EPA/635/R-02/006
c/EPA
TOXICOLOGICAL REVIEW
OF
Phenol
(CAS No. 108-95-2)
In Support of Summary Information on the
Integrated Risk Information System (IRIS)
September 2002
U.S. Environmental Protection Agency
Washington D.C.
-------
DISCLAIMER
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use. Note: This document may undergo revisions in the future. The most
up-to-date version will be made available electronically via the IRIS Home Page at
http://www.epa. gov/iris.
-------
CONTENTS - TOXICOLOGICAL REVIEW FOR PHENOL
(CAS No. 108-95-2)
Foreword vi
Authors, Contributors and Reviewers vii
1. INTRODUCTION 1
2. CHEMICAL AND PHYSICAL INFORMATION RELEVANT TO
ASSESSMENTS 2
3. TOXICOKINETICS RELEVANT TO ASSESSMENTS 5
3.1 Absorption 5
3.2 Distribution 10
3.3 Metabolism 12
3.4 Excretion 23
4. HAZARD IDENTIFICATION 24
4.1 Studies in Humans - Epidemiology, Case Reports, Clinical Controls 25
4.1.1 Oral 25
4.1.2 Inhalation 27
4.2 Pre-chronic, Chronic Studies and Cancer Bioassays in Laboratory Animals
32
4.2.1 Oral 32
4.2.2 Inhalation 52
4.2.3 Dermal 60
4.3 Reproductive/Developmental Studies 61
4.4 Other Studies 72
4.4.1 Initiation/Promotion Studies, Other Short-term Tumorigenicity
Assays, and Cancer Mechanism Studies 72
4.4.2 Genotoxicity 75
4.4.3 Neurological Effects 83
4.4.4 Immunotoxicity 84
4.4.5 Other Studies 85
4.5 Synthesis and Evaluation of Major Noncancer Effects and Mode of Action
86
4.6 Weight of Evidence Evaluation and Cancer Characterization - Synthesis of
Human, Laboratory Animal and Other Supporting Evidence, Conclusions
about Human Carcinogenicity, and Likely Mode of Action 92
4.7 Susceptible Populations 95
4.7.1 Possible Childhood Susceptibility 95
in
-------
4.7.2 Possible Gender Differences 96
5. DOSE RESPONSE ASSESSMENTS 96
5.1 Oral Reference Dose (RfD) 97
5.1.1 Choice of Principal Study and Critical Effect 97
5.1.2 Method of Analysis - Benchmark dose 100
5.1.3 RfD Derivation 100
5.2 Inhalation Reference Concentration (RfC) 102
5.2.1 Choice of Principal Study and Critical Effect 102
5.2.2 Methods of Analysis 103
5.3 Cancer Assessment 103
6. MAJOR CONCLUSIONS IN CHARACTERIZATION OF HAZARD AND DOSE-
RESPONSE 104
6.1 Human Hazard Potential 104
6.1.1 Oral Noncancer 104
6.1.2 Inhalation Noncancer 106
6.1.3 Cancer 107
6.2 Dose-response 107
7.0 REFERENCES Ill
Appendix A. Summary of External Peer Review Comments and Disposition 124
Appendix B. Benchmark Dose Modeling Results 129
Appendix C. Benchmark Dose Modeling Output 133
IV
-------
List of Tables and Figures
Table 1. Physical Properties and Chemical Identity of Phenol 4
Table 2. Summary of Oral Toxicity Studies 34
Table 3. Total Activity Counts in Rats Provided Phenol in Drinking Water
(ClinTrials BioResearch, 1998) 41
Table 4. Individual Data on Dehydration and Week 4 Motor Activity in Rats Provided
Phenol in Drinking Water
(ClinTrials BioResearch, 1998) 43
Table 5. Effects of Phenol Exposure on Spleen Cellularity
and Selected Blood Parameters in Mice and Rats 47
Table 6. Effects of Phenol Exposure on Spleen Cellularity and Selected Blood Parameters
in Mice and Rats 49
Table 7. Summary of Inhalation Toxicity Studies 54
Table 8. Selected Results of Two-Generation Drinking Water Study
(Ryan et al., 2001; IIT Research Institute, 1999) 63
Table 9. Key Results in Argus Research Laboratories, (1997) Rat Developmental Toxicity
Study 67
Table 10. Key Results from Developmental Toxicity Study in Rats
Administered Phenol by Gavage (NTP, 1983a) 69
Table 11. Summary of Genotoxicity Studies 77
Figure 1. Metabolism of Phenol 13
Figure 2. Plot of severity with dose for drinking water or gavage 90
-------
FOREWORD
The purpose of this Toxicological Review is to provide scientific support and rationale
for the hazard identification and dose-response assessment in IRIS pertaining to chronic
exposure to phenol. It is not intended to be a comprehensive treatise on the chemical or
toxicological nature of phenol.
In Section 6, EPA has characterized its overall confidence in the quantitative and
qualitative aspects of hazard and dose-response. Matters considered in this characterization
include knowledge gaps, uncertainties, quality of data, and scientific controversies. This
characterization is presented in an effort to make apparent the limitations of the assessment and
to aid and guide the risk assessor in the ensuing steps of the risk assessment process.
For other general information about this assessment or other questions relating to IRIS,
the reader is referred to EPA's IRIS Hotline at 301-345-2870.
VI
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
Chemical Manager/Author
Monica A. Barren
Office of Solid Waste
U.S. Environmental Protection Agency
Washington, DC
Contract Authors
Lynne Haber
Toxicology Excellence for Risk Assessment
Cincinnati, OH
Andrew Maier
Toxicology Excellence for Risk Assessment
Cincinnati, OH
Jay Zhao
Toxicology Excellence for Risk Assessment
Cincinnati, OH
Michael Dourson
Toxicology Excellence for Risk Assessment
Cincinnati, OH
Reviewers
This document and summary information on IRIS have received peer review by both
EPA scientists and independent scientists external to EPA. Subsequent to external review and
incorporation of comments, this assessment underwent an Agency-wide review process whereby
the IRIS Program Manager achieved a consensus approval among the Office of Research and
Development; Office of Air and Radiation; Office of Prevention, Pesticides, and Toxic
Substances; Office of Solid Waste and Emergency Response (OSWER) of Water; Office of
Policy, Economics, and Innovation; Office of Children's Health Protection; Office of
Environmental Information; and the Regional Offices.
Internal EPA Reviewers
Dorothy Canter
OSWER
vn
-------
Nancy Chiu
Office of Water, Health and Ecological Criteria Division (HECD)
Annie Jarabek
National Center for Environmental Assessment
Edward Ohanian
Office of Water/HECD
Diana Wong
Office of Water/HECD
Robert MacPhail (reviewed portions)
National Health and Environmental Effects Research Laboratory
Selene Chou
Agency for Toxic Substances Disease Registry
External Peer Reviewers
Rolf Hartung
University of Michigan
Michele Medinsky
ToxCon
Anthony Scialli
Department of Obstetrics and Gynecology
Georgetown University Medical Center
Summaries of the external peer reviewers' comments and the disposition of their
recommendations are presented in Appendix A.
Vlll
-------
1. INTRODUCTION
This document presents background and justification for the hazard and dose-response
assessment summaries in U.S. Environmental Protection Agency's (EPA's) Integrated Risk
Information System (IRIS). IRIS summaries may include an oral reference dose (RfD), an
inhalation reference concentration (RfC), and a carcinogenicity assessment.
The RfD and RfC provide quantitative information for noncancer dose-response
assessments. The RfD is based on the assumption that thresholds exist for certain toxic effects
such as cellular necrosis but may not exist for other toxic effects such as some carcinogenic
responses. It is expressed in units of milligrams per kilogram per day (mg/kg-day). In general,
the RfD is an estimate (with uncertainty spanning perhaps an order of magnitude) of a daily
exposure to the human population (including sensitive subgroups) that is likely to be without an
appreciable risk of deleterious noncancer effects during a lifetime. The inhalation RfC is
analogous to the oral RfD, but it provides a continuous inhalation exposure estimate. The
inhalation RfC considers toxic effects for the respiratory system (portal of entry) and effects
peripheral to the respiratory system (extrarespiratory or systemic effects). It is generally
expressed in units of milligrams per cubic meter (mg/m3).
The carcinogenicity assessment provides information on the carcinogenic hazard
potential of the substance in question and quantitative estimates of risk from oral exposure and
inhalation exposure. The information includes a weight-of-evidence judgment of the likelihood
that the agent is a human carcinogen and the conditions under which the carcinogenic effects
may be expressed. Quantitative risk estimates are presented in three ways. The slope factor is
the result of application of a low-dose extrapolation procedure and is presented as the risk per
mg/kg/day. The unit risk is the quantitative estimate in terms of either risk per ug/L drinking
water or risk per ug/m3 air breathed. Another form in which risk is presented is a drinking water
or air concentration that provide cancer risks of 1 in 10,000 1 in 100,000 or 1 in 1,000,000.
Development of these hazard identification and dose-response assessments for phenol has
followed the general guidelines for risk assessment as set forth by the National Research Council
(1983). EPA guidelines that were used in the development of this assessment may include the
following: Guidelines for Carcinogen Risk Assessment (U.S. EPA,1986a), Guidelines for the
Health Risk Assessment of Chemical Mixtures (U.S. EPA, 1986b), Guidelines for Mutagenicity
Risk Assessment (U.S. EPA, 1986c), Guidelines for Developmental Toxicity Risk Assessment
1
-------
(U.S. EPA, 1991), Proposed Guidelines for Carcinogen Risk Assessment (U.S. EPA, 1996a),
Guidelines for Reproductive Toxicity Risk Assessment (U.S. EPA, 1996b), Guidelines for
Neurotoxicity Risk Assessment (U.S. EPA, 1998a); Recommendations for and Documentation of
Biological Values for Use in Risk Assessment (U.S. EPA, 1988); (proposed) Interim Policy for
Particle Size and Limit Concentration Issues in Inhalation Toxicity (U.S. EPA, 1994a); Methods
for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry
(U.S. EPA, 1994b); Peer Review and Peer Involvement at the U.S. Environmental Protection
Agency (U.S. EPA, 1994c); Use of the Benchmark Dose Approach in Health Risk Assessment
(U.S. EPA, 1995); Draft Revised Guidelines for Carcinogen Risk Assessment (U.S. EPA, 1999);
Science Policy Council Handbook: Peer Review (U.S. EPA, 1998b, 2000a); Science Policy
Council Handbook: Risk Characterization (U.S. EPA, 2000b).
The literature search strategy employed for this compound was based on the CASRN and
at least one common name. At a minimum, the following databases were searched: RTECS,
HSDB, TSCATS, CCRIS, GENETOX, EMIC, EMICBACK, DART, ETICBACK, TOXLINE,
CANCERLINE, MEDLINE, and MEDLINE backfiles. Any pertinent scientific information
submitted by the public to the IRIS Submission Desk was also considered in the development of
this document. The literature search was conducted in June 1999; selected key articles published
after that date are also included.
2. CHEMICAL AND PHYSICAL INFORMATION RELEVANT TO ASSESSMENTS
Phenol is a monosubstituted aromatic hydrocarbon. In its pure state, it exists as a
colorless or white solid. This pure compound is mixed with water and commercially sold as a
liquid product. Phenol gives off a sweet, acrid smell detectable to most people at 40 ppb in air
and at about 1-8 ppm in water (ATSDR, 1998). It evaporates more slowly than water and is
moderately soluble in water. Phenol is also combustible.
Phenol is produced through both natural and anthropogenic processes. It is naturally
occurring in some foods, in human and animal wastes, and in decomposing organic material, and it
is produced endogenously in the gut from the metabolism of aromatic amino acids. Phenol has been
isolated from coal tar, but it is now synthetically manufactured. Currently, the largest use of phenol
is as an intermediate in the production of phenolic resins, which are used in the plywood, adhesive,
2
-------
construction, automotive, and appliance industries. Phenol is also used in the production of
synthetic fibers such as nylon and for epoxy resin precursors such as bisphenol-A. Phenol is toxic to
bacteria and fungi, and it is used as a slimicide and disinfectant. Because of its anesthetic effects,
phenol is used in medicines such as ointments, ear and nose drops, cold sore lotions, throat lozenges
and sprays (such as those sold under the Cepastat® and Chloraseptic® labels), and antiseptic lotions.
The greatest potential source of exposure to phenol is in the occupational setting, where
phenol is used in manufacturing processes. People are also exposed via consumer products, such as
medicines and lotions, and some foods and tobacco smoke. Phenol has been found in drinking
water.
The physical and chemical properties of phenol are shown in Table 1.
-------
Table 1. Physical Properties and Chemical Identity of Phenol
CAS Registry Number
Synonym(s)
Registered trade name(s)
Melting point, °C
Boiling point, °C
Vapor pressure, at 25 °C
Density, at 20 °C relative to
the density of H2O at 4 °C
Flashpoint (open cup)
Water solubility, g/L at 25 °C
LogKoW
Odor threshold
Molecular weight
Conversion factors
Empirical formula
Chemical structure
108-95-2
Benzenol, hydroxybenzene,
monophenol, oxybenzene,
phenyl alcohol, phenyl
hydrate, phenyl hydroxide
Carbolic acid, phenic acid,
phenic alcohol
43
181.8
0.3513
1.0576
85 °C
87
1.46
0.047 ppm (0.1 8 mg/m3) -
100% response
0.006 ppm (0.02 mg/m3) -
sensitive
94.12
1 ppm (v/v) = mg/m3 x 0.260
1 mg/m3 = ppm (v/v) x 3.85
QH60
^OH
Lide, 1993
ATSDR, 1998
ATSDR, 1998
Lide, 1993
Lide, 1993
HDSB, 1996
Lide, 1993
ATSDR, 1998
Lide, 1993
HDSB, 1996
U.S. EPA, 1986d
Calculated
Calculated
Lide, 1993
-------
3. TOXICOKINETICS RELEVANT TO ASSESSMENTS
Phenol is readily absorbed by the inhalation, oral, and dermal routes. The portal-of-entry
metabolism for the inhalation and oral routes appears to be extensive and involves sulfate and
glucuronide conjugation and, to a lesser extent, oxidation. The primary oxidative metabolites
include hydroquinone and catechol, which are also substrates for conjugation. Secondary products
of hydroquinone or catechol metabolism, including benzoquinone and trihydroxybenzene, can also
be formed. Once absorbed, phenol is widely distributed in the body, although the levels in the lung,
liver, and kidney are often reported as being higher than in other tissues (on a per-gram-tissue basis).
Elimination from the body is rapid, primarily as sulfate and glucuronide conjugates in the urine,
regardless of the route of administration. Phenol does not appear to accumulate significantly in the
body.
3.1. ABSORPTION
Extensive absorption following inhalation exposure has been demonstrated in both human
and laboratory animal studies. Piotrowski (1971) studied lung and skin absorption of phenol in
volunteers (seven male and one female) exposed to air concentrations of 6 to 20 mg/m3 for 8 hours.
The study subjects were staff of the research institute in Poland, and all had undergone a previous
medical examination. In the lung absorption studies, the subjects inhaled phenol through a face
mask, eliminating the potential for most dermal absorption. These subjects retained 60-88% of the
inhaled phenol, and the percent retained did not vary with exposure concentration. The absorption
rate leveled off after approximately 3 hours of exposure, indicating that absorption had reached
steady state. In the skin absorption studies, subjects wore underwear and denim coveralls or were
unclothed for different trials of the experiment; in each case they were supplied with fresh air from
outside the chamber for breathing. The absorption coefficient did not appear to vary greatly with
exposure for 6 hours to concentrations in air ranging from about 5 to 25 mg/m3, and clothing did not
appear alter the absorption rate. The mean absorption coefficient was 0.35 m3/hr, indicating that the
amount of phenol present in 0.35 m3 of air was absorbed through the skin per hour. These data show
that dermal absorption can contribute significantly to the systemic dose of phenol following
exposure to phenol in air. However, the quantitative data from the dermal exposure study are limited
for the development of an RfC because of the short duration of the exposure and the absence of a
direct determination of whether the absorption rate had reached steady state.
-------
Other studies of workers exposed to phenol provide evidence for significant absorption via
the inhalation route; however, the contribution of dermal absorption from direct contact with liquid
phenol or from phenol in air was not assessed in these studies. Ohtsuji and Ikeda (1972) studied the
urinary free and conjugated phenol levels in Bakelite® factory workers. The total and conjugated
phenol levels tended to increase with increasing air concentration, but free phenol levels were not
affected. This suggests that at the concentrations studied (up to 12.5 mg/m3), phenol conjugation
was not rate limiting. The investigators did not specifically evaluate the levels of oxidative phenol
metabolites, so no conclusion can be made regarding whether Phase I metabolism was rate limiting.
On the basis of mass balance analysis, the authors concluded that phenol is efficiently absorbed from
the lung, because the phenol dose (air concentration * air volume/hr) was similar to the total mass
excreted in the urine.
Other occupational studies provide qualitative evidence for lung absorption, reporting
increasing urinary excretion of phenol metabolites with increasing workplace air concentrations. In
many cases, the data are not adequate to estimate the rate and degree of absorption through this
route, and potential contributions of dermal absorption are often inadequately described (Ogata et
al, 1986;ACGIH, 1991).
Absorption through the lung has also been evaluated in laboratory animals report following
inhalation exposure or intratracheal administration. In an unpublished study, Dow Chemical Co.
(1994) studied the kinetics of 14C-phenol in Fischer 344 (F344) rats following inhalation exposure
to 25 ppm (96 mg/m3) for 6 hours (nose only).1 Radioactivity in the blood was at steady-state levels
at the first measured time point (120 minutes after beginning the 6-hour exposure), indicating rapid
absorption kinetics. Hughes and Hall (1995) evaluated the disposition of phenol following
intratracheal and intravenous (i.v.) administration of 63.5 nmol of 14C-phenol to female Fisher 344
rats. The recovery of radioactivity in tissues and excreta for both routes was approximately 90% of
the administered dose within 72 hours. Because the amount of radioactivity recovered was nearly
equal for the intratracheal and the i.v. dose routes (and fecal excretion is minimal), the authors
concluded that absorption was near 100%.
Hogg et al. (1981) administered 14C-phenol intratracheally in isolated perfused rat lungs
from MRC hooded rats. At the end of the experiment (perfusions were approximately 85 minutes),
'This study has not been peer-reviewed, but it was well-conducted according to EPA guidelines
for a pharmacokinetics study (with minor deviations).
-------
approximately 92% of the administered radioactivity was in the perfusate, 6% was recovered in the
lung lavage, and approximately 3% was associated with lung tissue. The high recovery in the
perfusate indicated that phenol is nearly completely absorbed across the airways.
Human evidence for oral absorption indicates rapid and complete absorption. In a study of
three human volunteers, Capel et al. (1972) found that 85-98% of a 0.01 mg/kg oral dose of 14C-
phenol was excreted in the urine—primarily as phenylsulfate and phenylglucuronide—within 24
hours. In addition, case reports of oral poisoning provide qualitative evidence for gastrointestinal
absorption of phenol, but the ingested and absorbed dose were not estimated in these reports, and in
some cases both oral and dermal exposure were involved (Tanaka et al., 1998).
Numerous laboratory animal studies have found that orally administered phenol is readily
absorbed. In most cases, absorption rates were not calculated, but the rapid recovery of high
percentages of administered doses in the urine—with only minimal recovery in
feces—demonstrated nearly complete absorption. In various studies in the rat (Kao et al., 1979;
Edwards et al., 1986; Kenyon et al., 1995), the percentage of the administered dose recovered in
urine ranged from 65 to 96.5% over a wide range of doses. Varying degrees of absorption have
been observed in a variety of other species. In a study of 18 animal species orally administered
single doses ranging from 20 to 50 mg/kg 14C-phenol, Capel et al. (1972) found that the percentage
of radiolabel recovered in the urine within 24 hours varied from 31% of the administered dose in
squirrel monkeys to 95% in Wistar rats. It is not clear, however, if these differences represent
differences in the degree of absorption or in the rate of elimination.
Hughes and Hall (1995) found that in female F344 rats administered 63.5 nmol of 14C-
phenol by oral gavage, total recovery of radioactivity (in tissues and excreta) was approximately
90% of the administered dose within 72 hours. Because fecal excretion is approximately 1-3%,
and the recovered dose was nearly equal for the oral and the i.v. dose routes, the authors concluded
that the absorption was near 100%. The difference between the 90% recovery and 100% total
absorption was attributed to losses that were consistent across doses.
Rapid absorption of orally administered phenol has been observed in a number of studies.
Dow Chemical Co. (1994) studied the kinetics of 14C-phenol in F344 rats following oral dosing by
gavage or in drinking water. Total recovery of the administered radioactivity in the urine, feces,
tissues and carcass, and exhaled air was approximately 95%, regardless of the dosing protocol. In
the high-dose gavage animals (150 mg/kg), peak levels of radioactivity in blood were detected 15
7
-------
minutes post-administration, indicating rapid uptake kinetics. Humphrey et al. (1980) found that
free phenol levels in the plasma of rats given an oral dose of 300 mg/kg radiolabeled phenol
reached a maximum of 26 |a,g/mL at the first measured time, about 10 minutes after dosing, and
declined rapidly to background by 60 minutes. They observed similar results in beagle dogs given
a 40 mg/kg dose, with rapid peak levels of 7.8 |o,g/mL and complete removal of free phenol by 1
hour.
More quantitative kinetics data are available from in situ perfusion studies. Humphrey et al.
(1980) administered 14C- phenol (1 mg/mL) to the gut lumen of rats by means of a duodenal
cannula. The remaining radioactivity was measured at 3-minute intervals over 30 minutes in
perfusate collected by an ileal cannula. The results from the intestinal perfusion studies indicated
that removal of 14C-phenol obeys first-order kinetics, with a lumenal T1/2 of 5.5 minutes and a rate
constant for absorption of 0.127 min"1. These authors also measured the plasma concentrations of
phenol in the portal vein and posterior vena cava of dogs following intraduodenal dosing with
either 40 or 160 mg/kg phenol. At either dose, the concentration was already maximal in the portal
vein plasma within 3 minutes after dosing (the first measurement taken) and had decreased to
nondetectable levels within 1 hour at the low dose and to 33% at the high dose. These data show
that in both species phenol is rapidly absorbed from the gut.
Powell et al. (1974) added 14C-phenol to the mucosal medium of isolated rat gut
preparations and measured the level of radioactivity in the mucosal and serosal medium over 2
hours. They found that 78% of the administered radiolabel had been transferred to the serosal
medium over this period. Kao et al. (1979) administered 14C-phenol (12.5 or 25 mg/kg) to rats
intraduodenally. Recovery of the radioactivity was rapid, with more than 70% recovered in the
urine within 2 hours.
The dermal route of exposure is an important one. Both absorption of phenol liquid directly
in contact with skin and dermal absorption from exposure to phenol vapor are of concern.
