommends that the following revisions be made to the document. This document can be finalized if the recommended revisions are made properly.

The revised document addressed the general and specific comments provided by NRC on the previous draft. An error in the Appendix C potency calculations was identified and the authors have noted and corrected that error.

Section 3.6, page 16, lines 37–38. Explain. The numbers do not add properly as presented unless one considers “altered liver foci” equivalent to a hepatic carcinoma? Although it is the practice of some regulatory agencies to combine benign tumors and frank neoplasic lesions in animals, the approach has not been described fully, nor have references to the approach elected been provided here.

Page 16, lines 4–5. Specify the dose of crotonaldehyde with which the mice were treated topically and clarify whether the small amounts refer to crotonaldehyde or to adducts.

Page 17, lines 3–4. How does the “anticipated reactivity of croton oil and aldehyde” lead EPA (1998) to classify crotonaldehyde as a “probable human carcinogen”? Explain.

AEGL-2s. Appropriate. Twenty-seven parts per million should be referred to as the 10-min AEGL and not as the 30-min AEGL (on page A-3).

AEGL-3s. No corrections required.

Preliminary Cancer Assessment. There are some calculation errors in this section. The first one relates to the total water uptake, which should be about 39 L and not 5.537 L as reported (because total water uptake=0.049×791). The crotonaldehyde intake then will not be 0.233 g at the low dose but rather 1.7 g during the exposure period. Consequently, the resulting inhalation exposure concentrations should be around 10 and 100 mg/m³ instead of 1.32 and 13.2 mg/m³ as reported. Furthermore, it is unclear whether the cancer risk calculations were based on interspecies adjustment for the slope factor (which should be the case).

At its January 27–29, 2003, meeting, the subcommittee reviewed the revised AEGL document on hydrogen sulfide. The document was presented by Cheryl Bast of Oak Ridge National Laboratory. The subcommittee recommends that the following minor revisions be made to the document. The document can be finalized if the recommended revisions are made properly.

Appendix C. There is no mention of the Fechner coefficient or its use in the “Level of Distinct Odor Awareness (LOA)” in the SOP. Cite the NRC publication where this method has been utilized and adopted. Fechner is not cited on page 32 of the References. The Fechner LOA has been added on page 27, lines 23–30, of the hydrogen sulfide document and did not appear in the previous drafts of the AEGL.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on phenol. The presentation was made by Peter Griem of Germany. The subcommittee recommends a number of revisions. A revised draft should be reviewed by the subcommittee at its next meeting.

The emergency response planning guideline (ERPG) values are far more consistent with the phenol toxicologic profile in the present context than are the AEGL values proposed in the NAC draft. The phenol AEGL-1 at 10 min to 1 hour is virtually identical to the occupational experience

reported by Shamy et al. (1994) (5.4 ppm) in 20 adult males over 14±6 years and to the data of Ogata et al. (1986) (1.2–5.0 ppm) for 82 adult males (duration of employment unspecified in the NAC draft). However, the rationale to suggest that an 8-hour AEGL-1 where people are to experience notable discomfort (page 2) at a value less than one-half the current REL, TLV, PEL, and MAC (Table 11) has not been addressed. What “notable discomfort” is associated with the AEGL-1 at the proposed values? It appears that NAC should carefully revisit the definitions of the AEGLs and ascertain whether the AEGL values proposed are consistent with the adopted SOP.

The phenol AEGL-2 at 8 hours (7.7 ppm) said by NAC to be disabling (page 37) and to impair one’s ability to escape (page 2) is not toxicologically different from the current REL, TLV, PEL, and MAC. The burden is on the NAC to demonstrate that the slope of the phenol concentration-response relationship is such that even a very small change in an 8-hour TWA causes “irreversible, long lasting health effects.” The proposed derivation of an AEGL-2 based on reduction of the AEGL-3 is arbitrary. The NAC would benefit from serious reconsideration of the derivations of the ERPGs for phenol. Typographical errors persist (page 3, line 18) throughout the document.

Page 1. Insert a brief summary of odor thresholds for phenol in air (from page 7, lines 19, 23, 27 and 29).

Page 1, lines 3–14. Explain why phenol was selected for AEGL development; there are no documented instances of human poisoning by inhalation, other than a peculiar infant mortality where the baby was placed in a confined space (Table 2). At line 22, were the routes listed involved in these fatalities or were the deaths restricted to ingestion and/or percutaneous absorption?

Page 1, lines 3–14. Explain how phenol is handled in commercial quantities and how it is transported in bulk, and cite the numbers of plants/facilities producing or using phenol in the United States. Insert the domestic quantities produced/consumed each year. Describe the disposal of phenol—presumably as a “toxic” waste.

Page 4, line 11 and page 24, line 42. Is “phenol of phenol extraction” the same as “phenol for DNA extraction”?

Page 5, line 6. Define and explain “toxic organ” damage?

Page 6, lines 3–5. Since phenol is (apparently) a solid at room temperature, please explain how it is that phenol vapor or aerosol can be generated in quantities sufficient to present a public health hazard in ambient air.

Page 6, lines 21–28. At each entry, clarify whether the exposure was aerosol or vapor.

Page 6, line 28. Under what conditions (oral bolus, inhalation, drinking water) was it that “no teratogenic effects [signs of developmental toxicity?] were found in rats and mice”? What

strains of rodents were studied in this context? Were doses sufficiently great to justify such a broad conclusion about all aspects of developmental toxicity?