Significant dermal absorption can result from phenol in air, so that phenol in air results in both
dermal and inhalation exposure (Piotrowski, 1971). On the basis of an analysis of the Piotrowski
(1971) data, ATSDR (1998) concluded that in air concentrations ranging from 5 to 25 mg/m3, the
amount of phenol absorbed through the skin will be about half of that absorbed through the lungs.
The conclusion was reached by estimating the amount of phenol absorbed through the lung as the
product of the human ventilation rate of 0.8 m3/hour and the steady-state lung retention fraction of
0.7 reported by Piotrowski (1971). The resulting lung absorption coefficient of 0.6 m3/hr is nearly
twice the skin absorption coefficient of 0.35 m3/hr. This analysis is limited, however, because it is
-------
not clear that the exposure duration was long enough for steady state to be reached in the dermal
absorption studies. Absorption via the dermal route may be lower at steady state due to the
potential for a back-pressure from phenol levels in blood.
A number of case reports and in vitro studies have led to quantitative estimates of phenol
absorption through the skin. Baranowska-Dutkiewicza (1981) applied a reservoir of 2.5, 5, or 10
g/L phenol solution on a small area of the forearm of 12 male volunteers. The absorption rate was
dependent on the concentration and ranged from 0.08 mg/cm2/hr at the low concentration to 0.301
mg/cm2/hr at the high concentration. At the low concentration, the total amount of phenol
absorbed—but not the absorption rate—increased with increased time; approximately 13% of the
applied dose was absorbed over a 30-minute period. In an in vitro study, 20% of applied doses
ranging from 1.3 to 2.7 |o,g/cm2 were absorbed from unoccluded human skin patches within 72
hours; addition of a Teflon cap resulted in 47% absorption over this same period (Hotchkiss et al.,
1992).
Bentur et al. (1998) reported an accidental dermal poisoning case in which a solution of
90% phenol was spilled on the left foot (3% of body surface). The exposure site remained
occluded, and no attempt at decontamination was made until the onset of symptoms, which began
within 4.5 hours. Following admission to the hospital shortly afterwards, peak serum phenol levels
of 21.6 jo-g/mL were measured. This study is presented here for completeness, but quantitative
exposure data from studies at lower phenol concentrations are more relevant to environmental
exposures.
The ability of phenol to be absorbed through the skin has also been evaluated in laboratory
animals. Hughes and Hall (1995) administered 63.5 nmol of labeled phenol to an occluded dermal
patch (2.5 cm2) of female F344 rats. Maximal recovery of the radioactivity was approximately
70%. The site of dermal application was washed 72 hours post-treatment and yielded 14% of the
recovered dose; 1.6% of the recovered dose was present in the skin at this site. Thus,
approximately 15% of the dose was not absorbed within 72 hours. In an in vitro study, Hotchkiss et
al. (1992) found that phenol absorption by rat skin is similar to that of human skin: approximately
20-50% in 72 hours, depending on the conditions.
Taken together, the human and laboratory animal data demonstrate that phenol is readily
absorbed following exposure by all dose routes. The recovery of greater than 90% of the
administered phenol dose as urinary metabolites provides direct evidence that the administered dose
was nearly completely absorbed. The route of administration appears to play a limited role, with
skin absorption reported as less extensive than absorption from the lung or gut. In most studies,
9
-------
absorption rate constants have not been calculated; however, the measurement of peak blood phenol
concentrations within minutes of dosing indicates that absorption is rapid
3.2. DISTRIBUTION
Studies in humans and laboratory animals indicate that phenol is widely distributed
throughout the body regardless of exposure route. Because phenol is rapidly excreted, studies on
tissue distribution typically evaluate only a small fraction of the absorbed dose.
Several fatal poisoning case studies evaluated phenol concentrations in multiple tissues
(Tanaka et al., 1998). Generally, phenol is widely distributed. Higher tissue concentrations relative
to blood have been reported for some organs, particularly for the liver and kidneys, although this
finding has not been reported consistently across all studies.
Morrison et al. (1991) reported on the kinetics of phenol injected intramuscularly in a motor
point block procedure in pediatric patients. Administered doses ranged from 6.7 to 70 mg/kg, and
the blood phenol concentration was measured at 5, 15, 30, 60, and 120 minutes after the last
injection. Phenol reached peak levels 5 to 15 minutes after administration and rapidly declined to 3
to 34% of peak levels within 120 minutes. Peak phenol concentration (|j,g/mL) in blood as a
function of administered dose (x, in mg/kg) was determined (y = 0.483x - 3.244; r = 0.873).
Pretreatment levels of blood phenol ranged from 0.3 to 0.8 |o,g/mL and post-treatment levels ranged
from 2.5 to 36 |o,g/mL.
The laboratory animal data provide additional evidence for elevated tissue concentrations in
the lung, liver, and kidney, although the magnitude of the tissue differences varies from study to
study. Liao and Oehme (1981) evaluated the tissue distribution of 207 mg/kg 14C-phenol orally
administered to male Sprague-Dawley rats. Total radioactivity in tissues declined rapidly from a
maximum of 28.4% of the administered dose at 0.5 hours to 16.6% at 1 hour and 0.3% at 16 hours.
Tissue concentrations of radioactivity measured at time points between 0.5 and 16 hours were
significantly greater than in plasma for the liver, spleen, kidney, and adrenal gland; tissue
concentrations in lungs and thyroid were also marginally elevated. The liver had the greatest
amount of radioactivity, accounting for 29-56% of the total radioactivity recovered from tissues at
the various time points. The study authors attributed the high levels in the liver to both an elevated
tissue concentration and the large relative organ size. Because the study measured total
radioactivity without further identification of the radio labeled compounds, it is not known whether
the observed radioactivity represented phenol or its metabolites.
10
-------
Dow Chemical Co. (1994) conducted a study of 14C-phenol administered to F344 rats by
oral gavage at 1.5, 15, or 150 mg/kg in drinking water at 5000 ppm or via nose-only inhalation at
25 ppm for 6 hours. Tissue levels of radioactivity were measured in the kidneys, liver, lung,
muscle, skin, spleen, testes, ovaries, and carcass 24 hours after exposure by the various routes. The
only sites with a statistically significant increase in radioactivity levels were the kidney and liver
(levels 5- to 10-fold higher than in other tissues). This finding was consistent across dosing
regimens.
Hughes and Hall (1995) evaluated the disposition of radiolabeled phenol administered
dermally, by oral gavage, intravenously, or intratracheally to female F344 rats. When the rats were
sacrificed 72 hours after administration by any of these four routes, tissue concentrations
represented only 1-5% of the recovered dose. No tissue appeared to have higher concentrations of
radiolabel following oral dosing, but the lung concentrations were markedly higher following
intratracheal administration. There was no substantive difference across tissues following dermal
dosing, although untreated skin had a slightly higher level. Marginal elevations in the liver and
kidneys were observed following i.v. dosing. The authors concluded that phenol is distributed
widely in tissues, with some accumulation in the large organs (lung, liver, and kidney, based on
within-route comparisons to the levels in blood).
Powell et al. (1974) treated juvenile rats (50 g) with less than 1 mg/kg 14C-phenol orally or
intraperitoneally. Whole-body radiograms indicated that the liver was not a site for accumulation
of the phenol; rather, it was widely distributed. It is not clear whether the difference between the
findings of this study and others is due to the differences in the sensitivity of the analysis or to
differences in dose levels. Thus, the data from animals studies at doses ranging from 1.5 to 207
mg/kg, which included several doses higher and lower than the chronic No Observed Adverse
Effect Level (NOAEL) of 60 mg/kg (NTP 1983a; Argus Research Laboratories, 1997) (see Chapter
5), showed that phenol is rapidly distributed to a wide range of tissues.
No direct studies of the placental transfer of phenol were identified. However, Ghantous
and Danielsson (1986) evaluated the placental transfer of benzene, of which phenol is a primary
metabolite. B6 mice were exposed for 10 minutes to benzene (at a target concentration of 2000
ppm) in air on gestation day (GD) 11, 14, or 17. The investigators conducted whole-body
radiography analysis and determined of tissue concentrations. Radioactivity was distributed to the
fetuses, but it was not specifically identified as phenol. The concentration of volatile and
nonvolatile radioactivity in the fetuses was, however, lower than that in maternal tissues.
11
-------
The human and laboratory animal data indicate that phenol is widely distributed in the body.
Although the human data are inconclusive, the laboratory animal data consistently indicate that
highly perfused organs such as the liver, kidney, and lung have higher tissue concentrations in
comparison to the blood concentration.
3.3. METABOLISM
Metabolic pathways for phenol are shown in Figure 1. Phenol is directly conjugated with
sulfate or glucuronic acid. Phenol that is not directly conjugated can also be a substrate for
oxidation reactions. The cytochrome P450 2E1 isozyme (CYP2E1) catalyzes the addition of one
oxygen atom to a variety of low-molecular-weight substrates such as benzene and chloroform, and
it is thought to be the primary P450 isozyme for phenol oxidation, although a minor role by other
cytochrome P450 enzymes cannot be discounted. The oxidation products of phenol generated by
CYP2E1 activity appear to be primarily hydroquinone and catechols, which can themselves
undergo further oxidation by CYP2E1 to trihydroxybenzene or by peroxidation to benzoquinone.
Alternatively, the hydroquinone or catechol metabolites can undergo conjugation reactions. In
addition to P450-mediated oxidation, some studies have suggested that peroxidative metabolism of
phenol can also take place, producing biphenols and diphenoquinones.
Direct sulfate and glucuronic acid conjugations are detoxifying mechanisms that represent
the bulk of phenol metabolism, as evidenced by the metabolic profiles observed in both humans
12
-------
Biphenol polymers
;
HO
Biphenol
Diphenoquinone
peroxidase
Glucuronide conjugate
UGT I ^J PST
CYP2E1 / Phenol \ CYP2E1
Sulfate conjugate
OH
Sulfate and glucuronide conjugates
PST
UGT
,OH
PST
-»• Sulfate and glucuronide conjugation
UGT
Catechol
OH
Hydroquinone
CYP2E1
OH
Sulfate and glucuronide
conjugates
PST
UGT
CYP2E1
OH
Glutathione
conjugate
Glutathione conjugate
OH
Trihydroxybenzene*
Figure 1. Metabolism of Phenol.
13
UGT= UDP-DEPENDENT
GLUCURONOSYL TRANSFERASE
PST= PHENOL SULFOTRANSFERASE
"INDICATES METABOLITES IDENTIFIED IN VITRO ONLY
-------
and laboratory animals. In humans and in most other species tested, sulfation predominates at
the lower doses. Capel et al. (1972) studied the urinary metabolites following oral
administration of 0.01 mg/kg 14C-phenol to three male volunteers. In these men, 85-98% of the
dose was excreted in 24 hours—69-90% as phenyl sulfate, 4-23% as phenyl glucuronide, and
trace amounts as hydroquinone conjugates. This high degree of conjugation indicates that, at
low doses, ingested phenol is nearly completely conjugated.
Laboratory animal studies have clearly shown that as the dose increases, the role of
glucuronidation increases, until at sufficiently high doses it becomes the predominant reaction.
The formation of oxidative metabolites (primarily by CYP2E1) also increases with increasing
dose. These dose-dependent changes have been best characterized in rat studies, which show
that at low doses sulfation predominates, with the glucuronidation beginning to predominate at
approximately 133 |^mol/kg (12.5 mg/kg) (Kao et al., 1979; Powell et al., 1974; Hogg et al.,
1981; Koster et al., 1981; Edwards et al., 1986; Meerman et al., 1987; Dow Chemical Co., 1994),
a dose that is below the chronic NOAEL of 60 mg/kg for phenol toxicity (NTP 1983a; Argus
Research Laboratories, 1997), as discussed later in detail. There is considerable interspecies
variation, however, in the contribution of the sulfation and glucuronidation pathways (Capel et
al., 1972;Mehtaetal, 1978).
The formation of oxidative metabolites increases at high doses. Dow Chemical Co.
(1994) reported that hydroquinone conjugates represented 3, 7.8, and 17.0% of the eluted
radioactivity in an analysis of urinary metabolites following single oral doses of 1.5, 15, and 150
mg/kg 14C-phenol, respectively. Legathe et al. (1994) administered an intraperitoneal (i.p.) dose
of 75 mg/kg to B6 mice and reported urinary metabolites as 34.5% phenyl sulfate, 28.5% phenyl
glucuronide, and 32.4% hydroquinone glucuronide, indicating substantial contribution of
oxidative metabolism at this high dose.
The formation of oxidative metabolites is thought to result primarily from reactions
catalyzed by CYP2E1. Koop et al. (1989), using hepatic microsomes prepared from male New
Zealand white rabbits, showed that CYP2E1 was the most active of six P450 isoforms tested.
Treatment of the lysates with an antibody to CYP2E1 inhibited hydroquinone formation by 68% and
89% in acetone-induced and uninduced microsomes, respectively. Snyder et al. (1993) studied
phenol metabolism in vitro in rat hepatic microsomal preparations. Addition of phenol to the
CYP2E1 microsome preparation yielded hydroquinone and, to a lesser degree, catechol metabolites.
Incubation of 14C-phenol and 3H-glutathione in the CYP2E1 microsome preparation yielded an
additional metabolite that cochromatographed with the compound formed from the reaction of
14
-------
benzoquinone with glutathione. The formation of the glutathione adduct was not dependent on
addition of glutathione-S-transferase. Lunte and Kissinger (1983) also reported the formation of
glutathione conjugates in microsomal (prepared from liver of male Swiss mice) metabolism of
phenol to hydroquinone. In addition to benzoquinone, hydroquinone and catechol can also be
oxidized to trihydroxybenzene (Sawahata andNeal, 1983).
Other in vitro studies using hepatic microsomes from rats treated with various P450 inducers
or inhibitors have also provided evidence for the importance of CYP2E1 in phenol metabolism
(Sawahata and Neal, 1983; Gilmour et al, 1986; Chapman et al, 1994; Kenyon et al, 1998).
CYP2E1 appears to predominate phenol oxidation. Contributions by other P450 enzymes cannot be
excluded, however, because only 68% of the induced hydroquinone formation was blocked by anti-
CYP2E1 antibody, and several inducers of other P450 enzymes (such as phenobarbital and arochlor)
enhanced phenol metabolism in these studies.
An alternative oxidative pathway involving peroxidation has been described for phenol.
Several investigators have used in vitro cell preparations with high peroxidase activity, such as
peritoneal macrophages or neutrophil preparations (Eastmond et al., 1986; Post et al., 1986;
Eastmond et al., 1987; Kalf et al., 1990), purified peroxidase enzymes (Smart and Zannoni, 1984;
Subrahmanyam and O'Brien, 1985), or cell lines that have high myeloperoxidase activity
(Kolachana et al., 1993), to show that phenol can be metabolized in these reactions. Metabolites
resulting from these reactions include 4,4'-biphenol and diphenoquinone. Although the peroxidation
of phenol has been demonstrated in vitro, no direct in vivo evidence for these peroxidative reactions
was identified.
The shift from sulfation to glucuronidation at increasing doses has been postulated to result
from depletion of sulfate pools (Kim et al., 1995). Alternatively, it has been suggested that the
difference between the K^ values for sulfate and glucuronide conjugation drives the conjugation
shift (Weitering et al., 1979). The effects of differing metabolizing enzyme activity across the zones
of the liver has also been suggested as an explanation for the metabolic profiles of phenol
(Medinsky et al., 1995). The functional units of the liver include lobules with blood supply
provided from the perimeter (periportal region) of the lobule though the portal vein and the hepatic
artery. The blood flows from the periphery of the lobule toward the terminal hepatic vein (also
called the central vein) at the center of the lobule through a series of differing metabolic regions or
zones. Both sulfotransferases and glucuronosyltransferases are present in periportal zone 1, with the
sulfotransferases predominating. Glucuronosyltransferases are present in zone 2, while both
15
-------
glucuronosyltransferases and monooxygenases such as CYP2E1 are present in pericentral zone 3.
According to the model, phenol entering the hepatic circulation would be metabolized first
in the periportal region, where sulfation predominates. Because the blood flows from the periportal
region to the pericentral region and then to the central vein and general circulation, little
unconjugated phenol is available for glucuronide conjugation or oxidation when it reaches the
pericentral regions of the liver. This model is consistent with the shift from sulfation to
glucuronidation at increasing doses. As the dose increases, more of the phenol reaches the
pericentral region unconjugated, and thus is available for glucuronidation. The model also explains
the increase in oxidative metabolites at high doses that exceed the conjugating capacity of zones 1
and 2 (Kenyon et al, 1995).
The model is also consistent with the observation that oral dosage with benzene results in
greater production of hydroquinone than that seen after oral dosage with phenol, even though
benzene is metabolized to hydroquinone via phenol. Benzene enters the liver in the periportal
region, is oxidized to phenol and other metabolites in the pericentral region, and then leaves the
liver via the hepatic vein. Because benzene must be oxidized before it is conjugated, more
unconjugated phenol would be released into the blood following benzene exposure than following
phenol exposure (Medinsky et al., 1995).
Direct evidence for this model was presented by Ballinger et al. (1995), who studied phenol
and metabolite transport kinetics in isolated perfused liver from rats, and by Hoffmann et al. (1999),
who conducted similar experiments in mice. The effects of enzyme distribution in the zones of the
liver were studied by contrasting phenol metabolite profiles resulting from antero- and retrograde
perfusions. It is noteworthy that the importance of the location of enzyme activities within the liver
would only be significant at oral phenol doses that were not conjugated at the portal of entry, and
thus were available for transport to the liver via the hepatic portal vein.
There is some evidence that the capacity for phenol conjugation varies with the portal of
entry. Cassidy and Houston (1984) conducted an elegant series of experiments in which they
injected rats intra-arterially, intravenously, or intraduodenally with phenol and then measured the
systemic availability of phenol. This allowed them to evaluate the first-pass metabolism by different
organ systems at doses ranging from 0.4 to 15 mg/kg. They were able to use this approach to
determine metabolism by the liver and gut. However, results on metabolism by the respiratory tract
from this study should be treated with caution, because environmental exposure results in exposure
of the epithelial respiratory tract (i.e., the portion exposed to the outside), whereas this study
16
-------
involved exposure of the endothelial respiratory tract (i.e., the portion exposed to the inside of the
body). Thus, any differences between the metabolic capacity of the endothelial and epithelial cells
would not be taken into account by the study design.
In this series of experiments, phenol that was systemically available had not been conjugated
or metabolized. The doses at which phenol became systemically available thus reflected the doses at
which the relevant metabolic enzyme systems became saturated. Metabolism became nearly
saturated between 4.5 and 15 mg/kg for the endothelial lung, and between 0.4 and 1.5 mg/kg in the
liver, and it was not saturated at the high dose for the gut. The affinity of metabolic pathways also
varied among the organ systems. The liver and gut, which removed 88% and 86% (respectively) of
phenol at the 0.4 mg/kg dose, demonstrated high affinity in comparism with the endothelial lung,
which removed 58% of the phenol at this same dose. Taken together, these data indicate that the gut
is a high-affinity and high-capacity site of metabolism, and the liver has high affinity but its capacity
is readily exceeded.
The data also suggest that the lung provides substantial metabolizing capacity but has lower
affinity than the gut and liver. Clear conclusions regarding the metabolic capacity of the lung
following exposure by the inhalation route are not possible because of the potential for differences
between the metabolic capacity of the epithelial and endothelial cells of the lung. However, one
would expect the potential for metabolism of inhaled phenol to be similar to that seen in this study,
because systemically absorbed phenol must pass through the endothelial cell layer.
The area under the blood concentration curve (AUC) for 14C-phenol was route- and dose-
dependent, reflecting the effects of portal-of-entry metabolism. In contrast, the AUC for phenol
metabolites did not differ by dosing route, indicating that phenol is extensively metabolized and the
effect of portal-of-entry metabolism is to reduce the amount of parent compound available for
metabolism by other organ systems.
Studies using isolated perfused rat liver were also conducted and correlated well with the in
vivo data. The percent of phenol removal from blood by first-pass metabolism declined from 73% at
a blood concentration of 2.8 |o,g/mL to 26% at 26 |o,g/mL, indicating extensive saturation at the
higher dose level.
Dow Chemical Co. (1994) studied the differential metabolism kinetics of phenol by differing
exposure routes. This study evaluated the kinetics of 14C-phenol in F344 rats following dosing
regimens that included single or eight daily oral gavage doses of 1.5, 15, or 150 mg/kg 5000 ppm in
17
-------
drinking water for 1 or 8 days or 25 ppm via inhalation for 6 hours (nose-only) for 1 or 8 days. The
authors estimated the doses resulting from the drinking water and inhalation exposures. For
drinking water administration (males only), doses were estimated by measurement of daily water
consumption. The administered dose was 291 mg/kg for the single-day protocol and 405 mg/kg for
the last day of the 8-day treatment, based on the water intake and the weight of each animal; thus,
the drinking water doses were higher than the oral gavage doses.
The doses used in the drinking water study and the high dose in the gavage study were all
higher than the chronic NOAEL of 60 mg/kg (NTP 1983a; Argus Research Laboratories, 1997), as
discussed in detail in Section 5. In contrast, the inhalation doses were estimated as 11.5 and 17.8
mg/kg for males and females, respectively, following a single exposure period, and the dose was
21.4 mg/kg (males only) on the last day of the 8-day exposure protocol. Thus, the inhalation route
more closely resembled the middle gavage dose level, and the absorbed dose via inhalation was
lower than the chronic NOAEL (NTP 1983a; Argus Research Laboratories, 1997).
Metabolic profiles revealed ratios of 0.61 for glucuronide/sulfate conjugates in urine at the
two lower gavage doses and were similar following inhalation (0.24-0.39). The ratio at the high
gavage dose was 1.16, and it was similar following drinking water exposure (1.43 and 1.87 for the
single and 8 day exposures). The observed formation of oxidative products, as shown by urine
levels of hydroquinone glucuronide, was also dependent on total dose. The formation of oxidative
metabolites following inhalation paralleled the low-dose gavage data, whereas the drinking water
levels paralleled the high-dose gavage levels. The pattern of phenol metabolism correlated with the
magnitude of the absorbed dose and did not appear to be dependent on the route of administration.
Metabolism of phenol appears extensive in the lung, liver, and gastrointestinal tract; however
limited data are available for other organs. Metabolism appears to be extensive in the kidney
(Tremaine et al., 1984). No data were identified that addressed portal-of-entry metabolism for the
skin.
One consequence of the portal-of-entry metabolism of phenol is that phenol serum levels are
not necessarily linear with dose or exposure levels. At low doses, almost all of the absorbed phenol
is conjugated and excreted, without entering the bloodstream. At higher doses, free phenol and its
metabolites appear in the blood and increase with dose. This nonlinearity of blood phenol levels
with dose is illustrated by the data of Dow Chemical Co. (1994). Peak phenol blood concentrations
in rats following an oral bolus dose of 150 mg/kg were 2320-fold higher than the peak blood
concentrations following an oral bolus dose of 1.5 mg/kg.
18
-------
The role of peak levels may be significant for induction of at least some aspects of systemic
toxicity. Dow Chemical Co. (1994) sheds some light on the relationship between metabolism and
toxicity. The high-dose gavage group in this study developed a cluster of behaviors that the authors
termed "phenol twitching behavior (PTW)" that included tremors, sudden jerks, hyper-reactivity to
stimuli, and excessive blinking. PTW began almost immediately after dosing and disappeared by 37
minutes post-dosing. Blood phenol levels also peaked almost immediately after dosing, and PTW
was not apparent at blood phenol concentrations below approximately 3 |o,g/mL. PTW was not
observed at the lower gavage doses or following inhalation exposure; peak blood phenol levels in
these groups were well below 1 |o,g/mL.
Interestingly, PTW was also not observed in the drinking water exposure groups, even
though the total dose in these groups was higher than the high gavage dose, and the drinking water
doses had a similar metabolic profile to the high gavage dose. Unfortunately, blood phenol levels
were not sampled in the drinking water groups, so the peak blood phenol level is not available.
However, given the rapid clearance of phenol from the blood, it is likely that the peak blood level
was much lower in the drinking water group than in the high-dose gavage group. This suggests that
PTW is more closely related to peak phenol blood levels than to a measure of total dose, such as
AUC. Because phenol metabolite levels paralleled those of phenol, these data cannot be used to
distinguish between phenol and its metabolites being the toxic agent. These data do not identify the
appropriate dose metric (e.g., peak concentration vs AUC) for other toxic endpoints.
One indication that the oxidative metabolites are important determinates of toxicity is based
on experiments by Chapman et al. (1994). They studied the dysmorphogenic and embryotoxic
effects of benzene and its metabolites to the whole rat conceptus in vitro. Phenol at 1.6 mM elicited
only minor effects, but inclusion of S9 microsomal fractions greatly increased the potency of phenol,
with significant effects observed at doses as low as 0.01 mM. Metabolite analysis indicated that
hydroquinone and catechol were the primary metabolites. When evaluated singly, hydroquinone,
catechol, and benzoquinone induced similar embryotoxicity, producing 100% lethality at 0.1 mM.
The addition of phenol and hydroquinone together induced a more-than-additive embryotoxicity,
which the authors suggested as evidence for a peroxidative mechanism for phenol bioactivation,
based on the potential for electron cycling between phenol and hydroquinone.
Intraspecies variability has also been studied. Campbell et al. (1987) isolated human liver
sulfotransferases, the enzymes responsible for the conjugation of phenol with sulfate, and
analyzed their apparent activities toward/>-nitrophenol (as a model compound for simple phenols).
19
-------
The average phenol sulfotransferase (PST) activity measured in liver samples of 20 patients (13
male, 7 female) was 35.8±10.6 standard error of the mean (SEM) units/mg protein. No correlation
between enzyme activity and patient age or gender was found, although the power to detect any such
correlation was not noted. Seaton et al. (1995) studied the kinetics of phenol sulfation and
hydroquinone conjugation, both of which varied over a range of approximately three-fold in a
sample of liver fractions from 10 humans. Using lysates from a single human liver, saturation of
phenol sulfation was apparent above 800 |oM; the observed kinetics were consistent with two
contributing enzymes, PST1 and PST2. The expression of two distinct PST enzymes has also been
demonstrated in human nasal epithelium (Beckmann et al., 1995).
Kawamoto et al. (1996) studied the effect of various lifestyle factors and of genetic
polymorphisms in five metabolizing enzymes, including aldehyde dehydrogenase (ALDH2), N-
acetyl transferase (NAT2), cytochrome P450 1A1 (CYP1A1), CYP2E1, and glutathione-5*-
transferase mu (GSTM1) on urinary levels of phenol in a cohort of men who were not
occupationally exposured. Step-wise multiple regression analysis was performed to identify
important determinates of urinary phenol levels. On the basis of this analysis, there was no
relationship between polymorphisms (including for CYP2E1) and background urinary phenol levels.
In the total sample (n = 351), the geometric mean urinary phenol level was 7.64 mg/L and the
geometric standard deviation was 2.9. No data are available, however, on how genetic
polymorphisms affect the levels of metabolites produced from exogenously dosed phenol.
The changes in enzyme activity or expression of genes that encode enzymes important for
phenol metabolism with age have been studied. The status of CYP2E1 in fetuses remains unclear,
with conflicting results reported. Most of the existing studies indicate that this enzyme is expressed
in human adults but not in human fetuses, even when measured using sensitive assays (reviewed in
Hakkola et al., 1998). However, at least two studies (Carpenter et al., 1996; Vieira et al., 1996)
indicate that CYP2E1 is expressed at least to some degree in fetal liver. Vieira et al. found that
CYP2E1 protein could not be detected immunochemically in fetal human liver, and there was only
minimal evidence of CYP2E1 mRNA or CYP2E1 activity in fetal liver microsomes. (The difference
in assay results may be due to differences in sensitivity or to cross-reaction of CYP1A1 activity.)
However, the authors found, that CYP2E1 protein levels rise rapidly in the first few hours after
birth, with a slow increase in protein levels and in CYP2E1 mRNA levels during childhood.
Results of animal studies of developmental CYP2E1 regulation are consistent with the
human data in providing uniform evidence of the rapid induction of this gene soon after birth (Song
20
-------
et al., 1986; Umeno et al., 1988; Schenkman et al, 1989; Ueno and Gonzalez, 1990). Thus, overall,
the data show that if CYP2E1 activity exists in human fetuses, levels are much lower than those in
adults. Regardless of fetal CYP2E1 expression, the enzyme is rapidly induced upon birth. For this
reason, children would be expected to be capable of phenol metabolism, although the amount of
CYP2E1 may be less than that present in adults.
Age-dependent changes in phase II conjugation have also been evaluated. In an evaluation
of how PST activity varies with age in rats, Iwasaki et al. (1993) studied B-naphthol metabolism by
PST in fetal rat liver, in the liver of 2-, 9-, 17-, and 25-day-old neonates, and in adult rats. Activity
was analyzed in the livers of both sexes. The fetal liver had little conjugating ability, but this
activity developed rapidly after birth. However, activity was substantially lower in neonates of all
ages evaluated when compared with adult levels. Heaton and Renwick (1991) administered i.p.
doses of 25 mg/kg 14C-phenol to rats varying in age from 3 to 16 weeks and measured metabolites in
urine collected in 24 hours. The percentage of the administered dose recovered in the urine in 24
hours ranged from 61-90% in males and 63-99% in females, with increasing recovery with age.
Importantly, the formation of hydroquinone conjugates was greater in the younger animals.
In males, 38% of the administered dose was recovered as hydroquinone conjugates in the 3-week-
old animals; 8.2% of the urinary metabolites was recovered in 16-week-old rats. In females, 17.8%
of the administered dose was recovered as hydroquinone conjugates in 4-week-old rats and 10.5%
was recovered in 15-week-old rats. Taken together, the evidence indicates that both sulfate
conjugation and P450 metabolism are lower early in life and increase as adulthood is reached.
However, even in the face of limited P450 activity, significant formation of oxidation products can
occur because of limited sulfation capacity. The oxidative products become substrates for
glucuronidation, and this does not appear to be limited in the young.
Phenol metabolism may also be gender dependent, although the data are less substantial than
those for differences due to age. Iwasaki et al. (1986) reported that PST activity was similar in both
sexes up to 3 weeks of age and was higher in males than in females in 7-week-old rats. Activity in
2-year-old rats of both sexes was similar and fell between the levels for males and females at 7
weeks to 1 year. Kenyon et al. (1995) administered 14C-phenol to B6 mice of both sexes and
observed that, males excreted a greater proportion of HQ-glucuronide than did females at all doses;
the difference was roughly twofold at a dose of 40 |^mol/kg. These results are consistent with the
greater degree of hydroquinone conjugates excreted in the urine of male versus female rats reported
by Heaton and Renwick (1991). Sex-based differences in metabolism have also been reported in
21
-------
rats (Meerman et al., 1987), with slightly lower total recovered radioactivity in the urine of females
versus males (i.e., more rapid metabolism in males). However, the magnitude of this difference
(91.2% vs 87.3%) was limited.
Interspecies differences in phenol metabolism have also been evaluated. Seaton et al. (1995)
found that the rates of both phenol sulfation and hydroquinone conjugation in mouse and rat liver
were comparable to those of human liver preparations. Schlosser et al. (1993) reported that mouse
liver microsomes metabolized approximately twice as much phenol as did rat liver microsomes,
although the relative proportions of metabolites were roughly similar.
Phenol is formed endogenously in the gut by bacterial metabolism of aromatic amino acids in
protein. The amount formed is related to the amount of protein ingested, but the amount in humans
typically varies from 1 to 10 mg/day, corresponding to approximately 0.014 to 0.14 mg/kg-day
(Bone et al., 1976; Lawrie and Renwick, 1987; Renwick et al., 1988).
A physiologically based pharmacokinetic (PBPK) model for the distribution of benzene and
metabolites was developed by Bois et al. (1991). The model was developed to predict phenol and
metabolite distributions to fat, well-perfused tissue, poorly-perfused tissue, bone marrow, liver, lung,
and gut using Monte Carlo simulations of 64 parameters. The model was not validated using
empirical data. The Bois et al. model consistently predicted that phenol administration would
produce higher levels of phenol and hydroquinone in the blood than seen following benzene
administration.
The first phase in the development of a model of the in vitro kinetics of phenol and benzene
biotransformation by liver microsomes was described by Schlosser et al. (1993) and enhanced by
Medinsky et al. (1995). The model described the following reaction sequences: benzene > phenol >
catechol > trihydroxybenzene and phenol > hydroquinone > trihydroxybenzene. All reaction steps
were assumed to be catalyzed by cytochrome P450 2E1, and benzene, phenol, catechol, and
hydroquinone were all assumed to compete through reversible binding for the same reaction site on
cytochrome P450. Parameters were identified that were successful at predicting the concentration
with time of all five chemicals in incubations with rat or mouse liver microsomes (Schlosser et al.,
1993). The observation of a lag time in the production of hydroquinone from benzene—in
comparison to the rate of production of hydroquinone from phenol—supported the assumption that
all of the substrates compete for the same enzyme reaction site.
Medinsky et al. (1995) extended the data into a conceptual model of the differences between
phenol and benzene metabolism. The goals of the conceptual model included explaining the
22
-------
observed differences between the carcinogenicity and genotoxicity of phenol and benzene and
explaining why urinary hydroquinone levels are higher after benzene dosing than after phenol
dosing. The latter observation would appear to be inconsistent with the prediction of the Bois et al.
(1991) model that blood hydroquinone levels are higher following phenol dosing than following
benzene dosing. As described earlier in this section, differences between benzene and phenol
toxicity were attributed to zonal differences in the distribution of hepatic metabolic enzymes.
In summary, phenol is an endogenous metabolite that undergoes further metabolism
efficiently. At low doses the bulk of the phenol appears to be conjugated with sulfate or glucuronide
at the portal of entry. As the dose increases, the sulfation pathway becomes saturated, and the
relative contribution of glucuronidation and oxidation reactions increases. Saturation of first-pass
metabolism may be important for producing peak levels of phenol that correlate with acute systemic
toxicity. In addition, saturation of conjugation, which leads to increases in oxidative metabolism,
may also be an important determinate of toxicity. The data on intraspecies variability are limited,
but they do not indicate great variation in metabolic capacity in humans. In rodents, males and
younger animals appear to rely more heavily than females and adult animals on oxidative
metabolism, respectively, but the differences are no more than twofold. The metabolism of phenol
in humans and rodents appears to be similar, although some evidence suggests that mice metabolize
phenol more rapidly than do humans or rats.
3.4. EXCRETION
The existing human and laboratory animal studies consistently report that phenol is rapidly
excreted, with little tendency for accumulation. Elimination is primarily in the urine in both humans
and laboratory animals, with only a minor contribution of elimination in the bile. Ohtsuji and Ikeda
(1972) studied the urinary free and conjugated phenol levels in Bakelite® factory workers. Workers
were exposed to phenol vapor by inhalation on a daily basis. The workers were also possibly
exposed by the dermal route, but the contribution of this route to the total exposure was not directly
measured. Analysis of urinary phenol levels at different times during the work shift and across work
shifts indicated that in workers exposed to 7.8 to 9.6 mg/m3, the urinary levels increased
significantly from the beginning of the work shift to the end of the work shift, but they did not tend
to accumulate across the work shifts. A slight increase in the morning sample on the sixth
consecutive work day was observed, but after two days off, pre-shift samples were no longer
elevated.
23
-------
Rapid clearance from the blood in humans has also been observed. Bentur et al. (1998)
presented a case report from a dermal poisoning in which a solution of 90% phenol was spilled on
the left foot (3% of body surface). Even at this high exposure level, clearance from the blood was
rapid, with blood levels decreasing from 21.6 to 2.8 |o,g/mL in the first 12 hours. The authors
estimated that the half-life elimination was 13.86 hours, but they did not include the initial rapid
decline in serum concentration that was apparent over the first 12 hours post-admission. Both
absorption and clearance would be expected to be more rapid at lower exposure levels, as high
exposure levels can lead to protein denaturation and saturation kinetics.
Laboratory animal studies have consistently found that phenol is rapidly excreted. Clearance
of phenol from the blood is rapid. Rats given an oral dose of 300 mg/kg, a level comparable to the
rodent median lethal dose (LD50)of 340 mg/kg (Deichmann and Witherup, 1944), had maximum
blood concentrations of 26 |o,g/mL at the first measured time point (about 10 minutes), and blood
levels declined rapidly to background by 60 minutes (Humphrey et al., 1980). Similar results were
observed by the same authors in dogs given a 40 mg/kg dose, with rapid peak levels (7.8 |o,g/mL)
and complete removal of free phenol by 1 hour. A half-life of 12 minutes in blood was reported for
rats administered 150 mg/kg by gavage (Dow Chemical Co., 1994). Legathe et al. (1994) reported
biphasic elimination kinetics from the blood, with a terminal half-life of 22 minutes. Similarly,
Cassidy and Houston (1984) reported biphasic kinetics with a half-life of approximately 5 minutes
following intra-arterial administration.
The elimination kinetics in multiple tissues was studied by Liao and Oehme (1981). Total
radioactivity in tissues was maximal within 30 minutes of dosing, representing 28.4% of the
administered dose. Tissue levels accounted for 16.6% of the administered dose at 2 hours and 0.3%
at 16 hours. Although maximum levels varied considerably across tissues, the rate of elimination
did not appear to differ with tissue type. Numerous laboratory animal studies indicate that urinary
elimination of sulfate and glucuronide conjugates accounts for most of the excretion, ranging from
70 to 90% of the administered dose within 24 hours, whereas excretion in feces represents only a
small fraction of the administered dose, approximately 1-3% (Edwards et al., 1986; Meerman et al.,
1987; Dow Chemical Co., 1994; Hughes and Hall, 1995).
4. HAZARD IDENTIFICATION
24
-------
4.1. STUDIES IN HUMANS—EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS
The epidemiology data on phenol are limited. Epidemiology studies have typically included
confounding exposures and have not adequately adjust for smoking. Kauppinen et al. (1986)
reported a significant increase in respiratory cancer in phenol-exposed workers, but this observation
appears to be due to confounding exposures, because there was no dose-response, and the effect
decreased after accounting for latency. No effect on cancer mortality was observed in workers
exposed to phenol in the rubber industry (Wilcosky et al., 1984) or in workers exposed to
formaldehyde and phenol (Dosemeci et al., 1991). An occupational study (Shamy et al., 1994) and
case studies (e.g., Merliss, 1972) have reported liver effects following exposure to phenol. Immune
effects were also been reported in an occupational study of workers exposed to phenol as part of a
mixture of solvents (Baj et al., 1994). Studies of populations whose drinking water was
contaminated with phenol found elevated incidences of diarrhea, nausea, mouth sores, and dark
urine (Jarvis et al., 1985; Baker et al., 1978).
4.1.1. Oral
Estimated lethal oral doses of phenol in adults vary widely, from 1 g (14 mg/kg, assuming an
adult body weight of 70 kg) to as much as 65 g (930 mg/kg) (Deichmann and Klepinger, 1981). In
another report (Bruce et al., 1987), the minimum lethal oral dose in adults was estimated as 140
mg/kg.
Jarvis et al. (1985) reported on illness associated with consuming water contaminated with
phenol from a spill into the river that served as the drinking water source. A retrospective mail
survey was sent to a total of 594 English households: 250 that were highly exposed, 94 that were
exposed at low concentrations (from a reservoir that diluted the contaminated river water), and 250
that were unexposed (selected from a telephone book). On the basis of data from the water
authority, the estimated phenol concentrations in drinking water in the low-exposure area (0.05
mol/L, equivalent to 4.7 |J.g/L) was roughly half that in the high-exposure area (0.11 |^mol/L,
equivalent to 10 |J.g/L) for the first 24 hours. The next day, the phenol concentration for both groups
was 0.05 |omol/L, and the concentration was < 0.01 |^mol/L (< 0.9 |J.g/L) by the third day after the
contamination incident.
Chlorination of the water resulted in production of chlorophenols. The chlorophenol
concentration followed a similar pattern, but the chlorophenol concentrations, which ranged from
25
-------
0.43 to 0.2 |^mol/L at the first measurement (84.7-39.4 |o.g/L, assuming all chlorophenols were in
the form of trichlorophenols), were higher than those for phenol. There were no data on phenol
concentrations in the unexposed area, but an unspecified local press report implicated a possibility
of phenol contamination. Due to the similarity of the two exposed areas in the measured
concentrations of phenol and chlorophenol, these two areas were combined in the data analysis.
The percentage of responding households was similar in all of the groups and ranged
from 69 to 77%, resulting in 172 households (448 people) in the unexposed area and 254
households (754 people) in the exposed area being evaluated. The two groups had similar
distributions in sex, age, and usual water usage. Compared with the unexposed individuals,
those in the exposed area had significantly higher incidences of gastrointestinal illnesses, such
as diarrhea, nausea, vomiting, and abdominal pain. Other symptoms, such as headache, rash,
and malaise, were also observed at a significantly elevated incidence in the exposed group. The
day of onset of symptoms corresponded with the period of elevated phenol concentrations in the
contaminated drinking water. The associations were stronger among those who reported that
they drank the water than among those in the exposed area who reported not drinking the water.
(Others may have consumed the water in cooking.) In another analysis, gastrointestinal
symptoms did not significantly correlate with whether the water tasted bad. However, because
of the confounding exposure of chlorophenol in the water, the usefulness of the result for risk
assessment is limited.
Baker et al. (1978) reported on phenol poisoning in humans due to an accidental
contamination of their drinking water on July 16, 1974. A train derailment resulted in a
spillage of 37,900 L of pure phenol onto the ground, and the spillage caused contamination of
drinking water in wells of nearby houses. Two wells near the spill were tested initially on
July 23 and were found to have phenol concentrations of 0.21 and 3.2 mg/L. Further testing
in late July and August of the six wells nearest the spill found peak concentrations between 15
and 126 mg/L. Within approximately 2 months after the spill, "most families" began to obtain
water from other sources (from neighbors or bottled water). Phenol concentrations in well
water as high as 1130 mg/L were reported over the next 6 months, with the higher levels
observed after flushing of the spill site (unspecified number of wells tested). The authors
investigated the health effects in three groups of people. Group 1 (n = 39) consisted of all
families living 120-310 m from the spill site and having at least one water test greater than
0.1 mg phenol/L (at least once between July and February). Group 2 (neighborhood control)
(n = 61) consisted of all families adjacent to Group 1 (210-670 meters from the spill) whose
wells had phenol concentrations of between 0.1 and 0.0001 mg/L. Group 3 (distant control)
26
-------
(n = 58) lived at least 1.9 km from the spill and had no phenol in its wells.
Group 1 reported significantly more diarrhea, mouth sores, burning mouth, and dark urine
than the combined control groups. About 44% of the individuals in Group 1 reported at least two
of these responses in the 7 months following the accident and were considered "affected
individuals"; only 8% and 3% of Group 2 or 3 subjects, respectively, had the same responses.
Responses in Group 1 were primarily restricted to the first 2 months of exposure, before the use of
bottled water began. Responses in the other two groups tended to occur throughout the 8-month
period (July through February). Other than the four reported symptoms, no abnormal observations
in physical examinations and serum biochemical evaluations were evident in Group 1 as compared
with controls when monitoring was done in February.
On the basis of water testing data and water preference histories, the authors estimated that
the daily oral dose of phenol for the 17 affected individuals in Group 1 was between 10 and 240
mg. However, this range may overestimate the amount of phenol ingested because phenol's
unpleasant odor might have discouraged ingestion of water with concentrations above 0.1 mg/L.
In contrast, this range does not include phenol that may have been absorbed during skin contact
with contaminated water. It was also not clear whether the subjects continued to shower with the
contaminated water after switching their drinking water source. Based on a default adult body
weight of 70 kg, this daily oral dose corresponds to 0.14 to 3.4 mg phenol/kg-day. Thus, there is a
considerable range in the estimated phenol dose associated with symptoms. In addition, because
"most" (but apparently not all) families switched to other water sources within the first 2 months
of exposure, the exposure duration for the affected individuals is not known. Therefore, it is
difficult to use these data for quantitative analysis, although they might be useful for placing
bounding estimates on the risk values for systemic effects estimated from laboratory animal
studies.
4.1.2. Inhalation
Kauppinen et al. (1986) reported a case-control study on respiratory cancers and chemical
exposures in the wood industry. A cohort of 3805 Finnish men who worked in the particle board,
plywood, sawmill, or formaldehyde glue industries for at least 1 year between 1944 and 1965 was
followed until 1981. From the cohort, 60 cases of respiratory malignant tumors were identified.
The tissue locations of these tumors included tongue (1), pharynx (1), larynx or epiglottis (4), and
lung or trachea (54). No cases with tumor in the mouth, nose, or sinuses were identified. Among
the 60 cases, 2 were rejected due to a false preliminary diagnosis of cancer and 1 was rejected as
27
-------
chronic lymphocytic leukemia. The final size of the group of cases was thus 57. The control
group contained three subjects for each case, selected from the cohort and matched by birth year, for
a total size of 171.
The job exposure was estimated from the industrial hygiene data of the plant, general
hygiene data on exposures, and information on ventilation, work procedures, and other relevant
factors at the plants. However, the authors, gave no information on direct phenol measurements;
thus, the quality of the estimated exposure levels could not be evaluated. The work histories of the
subjects were assessed primarily from plant registers and supplemented with personal interviews.
Individual phenol exposures were determined qualitatively as "yes" or "no" and as a function of
exposure time.
Smoking histories were determined by a mail survey that resulted in smoking information on
39 of 57 cases (68%) and on 130 of 171 controls (76%). Because there were few nonsmokers and
information on the amount smoked was not as complete as on years of smoking, the subjects were
compared only as light or heavy smokers, based on whether their years of smoking exceeded 35.