Page 6, lines 28–31. Describe the gender, strains of rodents, the organ(s) and tumor type(s) found, and duration of exposure. At line 31, explain how the “promoting activity” was measured. If carcinogenicity is an important end point for phenol AEGLs, explain the cancer risk associated with the proposed AEGLs. If cancer risk is incidental, provide a brief weight-of-evidence discussion from IARC or similar analyses on genotoxicity relative to doses/exposure duration necessary to exhibit such activity. If cancer is not an end point used in these AEGL derivations, that should be made clear.

Page 6, lines 36–37. As written, the following phrase conflicts with the SOP: “An uncertainty factor of 3 was applied for interspecies variability because a multiple exposure study was used for derivation of AEGL.” Certainly a subchronic or repeat-dose protocol that demonstrated neither clinical nor hematological nor histoapthological changes at 0.5–25 ppm would have greater (not reduced) confidence than those same results found after only a single exposure at the same concentration?

Page 6, lines 37–42 and page 7, lines 1–2. The calculations and verbal summersaults beg credibility. The NAC justifies unrealistic uncertainty factor selection, then provides additional verbiage to explain why the arguments are not persuasive. Uncertainty factor selection and justification for derivation of the phenol AEGLs should be revisited.

Page 7, lines 5–7. The NAC first faults the Hoffman et al. (1999) protocols as too extended for reliable AEGL development (page 6), then suggests brief exposure (10-min) AEGLs cannot be derived from a repeat-exposure protocol (6 hours per day). At a minimum, the text here should provide a rigorous explanation for why the 6-hour exposure at day 1 (which produced no adverse effects after another nine more such exposures) cannot be used even though it is a confirmed NOAEL for single (as well as multiple) inhalation exposures. The argument as presented could hold for a repeat-exposure protocol that induced toxicity, but the argument cannot be extended to a protocol that failed to identify any signs of toxicity even under repeated-exposure situations.

Page 7, line 10. On the basis of what empirical data does the NAC justify dividing an AEGL-3 by one-third to derive an AEGL-2? What concentration-response relation was utilized to justify that approach in the case of phenol? The AEGL-2 definition is not covered by any of the statements made on lines 10–19. The subcommittee recommends that the NAC provide a proper justification for the use of an uncertainty factor of 3.

Page 7, lines 10–13. List volunteer ages, gender, and health status.

Page 7, lines 15–16. Sentence fragment.

Page 7, line 22. No information is given for why the phenol vapor toxicity is or is not different from that of the same chemical when presented as an aerosol. If it cannot be demonstrated that there is no statistically significant difference between vapor and aerosol inhalation toxicity, a

clear explanation for why the particular aerosol concentration is both physically and biologically equivalent to the vapor concentration should be given. The comparisons between Flickinger and Brondeau as written are only circumstantial.

Page 7, line 26. From what section of the SOP was an uncertainty factor of 3 assigned because “variability for lethal effects [does this mean there is no significant difference in LD50 values across species?] after oral administration of phenol was limited”?

Page 7, line 31. Did Ruth (1986) report an air odor threshold for phenol?

Page 7, lines 35 and 41. Conflicting text.

Page 8, lines 10–17. List the duration of phenol exposure for the 39 most highly exposed individuals. Explain how this “most highly exposed” group was verified.

Page 8, line 14. Does the abbreviation “mg/d” stand for mg/day?

Page 8, line 21. List age, duration of employment in these activities, and health status of these phenol workers.

Page 8, lines 22–29. List lengths of employment in the phenol/synthetic fibers production of the Ogata plant.

Page 9, lines 10–13. Delete.

Page 9, lines 22–23 and page 10, lines 13–14. Include concentrations of phenol used in these in vitro assays.

Page 9, line 24. What does the verb “affected” mean or imply in the context used here? Explain whether the in vitro changes here occurred in the presence or absence of cytotoxicity and/or lethality.

Page 9, line 28. Does this sentence relate to total “respiratory” tract cancers (e.g., nasal, lung) or to a specific organ and tissue type?

Page 9, lines 28–32. The descriptions should include some indication of the retrospective study design, efforts to account for confounding (e.g., cigarette smoking), ages of these workers, numbers of subjects, and lengths of employment in these plants (e.g., duration at line 32)?

Page 10, line 3. What were the “white blood cells” that were counted (leukocytes, lymphocytes, etc.)?

Page 11, line 1. What assumptions did Flickinger (1976) make about percutaneous absorption of phenol when calculating the dermal LD₅₀?

Page 11–13, line 7. Why is it necessary to include this level of detail for these 1940s studies— especially considering none of the AEGLs are based on any of these oral data? Delete tables and condense the text presentation.

Page 11, lines 19–34. Delete.

Page 14, line 23. Indicate exposure duration each day (e.g., 24 hours/day).

Page 14, line 24. Delete background information on groups exposed to other materials and explain if the protocol exposure was via face mask, whole body, etc. Describe the chamber construction if whole-body protocols were used and whether the animals were confined to chairs and sedentary or were free to move about. At page 14, line 24, and page 17, line 11, what is the reliability of the colorimetric method used to measure phenol in air? This could be considered a key study if not for the antique analytical method used, but if that can be established as a reliable technique at the phenol concentration range of concern, the value in this controlled primate inhalation study could be used as one (but certainly not the highest) NOAEL in primates.

Page 14, lines 24–25; page 17, lines 19–20; and page 18, line 2. It is not necessary to describe the exposures to chemicals other than phenol.

Page 15, lines 34–37. This paragraph can be deleted inasmuch as the TSCA report could not be obtained and no useful information to AEGL derivation was evident from the brief sketch here.