Phenol exposure resulted in a statistically significant odds ratio (OR) of 3.98 or 4.94 for
respiratory tumors with or without the adjustment for smoking years, respectively. When the
duration of phenol exposure was considered, both exposures < 5 years and > 5 years resulted in a
statistically significant OR < of 5.86 or 4.03, respectively (i.e., no duration response). When a
provision for a 10-year latency was introduced (excluding exposure during the 10 years immediately
preceding the diagnosis of cases), phenol exposure resulted in a nonsignificant OR of 2.86 adjusted
for smoking years but a significant or of 3.98 without smoking adjustment. Of the 39 cases for
which smoking information was available, 12 had been exposed to phenol (9 to phenol in wood
dust), and 7 had been exposed to phenol with a 10-year latency (4 to phenol in wood dust). Because
the OR did not increase with duration of phenol exposure and the provision for the 10-year latency
period resulted in lower values of ORs, a confounding factor may have been responsible for the
observed statistically significant ORs.
One of the confounding factors could have been concurrent exposure to multiple pesticides,
which was in the same study shown to increase the OR for respiratory tumors. An exclusion of
workers exposed to both phenol and pesticides resulted in a change of the OR from a significant 4.9
to a nonsignificant 2.6. Thus, a confounding effect due to exposures to pesticides was very possible.
Considering the location of the tumors, formaldehyde exposure was also a likely confounder.
Generally similar results were observed in this study for workers exposed to phenol in wood
28
-------
dust. Exposure to phenol in wood dust resulted in a statistically significant OR with or without
adjustment for smoking. For the workers exposed to phenol but not wood dust, provision for a
latency period eliminated the observed statistically significant OR. Among the workers exposed to
phenol in wood dust, however, the OR did increase with exposure duration and was statistically
significant in those exposed > 5 years (OR of 4.77) but not in those exposed for < 5 years (OR of
3.84). On the basis of these results, the phenol-exposed workers had an elevated risk of respiratory
cancer, but phenol itself did not appear to be the causative agent; rather, it appears that there was a
confounding exposure.
Wilcosky et al. (1984) reported a case-control study of cancer mortality and solvent
exposures in the rubber industry. From a cohort of 6678 active and retired male rubber workers of a
large plant in Ohio, 183 decedents from stomach cancer, prostate cancer, lymphosarcoma and
reticulum cell sarcoma, lymphatic leukemia, and respiratory cancer were selected as cases. As a
control, 20% of an age-stratified random sample of the cohort (calculated as 1336 subjects) was
selected. Including phenol, a total of 25 solvents were authorized to be used in the plant. The
exposure to any particular solvent was determined from the records of annual authorization for use
of these solvents in each work area. Only workers who had cumulative exposures of more than 1
year were considered exposed.
On the basis of the analysis of the age-adjusted exposure ORs, no association was seen
between phenol exposure and mortality from stomach cancer, prostate cancer, lymphosarcoma and
reticulum cell sarcoma, lymphatic leukemia, or respiratory cancer. However, this study had several
major limitations. One limitation was that the estimation of whether workers were exposed to a
solvent was based solely on authorization and not on actual usage, which would tend to lead to an
overestimation of exposure. In addition, the analysis was based solely on a qualitative evaluation of
whether a given solvent was used; no estimates of exposure were made, and so no exposure-
response assessment was conducted. Although smoking can confound evaluation of cancer risk, this
factor was not investigated. Finally, it was common for workers to be simultaneously exposed to
multiple solvents; therefore, solvents other than phenol may have affected the study outcome. In
this study, phenol exposure was not associated with a risk of several cancers, but this lack of an
association cannot be considered definitive because of the study limitations mentioned above.
In an occupational epidemiology study, Dosemeci et al. (1991) evaluated mortality among
14,861 white male workers in five companies that used formaldehyde and phenol. Unfortunately,
the phenol exposure was confounded by co-exposure to other compounds, such as formaldehyde,
29
-------
asbestos, urea, melamine, hexamethylenediamine, wood dust, plasticizers, carbon black, ammonia,
and antioxidants. On the basis of phenol concentrations obtained from historical monitoring and
industrial hygiene surveys, the investigators assigned each job/department/year combination to
groups with no, low, medium, or high phenol exposure and then calculated cumulative exposure.
Compared with the entire U.S. population, the entire cohort, had no significant increases in
standardized mortality ratios (SMRs) for all causes of death or any diseases. The phenol-exposed
workers as a group had slightly elevated SMRs for cancers of the esophagus (1.6), rectum (1.4),
kidney (1.3), and Hodgkin's disease (1.7); however, none of these increases were statistically
significant when compared with those in general population. In addition, an analysis of mortality by
level of cumulative exposure showed that none of these increases had dose-response relationships
with exposure to phenol. The only significant observations were decreases of SMRs for infective
and parasitic diseases and for accidents in the entire cohort and exposed workers. These
observations were attributed to the healthy worker effect. This study provided no evidence of
phenol-induced morbidity, mortality, or carcinogenicity.
Baj et al. (1994) reported an epidemiology study of 22 Polish office workers (18 females and
4 males) exposed to Ksylamit® vapor for 6 months and 27 age- and sex-matched healthy volunteers
from the same town. The exact composition of the Ksylamit® vapor was not reported. The study
authors stated that Ksylamit® consists of "a mixture of chlorinated benzenes, pentachlorophenol, -
chloronaphthalene, chloroparaffin and kerosene." The only exposure information reported was that
at the end of 6-month exposure period the concentrations of formaldehyde and phenol in the
workplace atmosphere were 0.8 mg/m3 and 1.3 mg/m3, respectively. The study authors did not
address how exposure to formaldehyde or phenol resulted from the reported product constituents. In
addition, it cannot be determined from the presented information whether the analytical methods
used would differentiate between phenol and pentachlorophenol (ATSDR, 1998).
The exposed workers reported chronic symptoms such as headache, cough and sore throats,
burning eyes, and fatigue, but morbidity during the 6-month exposure period was comparable to that
of the controls. Although all evaluated hematological parameters were normal in the exposed
workers as a group, some statistically significant changes were observed in a subset of eight workers
who had elevated urinary phenol levels 3 days after the last day of exposure (mean of 18.2 mg/L,
compared with 12.1 mg/L in the exposed workers and 7.9 mg/L for the general population).
Compared with the matched controls, there was a small, but statistically significant decrease among
30
-------
in erythrocyte counts and a statistically significant increase in eosinophil and monocyte counts.
Levels of CD3, CD4, and CDS lymphocytes were also decreased in the exposed group, but there was
no effect on the CD4/CD8 ratio, and the effect was not stronger in the apparently more
highlyexposed subset. Decreases in lymphocyte proliferation induced by phytohemagglutinin and
alloantigens were also observed in exposed workers, whereas reactivity to concanavalin A (Con A)
was unchanged.
These results suggest that exposure to Ksylamit® could affect the immune and hematological
systems. However, the poor characterization of the chemical exposure, including uncertainties
regarding the source of the phenol as well as the marginal dose-response for phenol in urine, mean
that conclusions regarding the contribution of phenol to the observed effects are limited.
Shamy et al. (1994) reported a cross-sectional investigation of phenol-induced biochemical
changes in workers at an oil refining plant in Egypt. The study included 20 workers who were
exposed to a time=weighted average concentration of 5.4 ppm phenol and 30 office workers who
had no exposure to organic solvents. The phenol-exposed workers worked in the aromatic
extraction of distillates; other potential exposures were not described. The mean concentration of
phenol in spot urine samples was 68.6 and 11.5 mg/g creatinine in the exposed and control groups,
respectively. The average duration of exposure was 13.15 years. At the end of the shift of the last
working day of the week, blood samples were collected for hematological and serum biochemistry
evaluations.
Small but statistically significant increases (approximately 55% and 80%, respectively) were
observed in serum glutamic oxaloacetic transaminase (SGOT) and serum glutamic pyruvic
transaminase (SGPT). There were also small but statistically significant increases in hemoglobin,
hematocrit, mean corpuscular hemoglobin, and mean corpuscular volume, although there was no
effect on red blood cell count. This finding is in contrast with those of laboratory animal studies, in
which decreases in erythrocytes and hematocrit have been reported. Other small, but statistically
significant changes included increased basophils and neutrophils, decreased monocytes, and
increased clotting time. A nonsignificant increase in prothrombin time and decrease in platelets was
also observed.
Overall, these data suggest subclinical effects on the liver and hematopoietic system, based
on the small changes in SGOT and SGPT; the observed increases in the hematology endpoints are
not adverse. Although the authors described the workers as exposed to phenol alone and compared
them with other workers exposed to mixed solvents, it appears that the phenol-exposed workers may
31
-------
have also been exposed to other organic compounds that can cause hepatic or hematologic effects,
and the observed effects cannot be clearly attributed to phenol exposure.
Merliss (1972) reported a case of phenol marasmus. A laboratory technician had been
frequently exposed to phenol through vapor or skin contact for 13.5 years. He developed lessened
appetite, weight loss, muscle pain in his legs and arms, and dark color in his urine. Serum
biochemical evaluations indicated liver damage, with SGOT and SGPT at much higher than normal
levels. The patient's symptoms improved after the exposure ended.
4.2. PRE-CHRONIC, CHRONIC STUDIES AND CANCER BIO ASSAYS IN
LABORATORY ANIMALS
4.2.1. Oral
Acute lethality of oral phenol has been evaluated in numerous animal studies. Oral LD50
values in rodents range from 300 mg/kg to 650 mg/kg (Deichmann and Witherup, 1944; Von
Oettingen and Sharpless, 1946; Flickinger, 1976; Berman et al., 1995). The acute toxicity of phenol
when administered by gavage appears to be at least partly dependent on phenol concentration or
total administered volume (Deichmann and Witherup, 1944; NTP, 1983a), and it may be more
severe in young animals (Deichmann and Witherup, 1944). In addition to lethality, acute oral dosing
has been reported to induce adverse renal (tubular necrosis, protein casts, papillary hemorrhage),
hematological (reduction in poly/normochromatic erythrocyte ratio), respiratory (dyspnea and rales),
neurological (muscle twitching, decreased motor activity, ataxia, tremors, convulsions, coma), and
reproductive and developmental effects (Berman et al., 1995; Narotsky and Kavlock, 1995; Liao and
Oehme, 1981; Moser et. al. 1995; NTP, 1983a,b).
As shown in Table 2, there is an extensive database of oral studies relevant to the RfD.
Chronic drinking water studies have been conducted in rats and mice, but the only noncancer
endpoints evaluated were body weight and histopathology (NCI, 1980). Hematology and serum
biochemical evaluations were included in a recent two-generation drinking water study conducted in
rats (Ryan et al., 2001; available in unpublished form as IIT Research Institute, 1999). A specialized
subchronic neurotoxicity study was conducted with rats exposed to phenol in drinking water
(ClinTrials BioResearch, 1998).
These drinking water studies consistently found effects only at exposure levels where water
consumption was also decreased, sometimes by as much as 80%. The decreased water consumption
was presumably due to poor palatability of the drinking water. Effects seen in these studies included
32
-------
tremors, decreased final body weight as compared with the controls (possibly as a result of
dehydration), decreased motor activity, and kidney inflammation. Decreased pup survival was also
observed in the two-generation study. The only drinking water study in which effects were seen in
the absence of decreased water consumption was a 28-day study with mice by Hsieh et al. (1992).
In that study, decreased hematocrit and decreased immune endpoints were observed at doses at least
an order of magnitude lower than the NOAELs in the other drinking water studies. Although the
Hsieh et al. study is useful for hazard identification, its findings need to be confirmed before they
can be used in dose-response assessment. The differing findings of this study and the above-
mentioned two-generation study in rats in which no immunological effects were observed suggest
marked interspecies differences between rats and mice for this endpoint.
Toxicity in gavage studies with phenol is typically much higher than that in the drinking
water studies. NOAELs for systemic effects were 5-10 - fold lower in gavage studies (Berman et
al., 1995; Moser et al., 1995; Dow Chemical Co., 1945) than those seen in the drinking water
studies. Effects observed included tremor and liver and kidney histopathology. As described in
greater detail in Section 4.5, this difference between gavage and drinking water exposure is
33
-------
Table 2. Summary of Oral Toxicity Studies
Strain,
Species,
Sex
Reference
Dose
Route/
Volume
Duration
Response
NOAEL
mg/kg-day
LOAEL
mg/kg-day
Comments
Systemic Toxicity
F344 rat,
50/sex/
group
B6C3F1
mouse,
50/sex/
group
F344 rat,
10/sex/
group
B6C3F1
mouse,
10/group
NCI, 1980
NCI, 1980
NCI, 1980
NCI, 1980
0, 2500, 5000
ppm;
0, 260, 585 (M),
0, 280, 630 (F)
mg/kg-day
0, 2500, 5000
ppm;
0, 450, or 660
mg/kg-day
0, 100, 300, 1000,
3000, 10,000
ppm;
0, 16, 48, 160,
480, 800 (M),
0, 17,51, 170,
510, 1140(F)
mg/kg-day
0, 100, 300, 1000,
3000, 10,000
ppm;
0, 25, 75, 250,
450, 500 (M),
0, 26, 78, 260,
468, 520 (F)
mg/kg-day
Drinking
water
Drinking
water
Drinking
water
Drinking
water
103 weeks
103 weeks
13 weeks
13 weeks
Kidney inflammation
and decreased body
weight (compared to
controls) in both sexes,
decreased water
consumption at high
dose
Decreased body
weight, decreased
water consumption
Decreased body
weight, decreased
water consumption.
Decreased body
weight, decreased
water consumption.
260
450
480
450
585
660
800
500
Study authors stated there
were no noncancer effects,
but independent evaluation
for this assessment found
significant increase in
kidney inflammation.
Effect apparently
secondary to decreased
water consumption.
Range-finding for
bioassay; effect apparently
secondary to decreased
water consumption.
Range-finding for
bioassay; effect apparently
secondary to decreased
water consumption.
34
-------
Strain,
Species,
Sex
Sprague-
Dawley
rat, 15/sex/
group
CD-I
mouse,
5 M/group
Rat
10/group,
strain &
sexNS
F344 rat,
8 F/group
Reference
ClinTrials
BioResearch,
1998
Hsieh et al.,
1992
Dow
Chemical
Co., 1945
Berman et al.,
1995; Moser
etal., 1995
Dose
0, 200, 1000,
5000 ppm;
0, 18,83.1,308
(M),
0, 24.6, 107, 360
(F) mg/kg-day
0,4.7, 19.5,95.2
ppm;
0,1.8, 6.2 or 33.6
mg/kg-day
0, 50, 100
duration adjusted:
0,35.7,71.4
mg/kg-day
0, 4, 12, 40, 120
mg/kg-day
Route/
Volume
Drinking
water
Drinking
water
Gavage/
volume
NS
Gavage/
1 mL/kg
Duration
13 weeks
28 days
6 months,
5 days/wk
14 days
Response
Decreased motor
activity in females,
decreased body weight
in males and females,
decreased water
consumption
Decreased immune
endpoints (particularly
in plaque-forming cell
assay and ELISA),
decreased red blood
cells and hematocrit
Liver and kidney
histopathology;
mortality in 4/10 at
low and high doses
Tubular degeneration,
tremor, increased
rearing post-exposure
NOAEL
mg/kg-day
107
6.2
None
12
LOAEL
mg/kg-day
360
33.6
35.7
40
Comments
Specialized neurotoxicity
study; decreased body
weight apparently
secondary to decreased
water consumption;
unpublished GLP study
BMDL = 219 mg/kg-day
Study limited by small
sample size; confirmation
in a study conducted
according to modern
methods would be useful.
Unpublished study, small
group size, incomplete
reporting; authors raised
questions about the
mortality, although it is
unclear what the issue was.
Small group size, small
dosage volume
Reproductive and Developmental Toxicity
Sprague-
Dawley rat
30/sex/
group PI;
20/sex/
group Fl
Ryan et al.,
2001;IIT
Research
Institute,
1999
0, 200, 1000,
5000 ppm;
0, 14.7, 70.9, 301
(PI M),
0, 20, 93, 320.5
(P1F)
mg/kg-day
Drinking
water
2 genera-
tions
Decreased parental
and pup body weight,
decreased pup
survival, decreased
water consumption
70.9
301
Study also included
evaluation of hematology,
serum biochemistry, and
developmental landmarks.
Effects may be secondary
to decreased water
consumption. Decreased
uterine weight at all doses,
but not considered adverse
35
-------
Strain,
Species,
Sex
Sprague-
Dawley
Rat
25 F/group
CD Rats
20-22
F/group
CD Rat
5-10
F/group
CD-I
Mouse,
31-36
F/group
F344 Rat
15-20
F/group
Reference
Argus
Research
Laboratories,
1997
NTP, 1983a
NTP, 1983a
NTP, 1983b
Narotsky and
Kavlock,
1995
Dose
0,60,120 or 360
mg/kg-day
0, 30, 60 or 120
mg/kg-day
0-250
mg/dg-day
0,70, 140 or 280
mg/kg-day
0, 40, 53.3
mg/kg-day
Route/
Volume
Gavage
10
mL/kg
Gavage
5 mL/kg
Gavage
1-7.5
mL/kg
Gavage
10
mL/kg
Gavage
1 mL/kg
Duration
GD 6-15
GD6-15
GD6-15
GD6-15
GD6-19
Response
Decreased maternal
weight gain;
Decreased fetal body
weight and delayed
ossification
Decreased fetal body
weight
Toxicity (tremors,
liver and lung
pathology, death)
markedly higher in
smaller dosing
volume.
4/36 dams died,
tremors, reduced
maternal body weight
(10%); reduced fetal
body weight
Maternal rales and
dyspnea, marginal
decreases in maternal
body weight
NOAEL
mg/kg-day
60
(maternal)
120 (dev)
120
(maternal)
60 (dev)
N/A
140
(maternal)
140 (dev)
N/A
LOAEL
mg/kg-day
120
(maternal)
360 (dev)
None
(maternal)
120 (dev)
N/A
280
(maternal
PEL)
280 (dev)
N/A
Comments
Doses were divided into 3
administrations/day
One dam died at 360
mg/kg-day
Unpublished GLP study
BMDL = 93 mg/kg-day
None
Developmental LOAEL is
equivocal
BMDL not calculated,
because data on fetal
weight by sex was not
available
Range-finding studies.
None
Screening study
NS = Not Stated; dev = developmental; ELISA = enzyme-linked immunosorbent assay
36
-------
consistent with toxicokinetic data suggesting that toxicity is correlated with peak blood concentrations
rather than with total daily intake.
Several developmental toxicity studies are available (Argus Research Laboratories, 1997;
NTP, 1983a; NTP, Narotsky and Kavlock, 1995). All of these studies were conducted via the gavage
route, although the Argus Research Laboratories study used large dosing volumes and a divided-
dosing protocol, apparently to reduce the effect of peak blood levels. The developmental toxicity
studies found that the primary fetal effect is decreased body weight, which occurs at doses similar to
those that cause decreased maternal body weight gain. The National Toxicology Program (NTP,
1983a) also found that toxicity increased markedly if the same phenol dose was administered in a
lower dosing volume. Because the observed signs of toxicity included tremors and liver and kidney
pathology and were not targeted to the portal of entry, the effect of dosing volume is not related to the
concentration of a direct-contact toxicant.
The National Cancer Institute (NCI, 1980) conducted a carcinogenicity bioassay in which
F344 rats (50/sex/group) received analytical-grade phenol (approximately 98.5% pure) in drinking
water at concentrations of 0, 2500, or 5000 ppm for 103 weeks and were sacrificed 1-2 weeks later.
Using the reference water intake of 0.13 and 0.14 L/kg-day for chronic exposure of male and female
F344 rats, respectively (U.S. EPA, 1988), the doses can be estimated as 0, 260, and 585 mg/kg-day for
male rats and 0, 280, and 630 mg/kg-day for female rats. The doses shown here were adjusted to
account for the reported water consumption of 80% and 90% of control at the low and high doses,
respectively. The animals were observed daily for clinical signs and examined weekly for palpable
masses. Body weights and food consumption were recorded every 2 weeks for the first 12 weeks and
then monthly thereafter; water consumption was recorded weekly.
At the end of study, the animals were sacrificed, and complete gross and histopathological
examinations were performed. Organs and tissues examined included the bone marrow, spleen,
cervical and mesenteric lymph nodes, heart, liver, kidney, thyroid, reproductive organs, brain, and
other major tissues. No evaluation of other noncancer endpoints, such as hematological effects or
serum biochemistry, was conducted.
The survival rate at study termination was comparable among all three groups of males
(approximately 50%) and females (approximately 75%). Dose-related decreases in body weight
compared with the controls were observed in male and female rats, with a decrease of approximately
15% in high-dose males and approximately 10% in high-dose females. Water consumption was
reduced by approximately 10% at the high dose. The authors stated that the non-neoplastic lesions
were similar to those occurring naturally in aged F344 rats. However, an analysis conducted for this
assessment found statistically significant increases (using a chi square test) in chronic kidney
37
-------
inflammation in high-dose males and females; there were no significant changes at the low dose. No
other differences in the incidence of non-neoplastic lesions between the controls and exposed rats
were observed.
On the basis of increased kidney inflammation and decreased body weight as compared with
controls at the high dose of 5000 ppm (585 mg/kg-day for males and 630 mg/kg-day for females), the
NOAEL in this study can be considered to be the low dose, 260 mg/kg-day in males and 280 mg/kg-
day for females, resulting in an overall study NOAEL of 260 mg/kg-day. These effects also indicate
that the maximum tolerated dose (MTD) was reached.
In the NCI (1980) rat study, there were no dose-related trends in cancer incidence in male or
female rats, but the authors reported several tumors for which statistically significant increases were
seen in low-dose males only, based on pairwise comparisons. These increases were seen in the
incidences of pheochromocytomas of the adrenal medulla (13/50, 22/50, and 9/50 in the control, low-
, and high-dose groups, respectively) and "leukemias or lymphomas" (18/50, 31/50, and 25/50). The
historical control incidences of pheochromocytomas in the bioassay program was 9% (data for the
test laboratory were not reported), and the historical control incidence of leukemias or lymphomas in
the test laboratory was 26%. The study authors stated that the leukemias were "of the type usually
seen in untreated F344 rats." There were no significant increases in tumor incidence in any tissue in
female rats. Because there was no clear dose-response in males and the tumors were not observed in
female rats, an association between the tumors and phenol exposure cannot be established. NCI
concluded that phenol was "not carcinogenic in male or female F344 rats." However, the report
noted uncertainties regarding the possible increase in leukemia in male rats, and the NCI reviewers
recommended that phenol be considered for a retest.