Page 16, line 17. The CMA (1998) report/Hoffman et al. (1999) abstract was published in the peer-reviewed literature by Hoffman et al. (2001) (copy enclosed).

Page 16, line 18. The sentence “Following...sacrified.” should be moved to line 17.

Page 16, lines 31–34. Have this document’s authors made a distinction between chromodacyorrhea and red nasal discharge? Stressed rats commonly show evidence of chromodacyorrhea (i.e., red pigment) around the eyes and nares. The histopathological examination of the nasopharyngeal tissues would have yielded evidence of vascular damage and hemorrhage had the red nasal discharge been blood.

Page 16, line 39; page 22, line 24; page 17, line 9; page 17, line 21; and page 18, line 3. List hours/day of phenol exposure and whether a concurrent control group was included in the study design.

Page 17, line 1. What is a “labile sitting position” for a rat? How is it that one measures “disturbed grooming patterns” for a rat? Delete.

Page 17, lines 10, 21, and 27; and page 18, lines 3 and 9. Were these whole-body, head-only or another type of exposure?

Page 21, line 15. Ryan et al. (2001) (copy enclosed) recently published the findings of a 2-generation reproduction and immunotoxicity study of rats that ingested phenol in drinking water. The study results were essentially negative. This investigation should be described here.

Page 21, lines 9 and 23. Indicate whether the authors evaluated internal soft organs (e.g., Staples or Wilson technique) and measured fetal skeletal ossification.

Page 21, line 21. How many of the dams given phenol at 280 mg/kg died?

Page 21, lines 34–36. Did these changes occur in the absence of cytotoxicity in the Tsutsui study?

Page 21, line 9 and page 22, lines 37–39. List maternal parameters studied and indicate whether signs of maternal intoxication (e.g., reduced food consumption, reduced body weight gain, muscle twitching, CNS depression) were seen after phenol gastric intubation.

Page 21, lines 12–14 and page 22, lines 37–39. The data should be summarized to indicate clearly that signs of developmental toxicity in rats were evident only after doses that also produced overt maternal phenol intoxication.

Page 22, lines 10–13. Was benzene used as an initiator here? If not, what was the initiator used to study phenol promotion? Was benzene simply a solvent that confounds interpretation of the phenol applications? Describe in detail the “skin lesions” associated with topical phenol— were these pathologies so severe that the response seen is actually the result of overt dermal damage and subsequent regenerative hyperplasia?

Page 23, line 2. Please insert the NCI conclusion (e.g., clear, equivocal, no evidence) for carcinogenicity of oral phenol in rats and mice. Please insert the ACGIH and IARC weight-of-evidence classifications for phenol carcinogenicity.

Page 24, line 14. The 60–88% values represent percentage systemic uptake/absorption, rather than “retention rate.”

Page 24, lines 17, 22, 34. Provide details on gender, numbers of volunteers, ages, and smokers/nonsmokers in this population.

Page 24, line 19. It is important to note here that systemically absorbed phenol is extensively eliminated in the urine unchanged (other than being conjugated). Piotrowski (1971) reported 99% recovery of phenol in the urine of human subjects. This argues against hydroxylated metabolites (e.g., catechol, hydroquinone, benzoquinone) playing a significant role in acute systemic toxicity.

Page 24, line 25. By “percutaneous clearance,” do the document’s authors mean “percutaneous absorption”?

Page 24, lines 25. What is “percutaneous clearance”?

Page 24, line 37. What does the “80%” value refer to?

Page 25, line 1. Why are these values repeated here when they were already presented on page 4, line 20?

Page 25, line 8. As written, the sentence is confusing. If one sums the fractions of administered dose recovered (56+85%), it is not clear how the authors found more than the original administered dose of phenol in liver and in plasma?

Page 25, lines 11–18. The information presented in this paragraph on phenol metabolism is accurate, but more recent references should be cited. Kenyon et al. (1995) (copy enclosed), for example, characterized the dose-, route-, and gender-dependency of phenol metabolism in B6C3F1 mice. These findings are relevant to the topic discussed in lines 19–24 of page 26.

Page 25, line 41. Describe the extent of necrosis. Was the response seen in animals given repeated skin applications of phenol (page 22, line 11) similar to the skin burns observed in humans?

Page 26, lines 19–24. The cytotoxicity of phenol and its oxidation products in the bone marrow has received quite a lot of attention. It would be worthwhile to summarize some of those findings here. Eastmond et al. (1987) (copy enclosed), for example, were among the first to report that phenol appears to stimulate myeloperoxidase-catalyzed conversion of hydroquinone to a phenoxyl radical. This moiety binds to protein and DNA, apparently leading to bone marrow suppression.

Page 27, lines 5–6. Move this conclusion to page 29, line 41.

Page 27, line 17. Use a term other than “hypersensitivity” to describe the vulnerability of the newborn with a congenital pulmonary adaptive disorder. Hypersensitivity is synonymous with allergy.

Page 27, lines 33–34. Move this conclusion to page 30, line 1.

Page 28, lines 30–33. It should be pointed out that the “red nasal discharge” was very likely a manifestation of chromodacyorrhea. As mentioned previously, the presence of this pigment around the eyes and nares is commonly seen in stressed rats.

Page 28, lines 35–38. It would probably be better to include this information on nonhuman primates as the initial paragraph in this subsection (5.2).

Page 29, line 8. To what tissue or type and severity of irritation does the reference to rats pertain? What are the “higher” concentrations referenced here?