In a parallel study, NCI (1980) administered phenol at 0, 2500, or 5000 ppm in drinking water
to B6C3F1 mice (50/sex/group) for 103 weeks and sacrificed the mice 1-2 weeks later. For B6C3F1
mice, the reference water intake was 0.24 L/kg-day for both sexes. The study found that water
consumption was decreased to 75% and 50-60% of the control levels at the low and high doses,
respectively. The resulting doses (adjusting for decreased water intake) were 0, 450, and 660 mg/kg-
day for both sexes. Dose-related decreases in body weight as compared with the controls were
attributed to the decrease in water consumption. No other clinical signs of toxicity were observed,
and mortality rates (approximately 10% in males and 20% in females) were comparable between
experimental and control groups.
Histopathological examination and statistical analyses revealed no phenol-related signs of
toxicity or carcinogenicity; lesions in all systems observed in the dosed groups were comparable to
those in the controls. NCI concluded that, under the conditions of the assay, phenol was not
38
-------
carcinogenic in male or female B6C3F1 mice. On the basis of the decreased body weight as compared
with controls observed at 5000 ppm, the low dose of 2500 ppm (450 mg/kg-day) can be considered the
study NOAEL. The observed effect, however, is likely secondary to the decreased water consumption
due to poor palatability.
In light of the marked decrease in water consumption, higher doses of phenol in drinking water
probably could not have been tested. If the authors had attempted to overcome the palatability issue by
administering the high dose by gavage rather than in drinking water, high toxicity would have been
expected in light of the higher toxicity of phenol administered by gavage (see Section 4.5 and Table 2).
These considerations suggest that an MTD was also reached in mice, although the conclusion is less
clear than for rats.
In the range-finding test for the carcinogenicity bioassay (NCI, 1980), F344 rats and B6C3F1
mice (10/sex/group) were given drinking water containing 0, 100, 300, 1000, 3000, or 10,000 ppm
phenol (at a purity of 98.47%) for 13 weeks. Using the reference water intake of 0.16 and 0.17 L/kg-
day for subchronic exposure of male and female F344 rats, respectively (U.S. EPA, 1988), the doses
can be estimated as 0, 16, 48, 160, 480, and 800 mg/kg-day for male rats and 0, 17, 51, 170, 510, and
1140 mg/kg-day for female rats. The high doses shown here were adjusted to account for the
decreased water consumption described below. For B6C3F1 mice, the reference water intake was 0.25
L/kg-day for males and 0.26 L/kg-day for females. The corresponding doses (adjusting for decreased
water intake at the high dose) were 0, 25, 75, 250, 450, and 500 mg/kg-day for males and 0, 26, 78,
260, 468, and 520 mg/kg-day for female mice.
Body weights, appearance, behavior, and food and water consumption were recorded weekly.
After 13 weeks, all animals were sacrificed and tissues were subjected to histopathological
examinations. All of the rats and mice survived the phenol treatment. The only significant observation
was the decreased final body weights (compared to controls) in rats of both sexes (11-14%) and in
male mice (12%) that received 10,000 ppm. Because drinking water consumption in these groups was
decreased to 50-70% (rats) and 20-60% (mice) of the control value, the decreased body weight was
likely due to the low water consumption. No histopathological changes attributable to phenol
treatment were observed.
This study suggests that the second-highest dose (480 mg/kg-day for male rats, 510 mg/kg-day
for female rats, 450 mg/kg-day for male mice, and 470 mg/kg-day for female mice) was a NOAEL,
based on the decreased final body weight (compared to controls) at 10,000 ppm, which was secondary
to decreased water consumption due to poor palatability at the high dose.
In an unpublished 13-week neurotoxicity study conducted according to good laboratory
practices (GLP) guidelines (ClinTrials BioResearch, 1998), groups of 15 male and 15 female Sprague-
39
-------
Dawley rats received phenol via drinking water at concentrations of 0, 200, 1000, or 5000 ppm (at a
purity of 100%) for 13 weeks, followed by a 4-week recovery period.2 The authors calculated that the
average doses were 0, 18.1, 83.1, and 308.2 mg/kg-day for males and 0, 24.6, 107.0, and 359.8 mg/kg-
day for females. These concentrations were chosen on the basis of preliminary palatability studies
conducted at a different laboratory (IITRI project No. L08657).3
During the exposure period, clinical signs and water intake were recorded daily, and body
weight and food consumption were recorded weekly. In addition, a functional observational battery
(FOB) and a motor activity test were conducted pre-study and once each during weeks 4, 8, 13, and 17.
At the end of the exposure and the end of the recovery period, five rats/sex in the control and 5000
ppm groups underwent neuropathological evaluations (including a thorough evaluation of the brain and
several nerves). The rest of the rats in the ClinTrials BioResearch (1988) study were sacrificed at the
end of the 4-week recovery and were subjected to gross necropsy.
One high-dose female was euthanized on day 14 due to poor condition. Clinical signs observed
in this female prior to sacrifice included dehydration, hunched posture, tremors, reduced activity, and
cold to touch. Among the rest of the high-dose animals, the primary clinical sign was dehydration,
which was accompanied by reduced activity and tremors in one female and by a thin appearance in
additional animals. Dehydration was also observed in mid-dose rats (2/15 in each sex). Dehydration
was assessed qualitatively and independently of drinking water consumption by grabbing the scruff on
the back of the animal's neck; a delay in returning to the normal position was considered dehydration.
Dehydration was associated with marked decreases in water consumption at the high dose and
smaller decreases at the mid dose. Decreases in water consumption were more pronounced in females
than in males and were most evident during the first week of dosing. Water consumption was
decreased to approximately 90% of the control level in mid-dose males and females, to approximately
60% of control levels in high-dose males, and to approximately 55% (40% during the first week) of
control levels in high-dose females. Water consumption rebounded to levels higher than those of
2This study has not been peer-reviewed, but it was conducted (with minor deviations) according
to EPA guidelines for a neurotoxicity screening battery, it is well documented, and it contributes useful
information to the hazard identification and dose-response portions of the assessment. The study was
designed to comply with the U.S. EPA Enforceable Consent Agreement for Phenol (Docket No. OPPTS-
42150).
3Results of the palatability study were not provided in the IITRI study report (IIT Research
Institute, 1999; Ryan et al., 2001), which reports the results of the two-generation reproduction study.
The reproduction study was conducted during the same time period as the neurotoxicity study and
reported similar problems of markedly decreased drinking water consumption at the high dose of 5000
ppm (see Section 4.3).
40
-------
controls during the recovery period. The decreased water consumption was likely due to the poor
palatability of phenol at high concentrations rather than being a manifestation of an overt toxicological
effect. In addition, the high-dose group had decreased body weights when compared to controls (8%
for males and 12% for females) and decreased food intake (approximately 10% for males and 10-20%
for females).
The only lexicologically significant neurological effect was decreased motor activity in
females. There was a statistically significant reduction in total group mean motor activity counts at
week 4 in the 5000 ppm group. The authors reported that the rate of linear change of motor activity
with time was also significantly decreased at weeks 8 and 13 in the 1000 ppm and 5000 ppm groups,
although supporting data were not provided. Motor activity in females at week 4 exhibited a dose
dependency at the first five (of six) analysis intervals, although the total counts for the low- and mid-
dose groups were not significantly different from control. High-dose females also had markedly lower
total activity counts than did controls and lower counts in the first four intervals, at week 4, although
there was no statistically significant difference in mean total counts (Table 3). By contrast, the high-
dose males had markedly lower group mean activity counts pre-study but activity comparable to or
higher than the controls at weeks 4, 8, and 13.
The authors attributed the decreased activity to dehydration, noting that the control group mean
total activity increased by >20% at week 4 as compared with pre-study levels, whereas activity of
dehydrated females in the 5000 ppm group at week 4 was decreased by 17%, and activity of females in
this group that were not dehydrated increased by 2%. To address whether the decreased activity could
be attributed to dehydration, this assessment evaluated the data in greater detail. Table 4 presents the
individual animal data for week 4 total motor activity counts and compares them with the individual
animal dehydration data. If the individual clinical data reported an animal as dehydrated, the days of
that notation are shown. With the exception of animal 4502 (which died) and animal 4507 (which had
severe dehydration) dehydration was noted as slight or moderate. For clarity of presentation, the
individual animal data are shown for the control and high-dose groups, but only the average data are
shown for the low- and mid-dose groups.
Table 3. Total Activity Counts in Rats Provided Phenol in Drinking Water
(ClinTrials BioResearch, 1998)
Dose
Group
Prestudy
(Mean ±SD)
Week 4
(Mean ±SD)
WeekS
(Mean ±SD)
Week 13
(Mean ±SD)
Females
41
-------
Control
200 ppm
lOOOppm
5000 ppm
384±116
386±89
384±103
372±142
468±118
451±149
394±78
337±127**
436±75
440±99
436±1041J
363±llltTf
309±77
338±66
343±1241J
366±145tTf
Males
Control
200 ppm
1000 ppm
5000 ppm
354±109
340±107
335±126
277±59
339±89
346±132
356±137
321±95
320±90
323±88
359±105
352±91
260±68
256±78
274±103
275±116
** Significantly different from control, p<0.01 (T-test)
TJLinear constructed variable significantly different from control, p<0.05 (T-test)
^[Linear constructed variable significantly different from control, p<0.01 (T-test)
42
-------
Table 4. Individual Data on Dehydration and Week 4 Motor Activity in Rats Provided Phenol in Drinking Water
(ClinTrials BioResearch, 1998)
Group 1
(Control)
Animal #
1501
1502
1502
1504
1505
1506
1507
1508
Total
Counts
383
321
621
437
630
365
591
318
Days
Dehydrated
No
No
No
No
No
No
No
No
Group 2
(200 ppm)
Animal
#
2501
2502
2503
2504
2505
2506
2507
2508
Total
Counts
397
529
427
558
245
537
470
284
Days
Dehydrated
No
No
No
No
No
No
No
No
Group 3
(1000 ppm)
Animal
#
3501
3502
3603
3504
3505
3506
3507
3508
Total
Counts
322
402
270
370
572
452
461
342
Days
Dehydrated
No
18,21
No
No
21,28,35,
42, 49, 56,
70
No
No
No
Group 4
(5000 ppm)
Animal
#
4601
4502
4503
4504
4505
4506
4507
4508
Total
Counts
501
Days
Dehydrated
14,21
No data - sacrificed
day 14 due to poor
condition
227
258
396
277
399
271
7, 14, 21, 28
14,21,28,
35, 42, 70
No
70
7-9,11,12,
13-15, 17,
20,21,70,
77
No
43
-------
Group 1
(Control)
Animal #
1509
1510
1511
1512
1513
1514
1515
Overall
Average
Average -
Dehydrated
animals
Average -
non-
dehydrated
animals
Total
Counts
479
469
309
574
566
381
578
468
N/A
468
Days
Dehydrated
No
No
No
No
No
No
No
-
-
-
Group 2
(200 ppm)
Animal
#
2509
2510
2511
2512
2513
2514
2515
Total
Counts
527
823
561
424
302
289
386
451
N/A
451
Days
Dehydrated
No
No
No
No
No
No
No
-
-
-
Group 3
(1000 ppm)
Animal
#
3509
3510
3511
3512
3513
3514
3515
Total
Counts
462
452
390
383
320
403
311
394
487
(n=2)
380
Days
Dehydrated
No
No
No
No
No
No
No
-
-
-
Group 4
(5000 ppm)
Animal
#
4509
4510
4511
4512
4513
4514
4515
Total
Counts
387
450
439
130
242
180
556
337
315
365
Days
Dehydrated
No
No
7
No
49,56
7, 14,21,28
No
-
-
—
N/A = Not applicable
44
-------
Only two animals in the mid-dose group were reported as dehydrated on any day, and neither
of these animals had decreased motor activity. As shown, the average activity was lower in the
dehydrated high-dose females than in those not reported as dehydrated, but an association of
decreased activity with dehydration was not consistently supported on an individual-animal basis.
(For the purpose of calculating averages, animals were considered dehydrated if they dehydrated at
any point in the study. This is a limitation to the analysis, because some were reported as dehydrated
only prior to week 4 and others were reported as dehydrated only after week 4. In addition, basing
the analysis on the clinical sign of dehydration may not appropriately reflect whether the animals
were dehydrated, because no objective measure of dehydration was used and because decreased
water consumption in this group occurred throughout the study.)
As shown in Table 4, animal 4601 was reported as dehydrated on days 14 and 21, but it had
one of the highest total activity counts. Conversely, animal 4512 had the lowest activity count, but it
was never reported as being dehydrated. Furthermore, the mean activity of the dehydrated high-dose
females was 67% of concurrent controls, compared to 78% of concurrent controls for the
nondehydrated high-dose females. These data indicate that the difference between the control and
high-dose animals was greater than the difference between the dehydrated and nondehydrated
animals at the high dose.
Overall, the data indicate that there was not a tight linkage between dehydration and
decreased motor activity in the high-dose females. The data for high-dose males also did not indicate
a clear correlation between low activity and dehydration. The clinical signs for one high-dose male
(4003) for week 2 included severe dehydration and decreased activity, but no effect (i.e., no
dehydration or decreased activity) was seen when the animal underwent the objective activity
analysis in week 4. The finding of dehydration in males without the accompanying decrease in
activity further supports the conclusions that only severe (not mild or moderate) dehydration results
in decreased motor activity levels and that the decrease observed in females was phenol related.
Conversely, the absence of other findings in the FOB and the presence of a statistically significant
effect on motor activity only at 4 weeks and not at later time points argue against a neurotoxic effect
of phenol.
As an additional investigation of whether decreased motor activity was related to
dehydration, the very limited literature on water deprivation and motor activity was reviewed.
Campbell and Cicala (1962) evaluated the effects of terminal water and food deprivation (i.e.,
deprivation until death from dehydration or starvation) on motor activity of male and female Wistar
rats. Motor activity was measured using a stabilimeter, which is similar to the figure-8 mazes used in
45
-------
the ClinTrials BioResearch (1998) study in that ambulation (as opposed to simply movement) is
measured. The study found that water deprivation alone did not result in decreased motor activity
until approximately days 5-7 (depending on age), at which time activity continuously declined until
death. By contrast, food deprivation resulted in an initial increase in activity followed by decreasing
activity until death.
Only the pooled data for males and females were reported. These results are not directly
comparable to the results of the ClinTrials study because the latter involved long-term, lower-level
dehydration; however, they do support the conclusion that the decreased motor activity in high-dose
females was due at least partially to phenol exposure. The most appropriate way to address this issue
would be to conduct the neurotoxicity study with a water-restricted control group. Overall, based on
the decreased motor activity, the study NOAEL in females was 1000 ppm phenol (107 mg/kg-day)
and the Lowest Observed Adverse Effect Level(LOAEL) was 5000 ppm (360 mg/kg-day ). No
LOAEL was identified in males; the high dose of 308 mg/kg-day was a NOAEL. A 95% lower
confidence limit on the benchmark dose (BMD) of 219 mg/kg-day was calculated for decreased
motor activity in week 4 in this study (see Appendix B).
Hsieh et al. (1992) investigated the effects of phenol exposure on hematological, immune,
and neurochemical endpoints in a study of 6-week-old male CD-I mice administered actual
concentrations of 0, 4.7, 19.5, or 95.2 ppm in drinking water for 28 days. On the basis of measured
concentrations and water intake, the authors reported that the corresponding daily doses were 0, 1.8,
6.2, and 33.6 mg/kg-day.
The mice were housed in groups of five per cage. Drinking water was prepared and changed
every 3 days. Drinking water was provided in glass water bottles with stainless sipper tubes
containing ball bearings to minimize evaporation; the bottles were shaken frequently during
treatment. Food and water consumption were monitored continuously, and the animals were
weighed weekly. After 28 days, the mice were sacrificed by decapitation, gross pathological
examinations were performed, and the liver, spleen, thymus, and kidney were weighed. Blood was
taken at sacrifice for analysis. Splenocytes were prepared for analysis of mitogen-stimulated
lymphocyte proliferation, mixed lymphocyte response, and cell-mediated cytolytic response.
Data were reported for five animals per group for each assay. During the 28-day exposure,
no mortality and no overt clinical signs occurred in exposed mice. Phenol treatment had no effect on
food or water consumption or on body weight gain. Exposed mice had no gross lesions in the liver,
kidney, spleen, thymus, lung, heart, or brain, and there were no effects on organ weights for the
liver, kidney, spleen, and thymus. As shown in Table 5, a decreased antibody response to sheep red
46
-------
blood cells was observed, as indicated by both the plaque-forming cell (PFC) assay (expressed as
PFC/million spleen cells and PFC/spleen) and the antibody titer using an enzyme-linked
Table 5. Effects of Phenol Exposure on Spleen Cellularity
and Selected Blood Parameters in Mice and Rats
Concentration
(mg/L)
Dose (mg/kg-
day)
PFC/106 splenic
cells
PFC/total spleen
Antibody titer b
Hsieh et al. (1992) - 4-week study in CD-I mice
0
4.7
19.5
95.2
0
1.8
6.3
33.6
l,123±99a
896 ± 60
795 ± 49C
616±83C
265,947 ± 53,099
214,678 ±17,500
207,659 ±18,886
130,185 ±18,202C
0.446 ± 0.039
0.392 ± 0.068
0.325 ±0.019C
0.263 ± 0.037C
IIT Research Institute (1999); Ryan et al. (2001) - 2-generation study in Sprague-Dawley rats, effects in PI
generation
0
200
1000
5000
Positive control
(n=5)
0
15
71
301
cyclo-
phosphamide
1343±890d
1668±788
1781±1151
1880±865
0±0
5.54xl05±3.70xl05
6.42xl05±3.40xl05
9.01xl05±7.16xl05
9.81xl05±5.02xl05
0±0
Not assayed
Not assayed
Not assayed
Not assayed
Not assayed
a Values are given as mean ± S.E. (n=5). bArbitrary as change in absorbance at 490 nm, using 1:2000
diluted serum. ° Significant (P<0.05) difference from the control value. d Values are given as mean
±SD (n=8-9)
47
-------
immunosorbent assay (ELISA). Two of these measures were statistically significantly decreased at
the mid dose, and PFC/spleen was significantly decreased only at the high dose. Decreases in the
absolute splenocyte lymphoproliferative responses to mitogens and the mixed lymphocyte response
(the proliferative ability of splenic lymphocytes in response to alloantigens) were also observed at
the high dose; there was no effect on the cytolytic response to tumor cells at any dose.
Although these assays were conducted according to the methods of the day, the latter two
assays do not conform to modern protocols, and there is little biological significance in the results of
the mitogen response assay. In particular, the approach used cannot distinguish between an effect on
the lymphocyte's ability to initiate a proliferative response and confounding due to contamination by
nucleated red blood cell precursors.
A statistically significant, dose-related decrease in erythrocyte counts was observed at all
doses (Table 6). The hematocrit was decreased only at the high dose. A decreased erythrocyte count
in the absence of an effect on hematocrit may have been due to macrocytosis (enlarged erythrocytes),
but insufficient data were provided to evaluate this possibility. In the absence of such data, the
decreased erythrocyte counts are insufficient to form the basis for identification of a LOAEL. The
erythrocyte counts in all dosed groups were markedly lower than the historical control values
provided by the animal distributor (Charles River Laboratories, 1986), although the hematocrit
concentration in all groups was above the historical control mean. There was no effect on total or
differential leukocyte counts. Interestingly, total white blood cells for all groups, including the
controls, were below the historical control data provided by the distributor.
Hsieh et al. (1992) also observed dose-related decreases in the concentration of several
neurotransmitters and their metabolites in the brain, including levels of norepinephrine, indoleamine
serotonin, and dopamine and their metabolites. In the absence of a clear correlation with clinical
effects, the toxicological significance of these neurobiochemical findings is unclear.
Thus, this study found dose-related, statistically significant decreases in red blood cells at all
doses, but the significance of this finding is uncertain, because decreased hematocrit was observed
only at the high dose. Statistically significant decreases in antibody response were observed at the
mid dose, and these decreases reached 40% (a value often used by immunotoxicologists as a rule of
thumb for clinically relevant decreases) at the high dose. Identification of a NOAEL in this study is
somewhat problematic, because immunotoxicity risk assessment guidelines have not been developed.
The determination of what degree of decrease is adverse is also problematic, because the
clinical relevance of a decrement in immune function will depend on the magnitude and type of
48
-------
Table 6. Effects of Phenol Exposure on Spleen Cellularity and Selected Blood Parameters in Mice and Rats
Concentration
(mg/L)
Dose
(mg/kg-day)
Spleen
cellularity
(x 10'7)
WBCa
(x 1C'3)
RBC"
(x 10-6)
Hematocrit
%
Differential counts as % of WBCs
Lymphocyte
Neutrophil
Monocyte
Hsieh et al. (1992) - 4-week study in CD-I mice
0
4.7
19.5
95.2
0
1.8
6.3
33.6
Historical control valued
8.59±0.34b
7.94 ±0.20
7.31 ±0.40
7.26 ±0.55
Not available
6.06 ±0.17
5.82 ±0.60
5.05 ±0.53
5.68 ±0.69
9.0(8.9-9.1)
7.17 ±0.56
4.90±0.54C
4.64±0.76C
3.23±0.68C
7.6 (7.2-8.0)
48.00 ±0.52
49.10 ±0.68
48.20 ±1.24
44.10 ±0.81C
42 (36-48)
74.20 ±1.83
71. 80 ±2.06
69.20 ±3.25
73. 60 ±2.32
70 (52-86)
17.00 ±1.00
19.40 ±0.75
21.80 ±2.40
17.00 ±1.55
25 (10-42)
4.60 ±0.51
4.80 ±1.02
4.60 ±0.81
6.20 ±1.16
4 (0-8)
IIT Research Institute (1999); Ryan et al. (2001) - 2-generation study in Sprague-Dawley rats, effects in PI generation
0
200
1000
5000
0
15
71
301
43.2±13.5e
38.3±8.78f
48.1±11.0
52.7±13.4
13.1±2.01
13.8±1.98
14.5±2.42
14.9±2.93
9.22±0.37
9.08±0.62
9.03±0.34
8.81±0.44
46.5±1.44
46.2±3.65
46.4±1.56
45.1±1.75
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
aCells/mm3. bValues are given as mean ± S.E. (n=5); N/A = not assayed. Significant (P < 0.05) difference from the control value. dMean (Range ±2S.D) for mice 6-
8 weeks of age, based on 20 studies, from Charles River Laboratories (1986). eMean ±SD (n=9-10 for hematology endpoints and n=8-9 for spleen cellularity).
Standard deviation reported in Ryan et al. (2001) has a typographical error; correct standard deviation obtained from IIT Research Institute (1999)
49
-------
immune challenge, with a sufficiently large challenge resulting in illness even for unimpaired
individuals. In a paper on the use of immunotoxicity data for risk assessment, Belgrade (1999)
recommended that any statistically significant and consistent change be considered a risk for the
purposes of hazard identification, but the degree of change considered adverse for the purposes of
dose-response assessment was not addressed.