Page 29, line 10. As written, the text could be interpreted to suggest that the AEGL is based on an unreviewed abstract from 1999. What is the status of the full manuscript? At a minimum, the authors should indicate whether the CMA (1998) study was conducted under GLP.

Page 29, lines 12–14. It is stated here that (1) red nasal discharge in male rats was not dose-dependent; and (2) only one male showed this effect. This conclusion contradicts the information presented in lines 31–34 of page 16.

Page 29, line 14. To what does the phrase “and because only one males, but not females” refer?

Page 29, line 23. Move the sentence “Due to...exposure times” to line 36 after “...Appendix A.”

Page 29, lines 41–42. Describe the specific section of the SOP where a repeat-exposure protocol in an animal study justifies an uncertainty factor of 3 for intraspecies scaling?

Page 29, lines 41–42 and page 30, lines 1–7. It is not necessary to rationalize the use of an intraspecies uncertainty factor of 3 rather than 10. A factor of 3 is typically used when mild irritation at the initial site of contact is the toxic end point.

Page 30, lines 32–35: The NAC cannot justify the use of formaldehyde data to derive or support any aspect of a phenol AEGL. Delete.

Page 32, lines 1–5. There is not a suitable database for derivation of AEGL-2 values. Nevertheless, it may be most appropriate to utilize the 234 ppm exposure level of Flickinger (1976) as the starting point. Mention should be made in this paragraph that the slight loss of coordination in rats at 4 hours progressed to tremors and prostration (but not death) after 8 hours of exposure.

Page 32, line 23. It is worthy of note that Hoffman et al. (2001) saw no adverse effects in male or female F-344 rats that inhaled phenol at up to 25 ppm 6 hours daily for 10 days over a 2-week period. The recommended AEGL-2 values in the current document are all lower than 25 ppm.

Page 32, lines 25–28. This information should be deleted. It makes no sense to consider or to support AEGL-2s based on no-effect exposure levels in a subacute study (i.e., that of Hoffman et al. 2001).

Page 33, lines 1–5. What were the characteristics of the 17 people most affected (gender, age, health status)? If children or infants or others reported with gastrointestinal distress, then it appears they possibly received the highest administered dose? How was the 3.4 mg/kg·day (a rather precise dose metric) determined? Was there evidence for a dose-response relation among the group of 39?

Page 34, lines 1–9. Unfortunately, there do not appear to be LC₅₀ values or other inhalation data on which to base AEGL-3s. It should be recognized that Flickinger (1976) saw no fatalities in rats that inhaled 234 ppm for 8 hours. Similar effects (mild incoordination) were seen by Flickinger and Brondeau et al. (1990) in rats inhaling similar concentrations (234 and 211 ppm) for 4 hours. Rats have substantially higher respiratory and cardiac output rates than do humans, so the rats’ systemically absorbed dose (and toxicity) would be substantially greater than that of humans at the same vapor exposure level. Thus, the proposed AEGL-3 values are far below what would be an acute hazard to humans. As noted previously, Hoffman et al. (2001) observed no adverse effects in rats inhaling up to 25 ppm 6 hours daily for 10 exposure days. The proposed 4- and 8-hour AEGL-3 values are 29 and 23 ppm, respectively. They do not pass “reality checks.”

Page 34, lines 11–23. The calculation of the total dose absorbed systemically by female rats during a 900-mg/m³, 8-hour inhalation exposure is quite reasonable. It is not reasonable, however, to compare a dose absorbed over 8 hours with an acute oral LD₅₀. The oral dose is given as a bolus, resulting in attainment of a relatively high peak blood level. In contrast, inhalation results in constant infusion of a fixed amount of chemical over a prolonged period. As phenol is rapidly metabolized, blood levels in animals inhaling the chemical remain relatively low, and systemic effects are relatively mild. Sanzgiri et al. (1995) (copy enclosed) conducted a pertinent study in which rats received the same systemic dose of carbon tetrachloride (CCl₄) as an oral bolus and over 2 hours by inhalation. The gavaged rats exhibited much higher peak blood CCl₄ levels and greater liver damage than did their inhalation counterparts. The systemic toxicity of phenol is also dependent on its peak blood level, rather than on its area under the blood concentration (versus time) curve (AUC). The EPA (2002) (copy enclosed) acknowledges this and points out in its recent IRIS assessment that phenol is much more toxic when given by gavage than when given in drinking water (i.e., over a prolonged period). Berman et al. (1995) (copy enclosed), for example, reported renal and hepatic pathology in female F-344 rats given phenol at 40 mg/kg/day for 14 days by gavage. In contrast, no histopathological changes were seen in these organs in F-344 or Sprague-Dawley rats given much higher doses in drinking water (NCI 1980; Ryan et al. 2001).

Page 34, lines 29–40. In light of the comments above, the validity of the assumptions here is highly suspect. The use of the Flickinger (1976) study as the basis for calculation of 8-hour AEGL-3 values is questionable.

Page 34, lines 5–8. Was this among smokers or nonsmokers?

Page 38, Figure 1. Where is the “double triangle” referenced on page 37, line 20? It is not discernable in the figure itself or in the legend to the figure.

Page 48, lines 20–21. Please discuss the skin burns in the section on page 4, lines 23–29.

Page 71, line 28. What empirical basis or SOP directive determines whether a particular selection/identification of uncertainty factor is “considered adequate”?