On the basis of magnitude of the decreases in antibody response observed in three
related—assays, supported by decreased hematocrit and red blood cells—the high dose (33.6 mg/kg-
day) can be considered the study LOAEL, and the mid dose (6.2 mg/kg-day) can be considered the
study NOAEL. There is, however, considerable uncertainty regarding the reliability of these values
because of issues of study interpretation and because the study used only five animals per group as
opposed to the recommended eight per group (U.S. EPA, 1998c).
The results of BMD modeling conducted for this study are presented in Appendix B for
completeness. However, it is unclear what the appropriate benchmark response (BMR) would be for
an in vivo/in vitro immunotoxicity study, and so the modeling results are considered very
preliminary. In particular, it is unclear whether the default of one standard deviation is appropriate
as the BMR for this study design in light of the small sample size (n = 5) but the relatively tight data.
In contrast to the minimal effects observed in these drinking water studies, gavage dosing
with phenol produces severe toxicity, including liver and kidney pathology, and death at doses that
cause only minimal effects when delivered in drinking water.
Dow Chemical Co. (1945) administered 0, 50, or 100 mg/kg phenol by gavage 5 days/wk to
10 rats per group (sex and strain not reported) for 6 months (0, 35.7, or 71.4 mg/kg-day after
adjusting for intermittent dosing). The dosing volume was not reported. Mortality occurred in 1/10,
4/10, and 4/10 rats in the control, low-, and high-dose groups. The authors raised questions about
whether the mortality was treatment-related, but it is not clear whether they questioned whether the
deaths were due to phenol or to gavage accidents. Other observed effects were slight cloudy swelling
of the liver and of the tubular epithelium at the high dose and slight tubular degeneration at the low
dose. This unpublished study4 is limited by the incomplete reporting of methods and results, but the
low dose of 35.7 mg/kg-day appears to be a LOAEL.
4Although this unpublished study is not a primary reference for this assessment, it is presented
here because it contributes some useful information to the overall hazard identification phase of the
phenol assessment.
50
-------
In a series of toxicological screening tests, the systemic, neurological, and developmental
effects of phenol in F344 rats following acute and short-term oral exposure were examined
(Narotsky and Kavlock, 1995; Herman et al, 1995; and Moser et al, 1995; MacPhail et al, 1995).
In these tests, systemic and neurological effects were examined on the same animals following
exposure by gavage to a single dose of phenol or to 14 consecutive daily doses. Developmental
toxicity was also examined in pregnant rats that received phenol by gavage on GDs 6-19. The
dosing volume was 1 mL/kg (Moser et al., 1995; Narotsky and Kavlock, 1995).
In the acute toxicity study of this series, groups of eight female rats were given a single
gavage dose (1 mL/kg volume) of phenol at 0, 12, 40, 120, or 224 mg/kg in water (Berman et al.,
1995; Moser et al., 1995). An FOB evaluating autonomic and neuromuscular functions, activity,
excitability, and sensorimotor and physiological measures was conducted prior to the exposure and
at approximately 4 and 24 hours after exposure. Immediately after the 24-hour FOB, the animals
were sacrificed, blood samples were collected for serum chemistry analyses, and the liver, kidneys,
spleen, thymus, and adrenals were weighed and subjected to histopathological examinations. Two
rats (25%) died within 4 hours of exposure to 224 mg/kg, and one rat died 24 hours after exposure to
120 mg/kg phenol. The only treatment-related effects observed were confined to these two dose
groups, and they included tremor, decreased motor activity, and kidney pathology (necrosis, protein
casts, and papillary hemorrhage). Hepatocyte necrosis was also observed at 40 and 120 mg/kg but
not at 224 mg/kg. No other effects were reported at the lower doses, although the primary data were
not provided.
In the short-term study, groups of eight female rats were given daily gavage doses of phenol
in water at 0, 4, 12, 40, or 120 mg/kg-day for 14 consecutive days (Berman et al., 1995; Moser et al.,
1995). As in the acute study, the FOB was conducted prior to exposure as well as on days 4 and 9
(before the daily dose) and approximately 24 hours after the last dose. After the last FOB, blood
samples were collected for serum chemistry analyses, and internal organs were removed, weighed,
and subjected to histopathological examinations. All rats administered the high dose died during the
study, but deaths occurred over the entire dosing period. Tremor was also seen in the high-dose
(120 mg/kg) group immediately after the first administration but not after subsequent treatment.
Vacuolar degeneration of the liver, kidney necrosis and protein casts, and "necrosis or atrophy of
spleen or thymus" were reported at 40 mg/kg-day. The increased incidences were not large enough
to be statistically significant; the statistical power of the study was also low, with only 8 rats per
group.
Additional information on this study is available from a preliminary abstract (Schlicht et al.,
1992) and from a recent WHO (1994) review. According to these sources, the renal pathology
51
-------
consisted of 3/8 rats with renal vascular stasis, 2/8 rats with tubular degeneration in the papillar
region, and 1/8 rats with protein casts in the tubules. WHO (1994) states that, according to a
personal communication from one of the study authors, the pathology report attributed the renal
findings to decreased vascular perfusion.
The study also found slight but not statistically significant decreases in motor activity at 40
mg/kg-day. The only statistically significant effect in this group was increased rearing in the post-
exposure measurement. The only effect at 12 mg/kg-day was "necrosis or atrophy in the spleen or
thymus" in 1/8 rats. On the basis of the liver, kidney, and thymus/spleen pathology findings, which
are rarely observed in control animals in 2-week studies, and the decreased motor activity, the
second dose (40 mg/kg-day) was the study LOAEL, and the mid dose of 12 mg/kg-day was the study
NOAEL.
4.2.2. Inhalation
The laboratory animal inhalation data for phenol are very limited, with only one 2-week
toxicity study being conducted using modern methodology and documentation (Hoffman et al,
2001; additional details available in the unpublished version, Huntingdon, 1998). Although a
subchronic study conducted with multiple laboratory animal species is available (Sandage, 1961),
this latter unpublished study tested only one concentration and was insufficiently documented for
definitive risk assessment purposes. Other short-term (Dalin and Kristoffersson, 1974) or
subchronic (Deichmann et al., 1944) inhalation toxicity studies are limited by short duration,
inadequate documentation, or lack of a modern exposure protocol. Nonetheless, the data are
consistent that the respiratory tract, kidney, and nervous system are targets of inhalation exposures.
In conducting dosimetric conversions from animal studies to human exposure scenarios, U.S.
EPA (1994b) classifies gases according to their water solubility and reactivity. Category 1 gases are
highly water-soluble and/or rapidly reactive and do not penetrate the blood. Category 3 gases are
water insoluble, and uptake from the lungs is limited by perfusion. Category 2 gases are
intermediate between these two groups. They are moderately water-soluble and rapidly-reversibly
reactive or moderately-to-slowly irreversibly metabolized in the respiratory tissue. On the basis of
phenol's chemical/physical properties (see Table 1) of moderate water solubility and moderate
reactivity (based on the evidence of irritation and corrosivity seen following direct contact), it can be
considered a Category 2 gas. This conclusion is supported by the finding of both respiratory effects
(from direct contact) and systemic (extrarespiratory) effects (from absorbed phenol) following
inhalation exposure, as described below.
52
-------
Because the equations for the regional gas dosimetry ratio (RGDR) for Category 2 gases are
currently undergoing EPA reevaluation (eqs. 4-29 through 4-44, pages 4-52 through 4-57 of U.S.
EPA, 1994b), dosimetric adjustments for extrarespiratory effects were made using the Category 3
equations (eq. 4-48, page 4-60 of U.S. EPA, 1994b), in which the RGDR is based on the blood:air
partition coefficient for the chemical in the experimental animal species and in humans. No data on
the blood: air partition coefficient for phenol in laboratory animals or in humans were located.
Therefore, the default value of 1 for the ratio of the laboratory animal-to-human partition coefficient
was used, and the human equivalent concentration (HEC) for systemic effects was the same as the
duration-adjusted concentration.
HECs for respiratory tract effects were calculated using the equations of U.S. EPA (1994b)
for a Category 1 gas (eqs. 4-17 through 4-28, pages 4-47 through 4-51). When the EPA reanalysis is
complete, revised dosimetric conversions may be calculated. The inhalation toxicity data for phenol
are summarized in Table 7.
The acute toxicity studies support the findings of the short-term and subchronic studies that
the respiratory tract and nervous system are targets of inhaled phenol. For example, tremors were
seen in rats and guinea pigs exposed to 187 or 540 ppm (720 or 2080 mg/m3) phenol for 30 minutes
in a whole-body inhalation chamber (UBTL, 1991). By contrast, no tremors were observed in rats
exposed via nose-only inhalation at 25 ppm (96 mg/m3) for 6 hours (Dow Chemical Co., 1994).
Phenol also caused sensory irritation in mice, as evidenced by decreased respiratory rate (De
Ceaurriz et al., 1981). The concentration associated with a 50% decrease in rate (RD50) was
estimated to be 166 ppm (639 mg/m3). No acute lethality studies were identified for phenol
following exposure by the inhalation route.
In a 2-week inhalation study conducted according to GLP guidelines (Hoffman et al., 2001;
full study report available as Huntingdon, 1998), groups of 20 F344 rats per sex were exposed nose-
only to actual concentrations of 0, 0.52, 4.9, or 25 ppm phenol (0, 2.0, 18.9, or 96.2 mg/m3) 6
hrs/day, 5 days/wk for 2 weeks. The duration-adjusted concentrations were 0, 0.36, 3.4, and 17
mg/m3, respectively. The animals were observed twice daily for mortality and abnormal clinical
signs. Animal body weights and food consumption were recorded twice pre-test, weekly thereafter,
and just prior to sacrifice. At the end of 2 weeks of exposure, 10 rats of each sex in each group were
sacrificed. The rest of the rats were sacrificed after 2 weeks of recovery. Blood samples were
collected just prior to sacrifice for hematological (including differential leukocyte count) and
53
-------
Table 7. Summary of Inhalation Toxicity Studies
Strain
Species,
Sex
F344 Rat
(20/sex)
Rat, 7
exposed,
11-12
controls
(Strain &
sex NS)
Guinea pig,
12
(Strain &
sex NS)
Rabbit
6 exposed
(Strain and
sex NS)
Rat, 12
(Strain and
sex NS)
Rhesus
monkey
10 M/
group
Reference
Huntingdon
1998
Dalin and
Kristoffersson
1974
Deichmann et
al. 1944
Deichmann et
al. 1944
Deichmann et
al. 1944
Sandage 1961
Exposure
mg/m3
0,2.0,18.9,
96.2
Nose-only
0,100
100-200
100-200
100-200
0, 18.2
Duration
6hr/d
5 d/week
2 weeks
15 days
continuous
7 hr/day
5 d/week
6 weeks
7 hr/day
5 d/week
1 3 weeks
7 hr/day
5 d/week
74 days
90 days,
continuous
Duration-
Adjusted
mg/m3
0,0.36,3.4,17
0,100
3 1 (based on
midpoint of
range)
3 1 (based on
midpoint of
range)
3 1 (based on
midpoint of
range)
0, 18.2
Response
Red nasal discharge, but no
histopathology lesions.
Nervous system effects,
increased serum liver
enzymes
PEL -5/1 2 dead
Pneumonia, heart
inflammation, liver necrosis,
kidney tubular degeneration
No effect, no evidence of
histopathology
Liver and kidney pathology
(Not further described)
NOAEL/
LOAEL
mg/m3
17/None
None/100
None
None
31 /None
None/18.2
NOAEL/
LOAEL (HEC)
mg/m3
2.5 or 17
/None
None/100
None/
31 is PEL
None/
31
31 /None
None/18.2
Comments
Well-conducted study, but
authors did not note the
clinical signs
Exposure measurement not
done according to modern
methods, no histopathology
exam
Minimal documentation,
outdated exposure methods,
no controls
Minimal documentation,
outdated exposure methods,
no controls
Minimal documentation,
sensitivity of assay unclear,
outdated exposure methods,
no controls
Pathology reported to be
minimal, but limited by
minimal description
Unpublished study
54
-------
Strain
Species,
Sex
Sprague-
Dawley rat
50 M/
group
Albino
mouse 100
M/group
Reference
Sandage 1961
Sandage 1961
Exposure
mg/m3
0, 18.2
0, 18.2
Duration
90 days,
continuous
90 days,
continuous
Duration-
Adjusted
mg/m3
0, 18.2
0, 18.2
Response
Liver and kidney pathology
(Not further described)
Lung pathology
(Not further described)
NOAEL/
LOAEL
mg/m3
None/18.2
None/18.2
NOAEL/
LOAEL (HEC)
mg/m3
None/18.2
Not
determinable*
Comments
Pathology reported to be
minimal, but limited by
minimal description
Unpublished study
Pathology reported to be
minimal, but limited by
minimal description
Unpublished study
NS = Not stated
* HEC cannot be determined because the region of the respiratory tract affected is not clear.
55
-------
biochemical examinations. Gross pathological evaluations were conducted on all of the animals,
and organ weights were determined. Histopathological examinations were conducted on the liver,
kidney, and respiratory tract tissues (including three sections of the lungs with mainstem bronchi, the
pharynx, and three sections of the nasal turbinates) of the control and high-exposure groups; the
spleen of mid-concentration females was also analyzed.
During the exposure, one male rat in the low-concentration group accidentally died from
trauma caused by turning itself within the nose-only restraint tube. All of the other rats survived
until sacrifice at the end of the 2-week exposure or 2-week recovery periods. During exposure and
recovery, there were no treatment-related changes in weekly physical examinations, body weight,
weight gain, or food consumption. The authors reported no effect on clinical signs. However, there
was a concentration- and duration-related increase in the incidence of a red nasal discharge in the
males. The incidence was 0/20, 0/20, 3/20, and 4/20 at 0, 0.52, 4.9, and 25 ppm, respectively, in the
first week and 0/20, 0/20, 7/20, and 10/20 in the second week of exposure. This detailed information
on nasal discharge was presented only in the unpublished report (Huntingdon, 1998); the published
version reported scattered observations of chromodacryorrhea and nasal discharge.
In an analysis done for this assessment, the incidence at the mid and high concentrations was
statistically significant, using the Fisher exact test. In females, nasal discharge was seen in 1/20 at
the low concentration and 3/20 at the mid concentration in the second week, but no discharge was
reported in high-concentration females. Prior to exposure, a nasal discharge was observed in a
single control male and a single high-concentration female. Considering the exposure chamber
design, it does not appear that the discharge was an artifact of the rats' noses being in contact with
phenol condensate on the chamber walls. Instead, in the absence of nasal histopathology, it is likely
that the discharge reflected a nonspecific response to stress in the rats. A tear-like nasal discharge in
rats can be a generalized response to stress from a variety of causes. Porphyrins in the discharge
lead to a red color. In light of the dose-related response in males, it appears that the stress in this
study was related to an effect of phenol, either as an irritant or a toxicant.
Hematological and biochemical examinations showed slight but statistically significant
increases in prothrombin time at the low concentration only and in albumin concentration in high-
exposure females; these changes were not considered to be biologically significant. No other
significant changes in hematology or biochemistry were observed. The only statistically significant
changes in organ weights were decreases in liver-to-body, spleen-to-body and spleen-to-brain weight
ratios in mid-concentration (18.9 mg/m3) females. Because the changes in organ weights did not
occur at the highest phenol exposure concentration (96.2 mg/m3), and the same responses occurred in
56
-------
female rats but not in male rats, these responses were considered by the authors to be not
toxicologically significant.
Gross pathological and microscopic examinations of these organs did not exhibit any
differences from those of the controls. Although there were a number of histopathology findings in
the respiratory tract (e.g., inflammatory cells in the nasolacrimal ducts, alveolar macrophages, and
eosinophilic and basophilic material), these findings occurred at similar incidences in the control and
exposed groups. The lesions reported were also those typically seen in control animals. The only
lesion of concern was minimal to slight lung hemorrhage, which was reported in 4/10 control males
and 6/10 high-exposure males at the terminal sacrifice. However, there was no clear concentration-
related increase in incidence or severity, this lesion was not found in the females, and this lesion was
not seen in exposed animals post-recovery or in control animals.
Thus, it appears that the only effect of concern in this study was the red nasal discharge,
which was observed in males but not in females, and this effect was probably due to a nonspecific
response to stress. In addition, no supporting histopathology was observed in a thorough
examination. On the basis of these considerations, the highest concentration (96.2 mg/m3) in this
study was NOAEL. The HEC cannot be definitively determined in the absence of an affected
endpoint. However, assuming that the respiratory tract would be affected first, as shown in other
studies, a conservative NOAEL (HEC) based on a nasal effect would be 2.5 mg/m3. Assuming that
the nasal discharge reflects a nonadverse systemic stress response, the NOAEL(HEC) would be 17
mg/m3. No LOAEL was identified.
Deichmann et al. (1944) conducted subchronic inhalation studies of phenol toxicity in rabbits,
rats, and guinea pigs. Twelve guinea pigs, 6 rabbits, and 12 rats (strain and sex not reported) were
exposed (whole body) in a single exposure chamber to phenol vapor at "a concentration ranging from
0.1 to 0.2 mg/L (100-200 mg/m3)" for 7 hrs/day, 5 days/wk for 6 weeks, 13 weeks, or approximately
11 weeks, respectively. The actual exposure concentration apparently could not be controlled more
precisely. Using the midpoint of 150 mg/m3 as the exposure concentration, the duration-adjusted
concentration was 31 mg/m3.
Among the three tested species, the guinea pig was the most sensitive and the rat was the least
sensitive to phenol exposure. Deaths were observed in 5/12 guinea pigs during the 6-week exposure
period. Other signs of toxicity in the guinea pigs included decreased activity during the first week
and respiratory difficulties and paralysis of hind quarters after 4 weeks of exposure.
Histopathological evaluations revealed lesions of the lungs (pneumonia and bronchitis), heart
(inflammation, fibrosis, and necrosis), liver (fatty changes and necrosis), and kidneys (tubular
57
-------
degeneration and edema). At the end of exposure, the surviving guinea pigs had a concentration of
1.0 mg free phenol/100 mL blood and 0.4 mg conjugated phenol/100 mL blood. (Details on the
analytical procedures used to measure phenol in blood were not reported.).
No deaths or clinical signs of toxicity were observed in the rabbits exposed for 13 weeks, but
lobular pneumonia and fibrosis was observed in these animals. Histopathology lesions in the heart,
kidney, and liver were similar to, but less severe than, those reported in the guinea pigs. After 37
days on study, the rabbits had a concentration of 0.5 mg free phenol/100 mL blood and 0.7 mg
conjugated phenol/100 mL blood; similar concentrations were observed at the end of the exposure
period.
The rats did not show any clinical signs of toxicity during the 74-day exposure period
(approximately 11 weeks), and there was no histopathological evidence of any effect. Blood phenol
levels were not reported for the rats, but an analysis of carcass homogenate found 0.2 mg free phenol
and 0.35 mg conjugated phenol per 100 g carcass homogenate. These levels were reported to be
within the normal range in unexposed rats. This study is limited by the use of only one exposure
concentration, the absence of controls, the inadequate control of exposure levels, and the absence of
reporting of the primary data. However, the results do appear to show that rats are much less
sensitive than rabbits or guinea pigs to the inhalation effects of phenol. However, a comparison of
blood levels suggests that the interspecies differences are due to both toxicokinetic and
toxicodynamic differences.
In an unpublished 90-day study (Sandage, 1961), groups of 10 male rhesus monkeys, 50 male
Sprague-Dawley rats, and 100 male albino mice were exposed to average phenol concentrations of 0
or 4.72 ppm (18.2 mg/m3) continuously for 90 days.5 Exposure was interrupted for 14 hours on day
39 and for 36 hours on days 68-69. The frequency of monitoring of the test atmosphere was not
reported, but the phenol concentration was reported to remain in the desired range of 4.5-5.5 ppm
"after the first three days." No further information on the concentrations during the first 3 days was
reported.
During the exposure, no deaths were observed in the test animals. Body weight gain in mice
was comparable to that in controls but was slightly higher in exposed rats and monkeys. A complete
hematological examination showed no significant changes in the three test species following phenol
5Although this unpublished study is not a primary reference for this assessment, it is presented
here because it contributes some useful information to the overall hazard identification phase of the
phenol assessment.
58
-------
exposure. Blood biochemistry (alkaline phosphatase, cholinesterase, amylase, lipase, and glutamic
oxalacetic transaminase) was evaluated in monkeys only. Urinalysis was apparently conducted in all
species, but kidney function tests (urine volume and specific gravity) were conducted only in
monkeys and rats. The study authors reported that there were no effects on any of these endpoints
but did not provide any supporting data.
At the end of the exposure period, "approximately half of the animals underwent a stress test
in which the animals swam in a smooth-walled tank until exhausted. These animals were sacrificed
immediately after the test, and the other animals were held for a 2-week recovery period prior to
sacrifice. Histopathological evaluations were conducted on only 5-8 organs (including the liver,
kidney, and lung). It appears that all of the monkeys and about half of the rats and mice were
evaluated, although it is not clear whether some of the rodents were evaluated after the recovery
period.
The authors considered the histopathology findings "essentially negative" and did not provide
any description of the observed lesions or the number of animals examined histopathologically.
Liver and kidney pathology was observed in 30% and 20%, respectively, of the monkeys (compared
with 0% of the controls). However, the authors did not consider these changes to be significant, and
they noted that 6/7 reports of pathology in monkeys were considered "minimal or doubtful." Liver
and kidney pathology was also reported in 20% of phenol-exposed rats (compared with 0% of the
controls) and lung pathology was reported in 20% of the phenol-exposed mice (compared with 6% of
the controls). The incidences of liver and kidney pathology in the rat and lung pathology in the
mouse were statistically significant in a Fisher's exact test done for this assessment. Although the
incidence of lung pathology was not affected in monkeys and rats, a relatively high incidence of lung
pathology in the control animals (30% and 65%, respectively) decreased the sensitivity of the
evaluation. No other significant pathological changes were reported in the test animals.
Although the authors concluded that there was no evidence that phenol exposure resulted in
significant damage, there is some indication of liver, kidney, and lung pathology in this study, but the
inadequate reporting precludes the determination of whether there was a treatment-related effect. For
the purposes of this assessment, the single exposure level tested, 18.2 mg/m3, should be considered a
free-standing LO AEL, although it might be considered a minimal LOAEL if additional
histopathology data were available. The LOAEL (HEC) for the kidney and liver lesions is also 18.2
mg/m3. In the absence of additional information on the nature of the lung lesions, the LOAEL (HEC)
for the lung cannot be determined. The study is also limited by the poor control of exposure levels
and limited reporting of effects.