Conclusions on AEGL-1 Values: Data indicating the absence of histopathological effects in a 2-week animal study have been used to derive AEGL-1. It is important to look for data on the irritation/discomfort relating to phenol exposures and to use them as the basis for AEGL-1 derivation. Isn’t the TLV based on irritant effects of phenol? The subcommittee recommends that the choice of Hoffman et al. (1999) study as the key study as well as the rationale for the uncertainty factors applied be reconsidered. The NAC should reconsider the human data and review the basis for other exposure values, such as the current REL, TLV, PEL, and MAC values for phenol.

Conclusions on AEGL-2 Values: AEGL-2 values are derived from the AEGL-3 values. The latter is based on the study by Flickinger (1976). The subcommittee questions the determination of exposure concentration in this study (see AEGL-3 discussion). The approach and rationale for deriving the AEGL-2 values, which is based on reduction of the AEGL-3 values needs to be reconsidered.

The phenol AEGL-2 for 8 hours (7.7 ppm) that is considered by the NAC to be disabling and to impair one’s ability to escape is not toxicologically different from the current REL, TLV, PEL, and MAC. The burden is upon the NAC to demonstrate that the slope of the phenol concentration-response relationship is such that even a very small change in an 8-hour TWA causes “irreversible, long lasting health effects.” The proposed derivation of an AEGL-2 based on reduction of the AEGL-3 as written is arbitrary. The NAC should take into consideration how ERPG values were derived for phenol. The approach for deriving the AEGL-2 could be acceptable as a default, only if relevant data are not available.

In the rationale for derivation of AEGL-2s, the RD₅₀ (166 ppm) is not mentioned. Irritation is a key factor in AEGL-2 because it can cause escape impairment. Alarie and coworkers have recommended 0.3 times the RD₅₀ for EELs; however, this was for workers (Kane, 1979). Generally, a one-hour level-2 for an emergency exposure can be about 1/5^th of the RD₅₀. The proposed value (for 1 hour) is about 1/10^th of the RD₅₀. Therefore, the AEGL-2 values could be higher.

Conclusions on AEGL-3 Values: The key data used to derive AEGL-3 values are taken from Flickinger (1976). The use of this study as the basis for calculation of 8-h AEGL-3 values is questionable, primarily due to the determination of the exposure concentration. The study should be further evaluated. The subcommittee believes that there is inadequate justification of uncertainty factors and that the resultant AEGL-3 levels are too low. The concentrations reported by Flickinger (1976) were nominal, described in section 3.2.2 as determined by weight loss of the solution divided by the air volume. Was the concentration determined by correcting for the loss of material due to deposition of phenol on the equipment or was this correction not made? The use of nominal concentrations should be avoided if other data exist that can be better relied upon.

The Flickinger (1976) study involved an 8-hour exposure of rats to 900 mg/m³ of aerosol. In a liquid aerosol exposure, the rats would have been soaking wet with phenol; phenol can be lethal

when applied on the skin. This exposure was the equivalent to a combined inhalation, dermal, and oral study. Yet, there were no deaths. Therefore, the maximum non-lethal concentration for this study would have been significantly higher, probably at least a factor of two. It appears that the AEGL-3 levels could be increased substantially.

However, there do not appear to be LC₅₀ values or other concentration-time inhalation data on which to base AEGL-3s. It should be recognized that Flickinger (1976) observed no fatalities in rats which inhaled 234 ppm for 8 hours. Similar effects (mild incoordination) were seen by Flickinger and Brondeau et al. (1990) in rats inhaling similar concentrations (234 and 211 ppm) for 4 hours. Rats have substantially higher respiratory and cardiac output rates than do humans, so the systematically absorbed dose (and toxicity) in rats would be substantially greater than humans at the same vapor exposure level. Thus, the proposed AEGL-3 values are far below what would pose an acute hazard to humans. As noted previously, Hoffman et al. (2001) observed no adverse effects in rats inhaling up to 25 ppm 6 hours daily for 10 exposure days. The proposed 4- and 8-hour AEGL-3 values of 29 and 23 ppm, respectively, do not pass “reality checks.” The magnitude of the total uncertainty factor (10: interspecies×interspecies) is not properly justified.

At its January 27–29, 2003, meeting, the subcommittee reviewed the revised AEGL document on cyclohexylamine. The document was presented by Sylvia Milanez of Oak Ridge National Laboratory. The subcommittee recommends a number of revisions. The document can be finalized if the subcommittee’s recommended revisions are made properly.

Page v, line 20. Clarify the cyclohexylamine (CHA) exposure levels for Groups I and II.

Page 1, line 22. The meaning of the abbreviation “CHA” should be spelled out the first time it is used in the text.

Page 2, lines 37–38. The word “other” should be used to refer to chemical operators other than the individual who experienced symptoms. See changes made in the text.

Page 18, line 29. It would be useful to the reader for the authors to distinguish between sensory irritation and deep lung damage and/or systemic toxicity (if it occurs). See change made in the text.

Page 19, lines 1–2. Asthmatic patients and individuals with other respiratory difficulties would be expected to be more sensitive than healthy persons to a respiratory irritant like CHA. Might some persons who are exposed multiple times become allergic to CHA?

Page 19, lines 23–25. Dermal/systemic absorption of CHA would not be expected to “add to” irritant effects of the inhaled chemical.

Page 22, lines 39–40. The phrases “red nasal discharge” and “dried red facial material” are used here and in a number of other places in the document. The authors chose to retain the phrase, because they felt the condition was something other than chromodacyorrhea. Chromodacyorrhea is a very common manifestation of stress in rats. Rats exhibit the pigment around both their eyes and nares. If blood were discharged from the nose, it would turn dark/black within a short period of time. Histopathologic examination would reveal hemorrhaging in the nasopharyngeal area.

Page 26, line 26. It would be informative to summarize the manner in which ACGIH, NIOSH, and Germany derived their CHA occupational exposure guidelines. Each 8-hour guideline is higher than the 8-hour AEGL-1, -2, and -3. An explanation should be given for this substantial deviation from the previously recommended levels. The ACGIH value appears to have been based solely on the Watrous and Schulz (1950) study, but it would be interesting to learn whether the German MAK was obtained from more recent data.

Page 28, lines 26–35. A statement should be added that confidence in the AEGLs is low and should state clearly the reason for the low confidence. Apparently, low exposure levels are recommended because of the high uncertainty associated with the data?

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on ethylenediamine. The document was presented by Sylvia Milanez of Oak Ridge National Laboratory. All previous comments of the Subcommittee have been addressed. The subcommittee recommends that the section on data quality and research needs be revised to emphasize the limitations of the available data and the need for specific studies to be conducted to strengthen the existing database. The document can be finalized after the subcommittee’s recommended revisions to the data quality and research needs section have been made properly.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on HFE-7100. The document was presented by George Rusch of Honeywell, Inc. The subcommittee recommends a number of revisions. The revised document will be reviewed by the subcommittee at its next meeting. All previous comments of the subcommittee have been addressed. The document can be finalized if a justifiable argument for the use of modifying factor for deriving AEGL-1 is made.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on carbon monoxide. The document was presented by Peter Griem of Germany. The subcommittee recommends the following revisions. A revised draft should be reviewed by the subcommittee at its next meeting.

This document recommends that no value be established for the AEGL-1 because the first clearly evident symptoms in humans are life-threatening and would provide no margin of safety. It is also pointed out that the early symptoms in a sensitive population (impaired cardiac status) occur at carboxylhemoglobin levels seen in smokers and occupationally exposed workers. Another reason for not setting AEGL-1 levels on the basis of COHb would be the significant variation that occurs within individuals exposed to CO. The NRC subcommittee finds that in treating carbon monoxide exposure in the ER, headache and/or nausea and vomiting is a better predictor of CO exposure than COHb, which is variable depending on treatment during transport, or cherry-red liver (which is stressed in textbooks but is rarely documented).

The proposed AEGL-2 levels are based on findings in patients with coronary artery disease, which this document considers to be the most sensitive population, although infants, pregnant women, and the elderly should also be considered sensitive populations. Humans are also used for the AEGL-3 values, but those values are based on threshold lethality in normal subjects rather than in individuals with impaired cardiac status. Although the NRC subcommittee agrees generally with the rational for not recommending AEGL-1 values, there remain concerns with the recommendations for the AEGL-2 and -3. The first is that using cardiac patients for AEGL-2 values and normal humans for AEGL-3 values is comparing different subgroups because the individuals with impaired cardiac function would likely be more susceptible than normal individuals to myocardial infarction, although the papers by Dahms et al. and Ebisuno et al. do not seem to support that hypothesis. The second is that the discussion of the methodology used to convert COHb values to ppm and mg/m³ ignores the substantial individual variation that exists in COHb levels in subjects exposed to CO and the difficulty of obtaining reliable data in the field. The third concern is the reliance on very old data to set the AEGL-3 values (Haldane’s studies were pre-1900 and involved a single subject, himself). The protocols currently used in the emergency room and those used by HAZMAT teams rely not only on COHb values but use signs of toxicity in the two most sensitive organs (CNS and myocardium) to diagnose and treat CO exposure.

Although it might be possible to establish an AEGL-1 on frontal headache or nausea and vomiting since these effects can occur in patients who show no elevation of COHb, the subcommittee concurs with the NAC that the “not recommended” designation is appropriate. The ST-segment change and angina criteria used for the AEGL-2 is reasonable and the uncertainty value of 1 is appropriate. The authors conclude that an interspecies value of 1 would account for all sensitive species and provide supporting data for infants. However, pregnant women and the elderly are also sensitive species and might not be covered by this factor. The AEGL-3 is based on the papers by Chiodi et al. and Haldane using normal subjects, and it seems that the total uncertainty factor of 3 may be too low. After considering the references from the poison control protocol on CO and the documentation from the ACGIH TLV and BEI, there were references on the incidence of myocardial infarction in patients with impaired cardiac function but not much more. It is interesting however that researchers at the University of Pennsylvania and Harvard reported (January 20, 2003) that low-level exposure to CO will facilitate the repair of arteries and other vessels that have been damaged in angioplasty and transplants. One general comment is that the document seems to place excessive emphasis on the results in animals considering that these results are not used to establish or even to validate AEGL values from the human studies.

There was concern about the conservative nature of the AEGL-2 and -3 values and questions regarding the justification for the low values even though they are in the same range as other reference standards. It was suggested that this documentation be reviewed by an expert cardiologist with ER experience. Given the wealth of data on CO, it is difficult to believe that there aren’t more current studies that could be used to derive the AEGL values rather than Haldane and Chioti et al. Some of the Aranow et al. studies were not used by EPA in developing the CO NAAQS values because of controversy regarding the findings. If necessary for the current AEGL document, the authors should contact Lester Grant at EPA (North Carolina) to obtain further details about the studies and controversy. It was also recommended that the introduction to this TSD be enhanced to focus on manufacturing sources rather than motor vehicles given that the process should not be focused on non-point or endogenous sources.