59
-------
In a study published by Dalin and Kristoffersson (1974), seven white rats of an in-house strain
were exposed to phenol at a concentration of 100 mg/m3 continuously for 15 days. There is some
uncertainty about this exposure measure, however, because the exposure chamber was not set up
according to modern designs, and it does not appear that continuous monitoring of exposure levels
was conducted. Unexposed rats (n = 11-12) were used as controls. Nervous system effects were
observed from the first day after the start of exposure. These effects progressed from increased
activity to imbalance, twitches, and disordered walking rhythm on days 3-4. These signs disappeared
by day 5 and were replaced by sluggish behavior until the end of the exposure. A tilting-plane test
was conducted before and after exposure in both groups, and a significant effect was observed on the
exposed rats. There were no significant changes in food intake or water consumption during the
exposure period. Although there was no significant difference in body weight of the exposed group
compared with that of the controls, the average body weight of the exposed group decreased during
exposure, whereas the controls gained weight.
The serum biochemical evaluations showed large, statistically significant increases in SGOT,
SGPT, lactic dehydroganese (LDH), and glutamate dehydrogenase activities, indicating liver damage.
Plasma potassium and magnesium levels were also increased. Although the significance of these
changes is unknown, the authors suggested that the increased magnesium levels may have caused
some of the nervous system effects. Hemoglobin and hematocrit were unaffected. No histopathology
examination was conducted. On the basis of the observed nervous system effects as well as the
serum enzyme changes indicating liver damage, the only exposure concentration was a free-standing
LOAEL. The LOAEL (HEC) is 100 mg/m3, but the actual exposure measurement is of low quality.
4.2.3. Dermal
Phenol is quite irritating, and dermal exposure to liquid phenol or to concentrated phenol
vapor can result in inflammation and necrosis of the skin (Conning and Hayes, 1970; Patrick et al.,
1985; Pullin et al., 1978). As discussed in Section 3.1, phenol is readily absorbed from dermal
contact with phenol liquid or phenol vapor, so systemic effects can also result from dermal exposure.
Several acute lethality assays have been reported. Conning and Hayes (1970) reported a dermal LD50
of 669.4 mg/kg for undiluted phenol applied for 24 hours to the skin of female Alderly Park rats.
Acute dermal toxicity appears to be dependent on the concentration of phenol, with increased
lethality observed with decreased concentration when the same total dose is applied (Deichmann and
Witherup, 1944; Conning and Hayes, 1970). In addition to lethality, renal effects (severe
60
-------
hemoglobinuria and hematin casts in the tubules), cardiovascular effects (cardiac arrhythmias and
ventricular tachycardia), and neurological effects (severe muscle tremors, marked twitching,
generalized convulsions, loss of consciousness, and prostration) were observed at 107.1 mg/kg in
female Alderly Park rats following dermal exposure to undiluted phenol for 24 hours (Conning and
Hayes, 1970). A similar array of effects have been reported in humans following accidental dermal
exposures to large volumes of phenol (ATSDR 1998).
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES
No studies of the reproductive or developmental toxicity of phenol following inhalation
exposure of laboratory animals were located. Several developmental toxicity studies in rats and mice
conducted via the gavage route are available (Argus Research Laboratories, 1997; NTP, 1983a,b;
Narotsky and Kavlock, 1995); the only developmental effect reported in these studies was decreased
fetal body weight. In a two-generation drinking water study in rats (Ryan et al., 2001; available in
unpublished form as IIT Research Institute, 1999), decreased pup weight and decreased survival of
pups pre-culling were observed, but these effects appeared to be secondary to decreased water
consumption.
In the Ryan et al study, 30 Sprague-Dawley rats/sex/group were exposed to 0, 200, 1000 or
5000 ppm phenol in drinking water. Parental (PI) rats were given phenol for 10 weeks prior to
mating, during a 2-week mating period, throughout gestation, lactation, and until sacrifice. The
males were sacrificed after successful mating. All of the PI females were allowed natural parturition
and were sacrificed at Fl weaning. The authors calculated that the average daily phenol intake
during week 10 was 0, 14.7, 70.9, and 301.0 mg/kg-day for PI males and 0, 20.0, 93.0, and 320.5
mg/kg-day for PI females.
For the Fl generation, the average phenol intake during week 10 was 0, 13.5, 69.8, and 319.1
mg/kg-day for males and 0, 20.9, 93.8, and 379.5 mg/kg-day for females. The Fl generation (20
rats/sex/group) was treated following a protocol similar to that used for the PI generation, and F2
pups were sacrificed after weaning, on postnatal day (PND) 22. During treatment, rats were
monitored for mortality, clinical signs, body weight, and food and water consumption. At sacrifice,
the animals were necropsied, and reproductive organs from 20 animals per sex in the control and
high-dose groups from the PI and Fl generations were examined microscopically. In addition, the
spleen, thymus, liver, and kidneys from 10 randomly selected PI and Fl animals of each sex in the
control and high-dose groups were examined.
61
-------
Most of the treatment-related changes in PI rats were observed in the high-dose groups. No
treatment-related mortality was observed in PI rats of either sex. Three high-dose Fl female pups
died shortly after weaning. The deaths appeared to be associated with decreased water consumption
associated with poor palatability, as these pups refused to drink the water containing the phenol. No
other treatment-related mortality was reported in the Fl generation. The only significant observed
clinical sign was redness around the nose fur, which occurred in the high-dose males and females of
the Fl generation before mating and in PI dams during lactation. As noted in the context of the
Hoffman et al. (2001) inhalation study, this redness likely reflects a nonspecific stress response. A
significant decrease in water consumption was observed throughout the study in PI animals of both
sexes (up to 23% for males and up to 39% for females) and was attributed to poor palatability.
Decreased water consumption in the Fl rats at the high dose was of a similar or larger magnitude.
The low water consumption at the high dose was accompanied by decreased body weights as
compared with the controls (9% in PI males and 16% in PI females at sacrifice).
At birth, the high-dose Fl and F2 pups had decreased body weights as compared with the
controls, and the differences were larger by PND 21. The authors noted that pups began drinking the
water prior to weaning and that the decreased water consumption was also observed in the high-dose
pups. Decreased absolute organ weights and increased relative organ weights were observed for a
number of organs at the high dose in both the PI and Fl generations. Most of these changes likely
reflected the lower body weight and overall dehydration in these groups.
Fl females had a statistically significant, dose-related decrease in uterine weights at all doses,
but PI females were not affected. The authors suggested that some of this decrease may have been
related to a lower incidence of uterine dilatation at the high dose. Because the stage in the estrus
cycle can affect uterine weight, estrus cycle staging was also considered as an explanation, but the
authors did not consider the small decreases in the number of rats in estrus (16/24, 15/24, 13/25, 9/22
in the control, low-, mid-, and high-dose groups) sufficient to account for the decreased uterine
weight. Nonetheless, the decreased uterine weight was not considered adverse for several reasons.
There was no evidence of a dose-response relationship for relative uterine weight across the three
dose groups (Table 8). There was no effect on reproductive function and no histopathological
changes in the uterus. Finally, the uterine weight was below the control range for only a few rats in
each dose group, and the control group appeared to have greater variability (particularly at the high
end) than each of the experimental groups.
62
-------
Table 8. Selected Results of Two-Generation Drinking Water Study
(Ryan et al., 2001; IIT Research Institute, 1999)
Endpoint
PI body weight - males week 10
(g)
PI body weight - females week
10 (g)
PI water consumption - males
week 10 (g/day)
PI water consumption - females
week 10 (g/day)
FlAbsolute uterine weight (g)
Uterine weight relative to body
weight
Fl Absolute prostate weight (g)
Prostate weight relative to body
weight
Fl pndO litter weight (g)
F2 pndO litter weight (g)
Preputial separation (age, days)
Body weight at onset (g)
Vaginal patency (age, days)
Body weight at onset (g)
Control
414±39.8
259±24.5
30±4.6
24±3.7
0.81±0.33
0.27±0.11
0.90±0.20
0.18±0.039
6.73±0.42
6.67±0.38
44.0±2.16
229±23.0
34.6±1.85
126±13.7
200 ppm
(14-21 mg/kg-
day)
434±34.4
260±21.8
32±4.6
26±5.5
0.62±0.13*
0.21±0.044*
0.77±0.17*
0.15±0.031
6.85±0.51
6.80±0.44
45.4±2.05
236±23.4
34.5±1.83
127±13.5
1000 ppm
(70-94 mg/kg-
day)
423±40.3
258±19.3
30±3.3
24±4.1
0.60±0.17*
0.20±0.058*
0.74±0.17*
0.15±0.027*
6.63±0.51
6.58±0.42
44.4±2.022
227±19.7
34.1±2.25
121±12.6
5000 ppm
(301-380
mg/kg-day)
382±40.8*
234±21.2*
23±3.1*
15±2.1*
0.53±0.17*
0.20±0.068*
0.76±0.16*
0.18±0.036
6.38±0.27*
6.20±0.48*
47.8±3.13*
195±23.6*
38.3±2.21*
112±7.8*
'Mean ±Standard deviation
2 Standard deviation reported in Ryan et al. (2001) has a typographical error; correct standard deviation obtained from IIT
Research Institute (1999)
* Statistically significant, p 0.05
63
-------
Absolute prostate weights were significantly reduced—by approximately 15%—in all dose
groups of the Fl generation but not the PI generation. Relative prostate weight was decreased in the
low-dose (but not statistically significantly) and mid-dose Fl groups but not at the high dose. In the
absence of a dose-response, the changes in prostate weight were not considered adverse. Dose-
related statistically significant decreases in absolute adrenal and spleen weights were also observed in
the mid- and high-dose Fl males. However, there was no effect on relative weights of these organs in
these groups, there was no effect on these organ weights at any dose in the PI males, and PI females
were only affected at the high dose (where decreased body weight would have played a role).
The pathological examinations showed no treatment-related lesions in the kidneys, spleen,
liver, thymus, or reproductive organs. An immunotoxicity screen conducted with 10 male PI rats per
group found no significant effects on spleen weight, cellularity, or antibody-forming cells for any test
group when compared with the control group; the expected results were found with a positive control
group. The strength of this finding is reduced, however, by the relatively large standard variability in
response (based on the standard error as a percent of the mean).
Complete hematological evaluations (including hematocrit, erythrocyte count, and differential
white cell count) and serum biochemical evaluations were conducted on 8-10 PI males/group prior to
sacrifice. The only significant change in these evaluations was increased blood urea nitrogen (BUN)
in the high-dose group. Because this change was not accompanied by increased creatinine and there
was no associated kidney pathology, the BUN increase was not considered to be biologically
significant. The authors also noted that all but one of the high-dose BUN values were within the
control range.
There was no effect on fecundity or fertility in either generation. In addition, there was no
effect on other indicators of reproductive toxicity, including the frequency of estrus, testicular sperm
count, sperm motility, and sperm morphology. The survival of the high-dose Fl pups was
significantly decreased on PND 4 (pre-culling), although there was no effect on overall Fl pup
survival. In the F2 generation, high-dose pup survival was significantly decreased throughout the
lactation period. This decreased survival of both generations of pups was likely secondary to the
decreased maternal water intake and associated decreases in milk production. In the Fl generation,
delayed vaginal patency and delayed preputial separation were observed at the high dose. The delay
was considered secondary to decreased fetal growth at the high dose resulting from decreased water
consumption due to poor palatability and associated decreased food consumption.
Thus, all of the adverse systemic and reproductive effects of phenol in this study occurred at
the high dose, and they appear to be secondary to decreased water consumption due to poor
64
-------
palatability rather than to a toxic effect of phenol. On the basis of the decreased parental and pup
body weight (compared to the controls) and decreased pup survival, the high dose is a LOAEL. The
study NOAEL is 70.9 mg/kg-day (based on the NOAEL corresponding to the lowest LOAEL in this
study, in PI males). BMD modeling was not conducted for this study because the observed effects
appeared to be secondary to decreased water consumption and not reflective of phenol toxicity.
In an unpublished developmental toxicity study conducted according to GLP guidelines
(Argus Research Laboratories, 1997), pregnant CrhCDRBR VAF/Plus Sprague-Dawley rats (25 per
group) received phenol by oral gavage on GDs 6 through 15.6 Dosing was three times daily with 0,
20, 40, or 120 mg phenol/kg/dosage using a dosing volume of 10 mL/kg. The corresponding daily
doses were 0, 60, 120, and 360 mg/kg-day. The authors noted that the test material was 90% phenol
United States Pharmacopeia (USP); the authors adjusted the dosage calculations for test material
purity.
The exposed dams were observed twice a day for viability and daily for clinical signs,
abortions, and premature deliveries. In addition, the maternal body weights were recorded every day,
and food consumption was also recorded periodically (every 1-2 days). The rats were sacrificed on
GD 20, and gross necropsy of the thoracic, abdominal, and pelvic viscera was performed. The
number of corpora lutea in each ovary was recorded. The uterus of each rat was excised and
examined for number and distribution of implantations, live and dead fetuses, and early and late
resorptions. Each fetus was weighed, sexed, and examined for gross external alterations. One-half of
the fetuses were examined for soft tissue alterations, and the rest were examined for skeletal
alterations.
One high-dose dam died on GD 11. The study authors attributed this death to phenol
treatment because it occurred only at the high dose, although there were no adverse clinical
observations and no abnormal necropsy findings in this animal. Other high-dose animals exhibited
excess salivation and tachypnea (rapid breathing). There were no other treatment-related clinical
observations and no treatment-related necropsy findings. Dose-dependent decreases in body weight
of the exposed animals as compared with the controls were observed. Statistically significant
decreases in both maternal body weight (8%) and body weight gain (38% for GD 6-16) were
observed at the high dose; although a statistically significant decrease in body weight gain (11%) was
6This study has not been peer-reviewed, but it was conducted (with minor deviations) according
to EPA guidelines for developmental toxicity studies, it is well documented, and it contributes useful
information to the hazard identification and dose-response portions of the assessment. The study was
designed to meet the U.S. EPA Pesticide Assessment Guidelines, Subdivision F, 83-3.
65
-------
observed at the mid dose, the decrease at the mid dose (relative to controls) in absolute maternal
weight at the end of dosing (3%) was not statistically significant. Dose-dependent decreases in food
consumption were also observed during the dosing period (see Table 9).
Fetal body weights in the high-dose group were significantly lower than those of the controls,
by 5-7%. The high-dose group had a statistically significant decrease in ossification sites on the
hindlimb metatarsals, but it is unlikely that this small change is biologically significant. The
incidence of litters with incompletely ossified or unossified sternal centra was 0/23, 0/25, 3/23, and
3/24; this increase was not statistically significant (Table 9). There were small, dose-related
increases in the number of litters with fetuses with "any alteration" and with "any variation" at 120
mg/kg/day and higher. However, neither of these changes was statistically significant, and the
response was not clearly dose-related. In addition, an increase in total variations is of questionable
significance in the absence of any increase in individual variations. There were no other treatment-
related effects on uterine contents, malformations, or variations.
The maternal NOAEL was 60 mg/kg-day, based on small decreases in maternal body weight
gain at 120 mg/kg-day, and the developmental NOAEL was 120 mg/kg-day, based on decreased fetal
body weight and delayed ossification at 360 mg/kg-day. Benchmark dose (BMD) modeling was also
conducted for the decreased maternal weight. Defining the BMR as a one standard deviation
decrease in maternal body weight gain, the BMDL was 93 mg/kg-day. Details on the BMD modeling
are provided in Appendix B.
In a well-designed developmental toxicity study (NTP, 1983a), timed-mated CD rats were
administered phenol by gavage at 0, 30, 60, or 120 mg/kg-day in 5 mL/kg distilled water on GDs 6 to
15 and sacrificed on GD 20. Females were weighed on GDs 0, 6-15 (prior to daily dosing), and 20
(immediately following sacrifice), and they were also observed during treatment for clinical signs of
toxicity. A total of 20-22 females per group were confirmed to be pregnant at sacrifice on GD 20.
The dams were evaluated at sacrifice for body weight, liver weight, gravid uterine weight, and status
of uterine implantation sites. Live fetuses were weighed, sexed, and examined for gross
morphological abnormalities and malformations in the viscera and skeleton.
The results of this study did not show any dose-related signs of maternal toxicity or any
clinical symptoms of toxicity related to phenol treatment. The number of implantation sites was
slightly higher in the dosed groups, but this change could not be treatment related because
implantations in this strain
66
-------
Table 9. Key Results in Argus Research Laboratories, (1997) Rat Developmental Toxicity
Study
Maternal body weight
GD16(g)
Maternal body weight gain
(GD6-16)(g)
Maternal food
consumption (GD 6- 1 6)
(g/day)
Fetal body weight - all
fetuses (g)
Fetal body weight - males
(g)
Fetal body weight -
females (g)
% Fetuses with any
alteration/litter
% Fetuses with any
variation/litter
Incompletely ossified or
unossified sternal centra
litter incidence
Ossification sites per fetus
per litter, hindlimb
metatarsals
Control
363.9±22.2
64.0±10.7
26.7±2.7
3.62±0.30
3.71±0.30
3.53±0.30
1.0±2.6
1.0±2.6
0
4.00±0.00
60 mg/kg-day
359.0±18.7
58.0±9.4
26.0±2.6
3.63±0.24
3.73±0.25
3.53±0.25
2.4±4.6
2.4±4.6
0
4.00±0.00
120mg/kg/day
354.3±17.5
56.8±10.8*
24.8±1.9**
3.60±0.30
3.71±0.31
3.49±0.28
3.7±4.4
3.4±4.4
37337
4.00±0.00
360 mg/kg-day
334.2±20.1**
39.8±9.5**
21.9±2.1**
3.41±0.35*
3.53±0.34
3.29±0.35**
4.1±7.8
3.8±7.3
37338
3.98±0.07*
'Mean ±Standard deviation
* Statistically significant, p 0.05
**Statistically significant, p 0.01
67
-------
take place prior to GD 6 (prior to dosing). Significant increases in the litters with nonlive (dead plus
resorbed) pups were observed in the low- and mid-dose groups but not in the high-dose group (Table
10). Because this response was not dose dependent, and the response in the high-dose group was
comparable to that in the control group, this observation is not considered treatment-related. In
addition, there was no effect on the more appropriate measure of nonlive pups per litter.
There was also no effect on live fetuses, sex ratio, malformations, or variations, but there was
a clear dose-related downward trend in fetal body weight, although the changes at the two lower
doses were small and the effect was statistically significant only at the high dose (Table 10). Fetal
body weights in the high-dose group were 93% of the average of those in the control group; fetal
body weights were not reported separately for males and females. Historical control data from the
supplier report the average fetal body weight in this strain as being well below the weight in the high-
dose group (Charles River Laboratories, 1988). (Concurrent control weight was 4.14 g, high-dose
weight was 3.84 g, and historical control weight was 3.39 g.) The litter size in the high-dose group
was also somewhat higher (but not statistically significantly) than in the controls, possibly
contributing to the smaller fetal weight at the high dose.
As shown in Table 10, the total pup burden (total fetal weight) and the gravid uterine weight
were highest in the low-dose group and then in the high-dose group; both values higher than those in
the control group. In addition, the treatment-period maternal weight gain was very similar in the
control and high-dose groups (but higher in the low-dose group), but the absolute maternal weight
gain (i.e., adjusted for the gravid uterine weight) was much lower in the high-dose group than in the
controls. The results from the low-dose group suggest that the dams could have borne a somewhat
higher burden of the total in utero package. However, the results also suggest that the dams were
near the limit of what they could carry, considering the lower absolute weight gain but unaffected
treatment-period weight gain in the high-dose group. No dose-related signs of maternal toxicity and
no clinical symptoms of toxicity related to phenol treatment were observed in this study.
From these considerations and the potential for the decreased fetal weight to reflect primarily
the larger litter size, the decreased fetal weight in this study could be considered an equivocal
LOAEL. Thus, on the basis of decreased fetal body weight, the mid dose in this study of 60 mg/kg-
day was a NOAEL for developmental toxicity and the high dose of 120 mg/kg-day was an equivocal
LOAEL. The high dose was a maternal NOAEL. BMD modeling could not be done for the
decreased fetal weight because NTP did not have information on the fetal weight by sex,
68
-------
Table 10. Key Results from Developmental Toxicity Study in Rats
Administered Phenol by Gavage (NTP, 1983a)
Live fetuses/litter1
No. litters with nonlive (%)
Nonlive/liver
Control
12.23±0.51
0
(0%)
0
30 mg/kg-day
13.32±0.51
5
(22.7%)
0.41±0.18
60 mg/kg-day
12.14±0.56
7
(31.8%)
0.32±0.10
120 mg/kg-day
13.75±0.62
3
(15%)
0.25±0.14
Historical control data for this strain on nonlive/litter (Implants/pregnant female minus live fetuses/pregnant female) (Charles River
Laboratories, 2001): 0.6-0.8, based on 62 studies.
Males/litter
Average fetal body weight per litter (g)
Historical control fetal weight (from Charles
River Laboratories, 1988)
Total ave. fetal weight/dam (g)
(calculated)
Gravid uterine weight (g)
Treatment period maternal weight gain (g)
Absolute maternal weight gain (adjusted for
gravid uterine weight) (g)
6.23±0.38
4.14±0.07
Mean 3.39 g
Range 3.04-
3.52 g
50.6
76.9±3.0
41.0±1.1
58.0±3.1
7.36±0.52
4.10±0.05
54.6
82.9±3.4
47.2±1.7**
58.4±2.2
5.14±0.56
4.03±0.07
48.9
75.4±3.2
40.2±1.6
52.7±2.4
7.35±0.57
3.84±0.05**
52.8
81.7±3.7
41.2±2.9
51.8±3.2
'Mean ±Standard error of the mean.
**Statistically significant, p<0.01
69
-------
either in the report or in its archives.7 Data on fetal weight by sex is needed for meaningful modeling
because the average weight of males and females is different and the number of males per group
varied.
The preliminary rat developmental toxicity studies (NTP, 1983a) found that phenol toxicity is
increased by the use of small dosing volumes. For example, when phenol was administered by
gavage on GDs 6-15 to pregnant CD rats at doses of 0, 125, 160, 200, or 250 mg/kg-day in a volume
of 1 mL/kg, the mortality was 0% (0/9), 70% (7/10), 78% (7/9), 100% (9/9), and 100% (9/9),
respectively. The deaths were preceded by dose-related signs of toxicity, including tremors,
convulsion, and respiratory distress; mottled liver and congested lungs were found on necropsy. In
contrast, when the same doses were administered in a volume of 5-7.5 mL/kg, the respective
mortality was only 0% (0/24), 0% (0/5), 17% (1/6), 17% (1/6), and 71% (5/7), respectively. On the
basis of these results, a volume of 5 mL/kg was used in the main developmental toxicity study.
In preliminary toxicity studies conducted with doses of 60-250 mg/kg-day in a volume of
5-7.5 mL/kg, decreased maternal body weight gain (or body weight loss) during dosing was observed
at 160 mg/kg-day and up, doses at which mortality was also observed. In addition, tremors were
observed sporadically in the phenol-dosed groups, without any clear dose-response. There were no
treatment-related changes in prenatal viability, fetal sex ratio, or fetal structural development.
The study authors stated that when results of all of the preliminary studies were pooled, a
statistically significant trend of decreasing fetal weight was observed, but there were no statistically
significant differences from controls in pairwise analyses. The power of the pairwise tests was
limited because only 4-6 litters were produced in the dose range 100-200 mg/kg-day.