Page vii. The statement that CO is “tasteless, nonirritating, colorless, and odorless” appears so many times in this document that it is like a mantra. This point needs to be reduced, changed, or fixed so that is not annoying. Also, part of this statement is that exposures occur occupationally without specifics. It would be more informative if specific occupations were mentioned.

Executive Summary (ES), page viii, Summary Table. Expand the definition of “not recommended” to be consistent with the language used in the previously published AEGL documents when describing that AEGL-1 values were not derived. In this case, NAC might use “not recommended due to high acute toxicity.” This descriptor should be added as a footnote to the summary table here, in the text on page 40 (Table 14), and throughout the text where the AEGL-1 values are given or discussed.

ES, page viii, Summary Table. Is it possible to include the corresponding COHb levels in the table for each of the AEGL time frames? It will help the reader understand the discussion in the text that refers only to COHb levels and not to ambient exposure levels of CO.

ES, page viii, lines 5–8. The explanation given for the many case reports not used to derive the AEGL-3 value is not clear (poorly worded). The text should make it clear that the reasons for not using them included uncertainties about end of exposure COHb levels, whether oxygen had been provided, and how much time had passed prior to measuring blood COHb levels after the initial CO exposure. There is a much clearer explanation for why the anecdotal studies in humans (the case studies) were not used to derive the AEGL-3 values in the text (page 44, lines 30–32). That wording could be used here as well.

Page 1, lines 3–13, Introduction. Add something about industrial exposures to CO. It is not clear how, or why, there would be community-wide exposures to CO given this general background description.

Page 3: The source of Table 3 needs to be identified on the table.

Page 3, Section 2. Human Toxicity Data Table. All “exposure” values are defined in terms of COHb levels. No ambient CO levels are provided. Since final AEGL values are in ambient ppm values, data and discussion should include ppm values as well as COHb values.

Page 7, line 22. What is meant by ST-segment changes in the electrocardiogram? Briefly explain this term since this is a key health outcome.

Page 7, line 24. The explanation for why adults with coronary heart disease are the “most relevant for AEGL derivation” is not given. Expand this explanation and include proper references to the open peer-reviewed literature.

Page 7, line 25. Say more about the “other effects and additional high-risk groups” that have also been investigated. The text here is unclear. Are these other effects irrelevant or just redundant of what is in text?

Page 7, line 36. Typo “was” should be “as.”

Page 9, line 40. Typo “continuous.”

Page 10. Why is it that the Stewart et al. studies (25 total) were not used in derivation of the AEGLs?

Page 17, line 2. Typo “there” should be “these.”

Page 17, Table 6, line 27. What is “nocicptive”? Is that a typo?

Page 20, Table 10. Is there any way to include a comment on when the treatment occurred in this summary chart? In two of six cases, it is clear that oxygen was provided in an ambulance; in the other cases, did the treatment begin “at time of exam after exposure”?

Page 21, Section 2.6, Summary. This summary does not provide data to support the statement that children are more sensitive than adults to CO exposure.

Page 25, Table 12, line 18. Typo. Exposure time should be 15 min, not 30.

Page 28, line 37. Typo “the” should be “they.”

Page 30, lines 32–33. Typos “pubs” should be “pups.”

Page 31, line 21. Add “in animals” to statement.

Page 31, line 25. Add “in animals” to statement.

Page 34, line 3. This is ambiguous; either a word is missing or possibly a typo (“and” should be “an”).

Page 34, Section 4.3, mathematical models for COHb formation. Is this section necessary, especially in such detail? This section is very technical and does not seem germane or integral to understanding the derivation of AEGL values. The subcommittee suggests dropping it or substantially cutting it down with the idea of making it simpler and easier to understand.

Page 38, line 33. Typo “anemia.”

Page 39, line 9. The text refers to “The LC₅₀ values shown in Figure 1.” Actually, Table 12 shows the LC₅₀ values. Figure 1 shows a log representation of the LC₅₀ values. The text should be more precise.

Page 39, Section 4.4.3, time scaling. This paragraph comes after the introduction of the model used to determine COHb levels for different exposure times. It would be more logical to move it up in the text prior to that discussion. Tell the reader that time scaling is not going to be used in this instance, provide the reason for that decision, and then go into discussion of the model.

Page 40, Summary Table. As discussed in the ES, expand the definition of “not recommended” to be consistent with the language used in previous documents when describing that AEGL-1 values were not derived. In this case, NAC might use “not recommended due to high acute toxicity.” COT used a similar descriptor for arsine and in one or two other documents. This descriptor should be added as a footnote to the summary table here, in the ES, and throughout the text where the AEGL-1 values are given or discussed.

Page 44. Why not use the results of Atkins and Baker, Ebisuno et al., and Grace and Platt to establish the AEGL-3 rather than the data of Haldane and Chiodi?

Page 44, lines 17–20. This paragraph describes nonlethal studies in animals. These studies should not be discussed in the section on the animal data relevant to the AEGL-3 values, which are based on lethality.

Page 44, lines 30–32. This is a good explanation for why the anecdotal studies in humans (the case studies) were not used to derive the AEGL-3 values. This explanation should be included in the Executive Summary as well as on page 44.

Page 44, lines 34–35. The text here refers to “newer” studies that are not available. The implication here is that there are other relevant data that is more recent, but not “available.” What does “not available” mean in this case? Reword to make it clear what is intended here.

Page 45. The statement that an uncertainty factor of 3 was used rather than 10 because the value derived using 10 would be in the range found in the environment is weak and arbitrary. The statement on page 50 may be considered appropriate here.

Page 45, lines 1–4. The explanation for only using an UF of 3 is not scientific. As discussed previously, the explanation sounds like “we didn’t like the data we got, so we decided not to use it.” The text should be reworded.

Page 48, Table 18. Several of the values are not consistent (see WHO Air Quality Guidelines and AEGLs; EEGLs and AEGLs; MAK at 30 min and 30-min AEGLs). The text should discuss these differences and offer explanations for why they exist.

Page 51. The conclusion about lack of exposure duration is justified even though there is considerable duration information included in the human effects section of this report.

Page 64, Appendix B. This appendix provides formulas used to calculate the ambient CO exposure values (the AEGLs in this case) for different blood COHb values. Yet how these values are derived is not clear from the formulas provided in this appendix. At least one sample calculation for each AEGL should be provided in this appendix (or in Appendix A, where a similar calculation is typically shown for the time scaling extrapolations done using the ten Berge equation).

Appendix C, page 76, lines 1–9. Why is this detailed explanation for why several studies were not used to derive the AEGL included here? This is confusing and not germane to the derivation of the AEGL value. Delete and only discuss the studies that were used.

The papers for the AEGL-2 should be rechecked to insure that they support a COHb level of 4 since that is very low.

The current wording emphasizes the symptoms reported in the Chiodi and Haldane studies (see Section 7.1). In the paper by Chiodi et al., no mention is made of what, if any, symptoms were experienced. This should be clarified in the text in Section 7.1.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on ethylenimine. The document was presented by Sylvia Milanez of Oak Ridge National Laboratory. The subcommittee concurs with the NACs recommendation that an AEGL-1 not be recommended. However, the subcommittee recommended that an attempt be made to develop an LOA. The document can be finalized after the recommended revisions have been made properly.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on propylenimine. The document was presented by Sylvia Milanez of Oak Ridge National Laboratory. The subcommittee concurs with the NACs recommendation that an AEGL-1 not be recommended.

However, the subcommittee recommended that an attempt be made to develop an LOA. The document can be finalized after the revisions have been made properly.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on allylamine. The document was presented by Sylvia Milanez of Oak Ridge National Laboratory. The subcommittee recommends the following revisions. The document can be finalized after the revisions have been made properly.

Page 14, lines 14–16. Guzman et al (1961) did not see cardiovascular lesions in rabbits or in monkeys subjected repeatedly to allylamine (AA) vapor at 40 ppm. Boor and Hysmith (1987), however, reviewed numerous studies describing such lesions in several species administered high doses of AA by various routes (see page 15, lines 23–28). The introductory sentence to the summary (page 14, lines 1–16) should be amended to reflect this information.

Page 21, line 42 and page 22, line 14. Derivation of the AEGL-2s would be easier to understand if the explanation of the 10-, 30-, and 60-min values methodology and the 4- and 8-hour methodology were placed into separate paragraphs. See suggested paragraph beginnings in the text.

At its January 27–29, 2003, meeting, the subcommittee reviewed the AEGL document on chlorine dioxide (CD). The document was presented by Cheryl Bast of Oak Ridge National Laboratory. All previous comments by the subcommittee have been addressed; however, a computational correction is needed to the application of the interspecies uncertainty factor. The document can be finalized after the recommended revisions have been made properly.

Page v, line 13. ATSDR (2002) is not included in the References. The final version will probably be published in 2003.

Page 12, lines 12–41. It is recommended that the authors of this document further explore the chemistry of CD in inhaled air. What is the source of the statements that inhaled CD dissociates into chlorine gas and oxygen? It would be worthwhile for the authors to determine the yield of chlorine from inhaled CD. Compare the AEGLs for inhaled CD and chlorine on a molar basis. If the AEGLs for CD are substantially lower than the chlorine AEGLs, that

might argue in favor of the use of a total uncertainty factor of 10 rather than 20. This could be accomplished by omitting the modifying factor of 2.

Page 13, lines 8–11. If possible give more specific information on the extent to which asthmatic patients are more susceptible to inhaled chlorine than are healthy individuals. This should provide support for the use of the intraspecies uncertainty factor of 3. The subcommittee typically advocates the use of an uncertainty factor of 3 to protect asthmatic patients, although a factor as high as 5 has been recommended for some chemicals.

Page 14, lines 14–17. The phrase “supported by the steep concentration-response curve (0% mortality ...for 6 hr)” should be deleted. A steep response curve for lethality would argue for a factor larger than 3. The phrase “which implies little individual variability” should also probably be deleted. The applicability of the lethality study results to minor irritation is questionable. Corresponding changes in the Executive Summary should be made.

Page 14, lines 24–26. Retain the two sentences in these lines. They provide important information that supports the use of inter- and intraspecies uncertainty factors of 3 and 3. Add a sentence describing protection of asthmatic subjects.

Page 15, lines 25–26. Mention of the steep concentration-response curve for lethality should be avoided. This is not supportive of an intraspecies uncertainty factor of 3.

Section 3.6. The document’s authors note here and elsewhere that the toxicity database for CD is sparse. It is recommended that the authors add a statement that their confidence in the database is low (hence the conservative proposed AEGLs.) The low degree of confidence can also be pointed out in a new section entitled “Data Quality and Research Needs.” Specific research needs should also be included there.

Page 18, lines 13–14. Ascertain the bases for calculation of the MAK and MAC. They might be derived from more current data (than are the NIOSH, OSHA, and ACGIH values) and may be more appropriate than the U.S. values.