In a standard mouse developmental toxicity study (NTP, 1983b), phenol was administered by
gavage in water at 0, 70, 140, or 280 mg/kg-day on GDs 6 to 15 in a volume of 10 mL/kg. Groups of
31-36 plug-positive female CD-I mice were used in each treatment group. The pregnancy rate in the
controls was only 83%; the pregnancy rate in dosed animals ranged from approximately 83% in the
low- and mid-dose groups to 71% at the high dose. In addition, 4/36 high-dose mice died; no deaths
occurred in any other groups. The average maternal body weight gain during treatment was
statistically significantly reduced at the high dose, as was the maternal body weight at terminal
sacrifice on GD 17 (by 10%, compared with the control group). In addition, tremors were observed
at the high dose throughout the dosing period. As in the rat study, there was a highly statistically
significant decrease in fetal body weight per litter (18%) at the high dose. An increased incidence of
cleft palate was also reported at the highest dose level, although the incidence was not significantly
7Michael Shelby, NTP, personal communication to Lynne Haber (TERA), March 13, 2002.
70
-------
different from that of the other groups and there was no statistically significant increase in the
incidence of litters with malformations. There was no other evidence of altered prenatal viability or
structural development.
Thus, the high dose of 280 mg/kg-day was a maternal frank effect level based on the observed
deaths; tremors and decreased body weight also occurred at this dose. The high dose was also a
developmental LOAEL based on decreased fetal body weight (accompanied by a possible increase in
the incidence of cleft palate) in the fetuses, an effect that was likely secondary to the severe toxicity
in the dams. The study NOAEL for maternal and developmental toxicity was 140 mg/kg-day.
The series of oral screening studies mentioned above (Narotsky and Kavlock, 1995; Berman
et al., 1995; Moser et al., 1995; MacPhail et al., 1995) also included a developmental toxicity
screening study in which groups of pregnant F344 rats (15-20 animals/group) were given phenol at
doses of 0, 40, or 53.3 mg/kg by gavage once daily on GDs 6-19 and then sacrificed on postnatal day
(PND) 6 (Narotsky and Kavlock, 1995). The dosing volume was 1 mL/kg. Pups in each litter were
examined and counted on PND 1,3, and 6 and were weighed on PND 1 and 6. Uterine implantation
sites were counted after the dams were sacrificed. Only minimal quantitative data were presented.
No maternal deaths were observed. The authors reported that phenol treatment altered respiration
(rales and dyspnea) at both dose levels, but no quantitative data were presented. Decreased (but not
statistically significant) maternal body weight (compared with the controls) and decreased
(statistically significant) maternal body weight gain were also reported at both doses, but there was
no clear dose-response.
No statistically significant evidence of developmental toxicity due to phenol exposure was
observed. The only evidence of developmental toxicity came from dams that exhibited severe
respiratory signs. These signs of developmental toxicity included a dose-dependent increase in full-
litter resorptions (one at the low dose and two at the high dose) and an excessive incidence of
perinatal mortality and reduced pup weights on PND 1 in one litter at the high dose. However, these
changes as a group were not significantly different from those in the controls. Nonetheless, the
respiratory effects from oral dosing reported in this study are of interest, particularly as they were not
reported in the related study of systemic toxicity (Moser et al., 1995). This difference may reflect
differences in the completeness of the study reporting. Alternatively, it may suggest that pregnant
females may be more sensitive than nonpregnant females to the toxic effects of phenol.
71
-------
4.4. OTHER STUDIES
4.4.1 Initiation/PromotionStudies, Other Short-Term Tumorigenicity Assays, and Cancer
Mechanism Studies
Several studies have tested the promotion potential of dermally administered phenol. These
studies found that phenol promotes tumors initiated with dimethyl=benzanthracene (DMBA), but the
phenol doses tested caused ulceration (Salaman and Glendenning, 1957; Boutwell and Bosch, 1959)
and thus were well above the MTD. One study (Wynder and Hoffman, 1961) found promotion of
DMBA-initiated tumors by dermally administered phenol at a concentration that caused "no toxic
reactions."
Salaman and Glendenning (1957) conducted an initiation/promotion study in which groups of
20 male "S" strain mice were initiated with a single dermal treatment with DMBA and promoted with
dermal treatment with 0.5 mg/mouse phenol in acetone using two different treatment concentrations
for the same applied dose. The phenol was applied beginning 3 weeks after the DMBA application
for either 24 weeks in a volume of 0.025 mL as a 20% solution or for 32 weeks in a volume of 0.1
mL as a 5% solution (rotating the weekly applications among four application sites for both
concentrations). The study did not report whether the application site(s) were covered or whether the
animals were restrained from licking the site.
The high concentration produced local ulceration that healed just in time for the next
treatment 4 weeks later, whereas the low concentration produced only transient light crusting that
tended to decrease as the experiment progressed. It is unclear how severe the skin effects would have
been if the low concentration had been repeatedly applied to the same site rather than being rotated
among four sites. Tumors were observed in both treatment groups, with a shorter time to first tumor
and a higher tumor burden in the group treated with the higher phenol concentration. A few
histologically confirmed malignant tumors (primarily squamous epitheliomas) were observed in both
groups. In mice that underwent the same phenol treatment but were not pretreated with DMBA,
seven papillomas were observed at the high concentration. No tumors were observed at the low
concentration, even though the weekly dose was the same and the total dose per mouse was higher
(because the duration was longer).
This study had no control group on DMBA-only group, but the absence of tumors at the low
concentration indicates that the observed tumors were phenol related. The authors noted that the
observed tumors might have been related to the significant skin injury produced by phenol. This
suggestion is supported by the strong effect of the concentration of applied phenol at the same total
dose.
72
-------
Boutwell and Bosch (1959) conducted a series of initiation/promotion studies with different
strains of mice. The mice were pretreated with a single application of 75 |o,g DMBA in benzene
followed by 5% or 10% phenol (1.25 or 2.5 mg per application) in benzene or with either dose of
phenol alone twice weekly for 52 to 72 weeks. An additional group received DMBA alone,
apparently followed by benzene vehicle, although there is some inconsistency between the text and
the summary tables regarding whether the control group received the benzene vehicle. At the high
phenol dose, dermal treatment with phenol resulted in decreased body weight (compared with the
controls) and decreased survival. Skin wounds, hair loss, and reparative hyperplasia were also seen
at the high dose, with the wounds predominantly seen in the first 6 weeks of treatment. By contrast,
the authors stated that there was no evidence of ill effects of 5% phenol except for its promoting
activity. This statement was based on external observation; no histopathology was conducted.
A dose-related increase in papillomas and in carcinomas was observed in the groups initiated
with DMBA and promoted with phenol. Increased papillomas were also observed in one strain
treated with the high dose of phenol alone. There was evidence of decreased activity when phenol
was further purified, indicating that the activity was not due to a contaminant. Because the benzene
vehicle is a defatting agent, it is unclear whether it would have contributed to the effect of phenol.
Wynder and Hoffmann (1961) also found that dermally applied phenol is a promoter. Female
Swiss mice (28-30/group) were initiated with a single application of DMBA followed by treatment
with 5% phenol three times weekly or 10% phenol two or three times weekly. The dilution vehicle
was not reported. "No toxic reactions" were reported at 5% phenol, although the higher
concentration was reported to be "rather toxic." Treatment was for 12 months, and the mice were
observed for an additional 3 months; the percentage of animals with papillomas and with cancers was
recorded monthly.
At 10 months, papillomas were seen in 33% of the low-dose group and > 80% of the high-
dose group; cancer was seen in 3% of the low-dose animals and 30-60% of the high-dose groups. By
contrast, there were no papillomas or cancers in female Swiss mice treated with phenol alone and
only 10% papillomas (no cancer until week 12, and only 7% of the animals had cancer at study
termination) in the mice treated with DMBA alone. Survival decreased markedly after week 10 in
the high-dose groups but not the other groups. In another experiment, the onset of tumor formation
was much earlier in mice treated with 0.005% benzo[a]pyrene three times weekly plus 5% phenol
twice weekly than in mice treated with benzo[a]pyrene alone. Papillomas were observed by the
second week in the groups receiving benzo[a]pyrene and phenol and were present in at least 33% of
the animals by week 5, compared with 3% of the mice at week 5 in the benzo[a]pyrene-only group.
73
-------
In a short-term assay, Stenius et al. (1989) found that phenol did not increase the production
of gamma-glutamyl transpeptidase (GGT)-positive foci. Groups of 7-9 male Sprague-Dawley rats
were partially hepatectomized and treated with diethylnitrosamine and then with 100 mg/kg phenol
by gavage for 5 days (gavage volume not reported). Phenobarbital, the positive control, produced a
marked increase in GGT-positive foci. This assay was based on the assumption that GGT-positive
cells in enzyme-altered foci represent initiated cells and the observation that these cells are often
resistant to toxic effects.
A decrease in tumor formation was seen in a co-carcinogenesis study of phenol and
benzo[a]pyrene (Van Duuren and Goldschmidt, 1976; Van Duuren et al., 1971, 1973). Phenol was
applied at a dose of 3 mg/application in acetone to the clipped skin of female ICR/Ha Swiss mice
(50/group) three times a week for 1 year simultaneously with 5 [ig of benzo[a]pyrene. The resulting
number of tumors (both papillomas and squamous cell carcinomas) was markedly lower than in the
animals receiving the benzo[a]pyrene alone. Phenol alone did not cause skin tumors. Neither the
application volume nor the application surface area were reported, and no information was provided
on any skin effects other than tumors.
In a test of a (TG • AC) transgenic mouse line carrying a v-Ha-ras gene fused to a £ globin
promoter, Spalding et al. (1993) found that phenol did not produce papillomas. This strain has
genetically initiated skin and has been shown to be sensitive to the known promoter 12-0-
tetradecanoylphorbol-13-acetate (TPA). Five male mice were dermally treated twice per week for 20
weeks with 3 mg phenol. Only one papilloma was observed; in contrast, strongly promoting agents
produced five or more papillomas per mouse. This dose did, however, cause chronic irritation and
hair loss.
A number of studies have investigated the reason why benzene is carcinojenic but orally
administered phenol is not, in light of the fact that when phenol and many of its metabolites are
significant products of benzene metabolism. Medinsky et al. (1995) noted that, on the basis of the
urinary metabolite profile, higher levels of hydroquinone are produced after benzene exposure than
after exposure to comparable doses of phenol. The potential production of other toxic metabolites,
such as muconaldehyde, following benzene exposure but not phenol exposure was also noted. The
authors explained the different metabolite profiles of phenol and benzene using the zonal distribution
of metabolizing enzymes in the liver. As described in Section 3.3, phenol is conjugated in the gut
and in zone 1 of the liver. This reduces the amount of phenol that reaches zone 3 of the liver, where
oxidative activity is highest, and so decreases hydroquinone production. By contrast, conjugation of
benzene in the gut and zone 1 is low, because benzene must be oxidized prior to conjugation. This
74
-------
results in more free phenol reaching zone 3 of the liver following benzene exposure than after phenol
exposure, and hence more production of hydroquinone.
Equivocal or negative results were obtained with phenol in a well-conducted and well-
controlled interlaboratory study evaluating the usefulness of the Chinese hamster V79 cell metabolic
cooperation assay for detecting tumor promoters (Bohrman et al., 1988). The study authors noted,
however, that the assay was conducted in the absence of exogenous metabolic activation, and V79
cells have low intrinsic metabolic capacity.
Miyagawa et al. (1995) conducted a validation test in male B6C3F1 mice of the in vivo-in
vitro replicative DNA synthesis test. The test was based on the hypothesis that nongenotoxic
carcinogens are likely to increase cell proliferation. Phenol was negative in this assay, which was
conducted at 0, 300, and 600 mg/kg administered via oral gavage.
4.4.2. Genotoxicity
The genotoxic potential of phenol appears to depend on the competing processes of activation
to a genotoxic form and metabolic inactivation (e.g., via conjugation). Phenol tended to be negative
in bacterial gene mutation assays (Pool and Lin, 1982; Rapson et al., 1980; Haworth et al., 1983) but
was positive or equivocal in mammalian cell gene mutation assays (McGregor et al., 1988a,b;
Paschin and Bahitova, 1982; Tsutsui et al., 1997) (Table 11). Phenol tended to induce micronuclei in
mice when administered intraperitoneally (Marrazzini et al., 1994; Chen and Eastmond,1995a;
Ciranni et al., 1988b) but was negative (or positive only at very high doses) when administered orally
(Ciranni et al., 1988b; Gocke et al., 1981). This difference is likely due to the first-pass conjugation
and inactivation of orally administered phenol. Phenol was also positive in in vitro micronucleus
tests with human lymphocytes (Yager et al., 1990) and Chinese hamster ovary (CHO) cells (Miller et
al., 1995) and caused chromosome aberrations in the presence of S9 activation in CHO cells (Ivett et
al., 1989). Results from DNA damage assays are inconsistent, but they tend to show that phenol can
cause sister chromatid exchanges (Erexson et al., 1985; Ivett et al., 1989) or cell transformation
(Tsutsui et al., 1997) if it is not metabolically inactivated.
Phenol was negative in a well-conducted assay Salmonella typhimurium reverse mutation
assay performed with up to cytotoxic doses in the presence and absence of varying concentrations of
S9 activation with strains TA1535, TA1537, TA1538, TA98, and TA100 (Pool and Lin, 1982).
Phenol was tested in two independent laboratories as part of a large-scale test by NTP in salmonella
strains TA1535, TA1537, TA98, and TA100 in the presence and absence of S9 activation (Haworth
et al., 1983). Both laboratories found that phenol was negative. Rapson et al. (1980) also reported
75
-------
that phenol was negative in a test in strain TA100, although no primary data were presented and it
was unclear whether sufficiently high doses were tested. A weak positive response was reported with
phenol in strain TA98 in the presence of S9 but not in the absence of S9 (Gocke et al., 1981). Other
strains were also tested in that assay, but the results were unclear.
Positive or equivocal results have been reported in mammalian cell gene mutation assays.
McGregor et al. (1988a, b) reported on a well-conducted mouse lymphoma L5178Y tk+/tk- assay of
phenol performed as part of a test of 72 coded chemicals. In the absence of S9, the results were
considered questionable or inconclusive in two independent assays because of the absence of a dose-
related trend and increases occurring only in the presence of high cytotoxicity. In the presence of S9,
the first test was questionable (no dose-related trend but statistically significant results at several
doses), but a clear positive result was obtained in the confirmatory test. Overall, the study authors
concluded that no definitive conclusion was possible.
76
-------
Table 11. Summary of Genotoxicity Studies
Endpoint
Assay system
Results (wo/w
Activation)
Comments
Reference
In Vitro Studies
Gene mutation-
bacteria
Gene mutation -
mammalian cell
Clastogenicity
Chromosome
aberration
Salmonella
typhimurium
TA1535, TA1537,
TA1538,TA98,
TA100
Salmonella
typhimurium
TA1535, TA1537,
TA98, TA100
Salmonella
typhimurium
TA100
Salmonella
typhimurium
TA98
Mouse lymphoma
L51 78 Y cells
Chinese hamster
V79 cells
Syrian hamster
embryo (SHE) cells
Micronuclei in
human lymphocytes
Micronuclei in CHO
cells
CHO cells
-/-
-/-
-1
-/w2
?/?2
NT2/+
NT/+
+/NT
+/+
-/+
Tested to cytotoxic
doses, varying S9
concentrations
Part of NTP testing;
tested in 2
laboratories
Unclear if
sufficiently high
doses tested
Other strains also
tested, but results
with them unclear
Two independent
assays conducted +/-
S9
S9 from
phenobarbital-
induced mice
None
None
S9 from
phenobarbital/beta-
naphthofiavone
induced rats
Part of NTP testing
Pool and Lin,
1982
Haworth et al.
1983
Rapson et al.
1980
Gocke et al.
1981
McGregor et
al. 1988a,
1988b
Paschin and
Bahitova 1982
Tsutsui et al.
1997
Yager et al.
1990
Miller et al.
1995
Ivett et al.
1989
77
-------
Endpoint
DNA damage
Cell transformation
Assay system
Sister chromatid
exchange, human
lymphocytes
Sister chromatid
exchange, human
lymphocytes
Sister chromatid
exchange, human
lymphocytes
Sister chromatid
exchange, CHO cells
Unscheduled DNA
synthesis, SHE cells
Single strand breaks
mouse lymphoma
cells
Single strand breaks
CHO cells
Syrian hamster
embryo (SHE) cells
Results (wo/w
Activation)
-/NT
+/NT
+/NT
+/W
+/NT
"
-
+/NT
Comments
Unclear if
sufficiently high
doses tested
Small increases
None
Part of NTP testing
None
None
None
None
Reference
Jansson et al.
1986
Morimoto et
al. 1983;
Morimoto and
Wolff 1980
Erexson et al.
1985
Ivett et al.
1989
Tsutsui et al.
1997
Pellack-
Walker and
Blumer, 1986
Szeetal. 1996
Tsutsui et al.
1997
In Vivo Studies
Gene mutation
Clastogenicity
Drosophila sex-
linked recessive
lethal
Drosophila sex-
linked recessive
lethal
Drosophila sex-
linked recessive
lethal
Mouse micronucleus
i.p.
"
~
+
None
None
None
None
Gocke et al.
1981
Sturtevant
1952
Woodruff et
al. 1985
Marazzini et
al. 1994
78
-------
Endpoint
Chromosome
aberration
DNA damage
Assay system
Mouse micronucleus
i.p.
Mouse micronucleus
i.p.
Mouse micronucleus
i.p.
Mouse micronucleus
oral
Mouse micronucleus
i.p.
Mouse micronucleus
oral
Mouse micronucleus
oral
Mouse,
spermatogonia and
spermatocytes
Single strand breaks,
testicular cells, i.p.
Results (wo/w
Activation)
+
-
w
+
-
+
+
-
Comments
Weak response
No positive control,
unclear if
sufficiently high
doses tested
Small sample size
None
Clear effect at same
dose as oral study
Number tested not
reported
Pregnant females on
GD13
Inconsistencies in
reporting
None
Reference
Chen and
Eastmond
1995a
Barale et al.
1990
Gocke et al.
1981
Ciranni et al.
1988b
Ciranni et al.
1988b
Gad el-Karim
etal. 1985
Ciranni et al.
1988a
Bulsiewicz
1977
Skare and
Schrotel 1984
1 Apparently in the absence of S9 - the presence of absence of S9 was not addressed.
2w = weak positive response; ? = questionable or inconclusive; NT = not tested
79
-------
Other authors have also reported positive results in mammalian gene mutation assays.
Paschin and Bahitova (1982) found that phenol was mutagenic in an in vitro assay for mutagenicity
at the HGPRT locus of V79 Chinese hamster cells in the presence of S9 from the livers of
phenobarbital-induced mice. Tsutsui et al. (1997) also reported that phenol induced a dose-related
increase in mutation frequency in Syrian hamster embryo cells.
In contrast with these positive results in mammalian cells, a number of authors (Gocke et al.,
1981; Sturtevant, 1952; Woodruff et al., 1985) found that phenol was negative in sex-linked
recessive lethal assays conducted in drosophila using the feeding and injection routes.
The potential for phenol to induce micronuclei (a measure of clastogenicity) appears to be
related to the route of dosing, with generally positive results when phenol is administered
intraperitoneally but negative or equivocal results when it is administered orally. This route-related
difference is likely due to the potential for first-pass detoxification of phenol when it is administered
via the oral route but not when administered intraperitoneally. Several authors have suggested that
the chromosomal effects of phenol result from phenol interactions with the spindle apparatus
(Bulsiewicz, 1977; Yager et al., 1990). No in vitro studies of phenol clastogenesis were located.
Phenol was positive in the micronucleus test in male mice at an i.p. dose of 120 mg/kg
(Marrazzini et al., 1994). Similarly, Chen and Eastmond (1995a) reported a weak increase in bone
marrow micronuclei of male CD-I mice treated with up to 160 mg/kg-day phenol intraperitoneally
for 3 days. By contrast, Barale et al. (1990) found that phenol was negative when administered at
i.p. doses of up to 160 mg/kg to male CD-I mice. However, it is unclear whether sufficiently high
doses were tested in that study, since no cytotoxicity and no clinical signs of toxicity were reported;
also, the study did not include a positive control. Gocke et al. (1981) also found that phenol was
negative in male and female NMRI mice treated with i.p. phenol at two daily doses of up to 188
mg/kg per dose, although the sample size of tested animals was small.
Ciranni et al. (1988b) found that an oral dose of 265 mg/kg phenol caused a slight increase in
micronuclei and some myelotoxicity (decreased ratio of polychromatic to normochromatic
erythrocytes, PCE/NCE ratio), but i.p. administration of the same dose caused clear increases in
micronuclei and stronger myelotoxicity. Gad-el-karim et al. (1985) reported that a single oral dose
of phenol (250 mg/kg) was negative for micronucleus formation in male CD mice, but they did not
report the number of animals tested. This difference between the effects of i.p. and oral
administration of phenol is also consistent with the metabolic effects of first-pass metabolism
mentioned in Section 3.3.
80
-------
Ciranni et al. (1988a) found that a single gavage dose of 265 mg/kg phenol caused a small
but statistically significant increase in bone marrow micronuclei—accompanied by cytotoxicity—in
pregnant female CD-I mice treated on GD 13. There was no effect on fetal liver. Although no
positive control was included, benzene did cause micronuclei in fetal liver, confirming the
sensitivity of the assay.
Phenol was positive in an in vitro assay for the development of micronuclei in human
lymphocytes in the absence of exogenous metabolic activation, although the dose-response was
weak (Yager et al., 1990). Miller et al. (1995) also found that phenol was positive in an in vitro
micronucleus test in CHO cells in the presence or absence of S9 from livers of phenobarbital/beta-
naphthoflavone induced rats, although a stronger response was observed in the presence of S9.
Phenol was evaluated in the chromosome aberration assay in CHO cells as part of a series of
tests by NTP to evaluate genotoxicity assays (Ivett et al., 1989). No significant increase was
observed in the absence of S9 activation. In the presence of S9, significant increases in both simple
and complex aberrations were observed. A delayed harvest time was used due to cell cycle delay.
In a five-generation study of chromosome aberrations in spermatogonia and spermatocytes in
Porton strain inbred mice, Bulsiewicz (1977) observed dose-related increases in aberrations that
tended to increase with successive generations. Polyploidy was also observed. Three dose groups
and a control were treated by oral gavage. The dosing volume was reported as "2 mL of a solution
of phenol" (sic) for the low-dose group; volumes were not reported for the other groups. Phenol was
reported as being administered in a concentration of 0, 0.08, or 0.8 mg/L per day, or "8 g per liter"
(sic).
A number of studies reported synergistic effects between phenol and hydroquinone in the
micronucleus assay in mice (Marrazzini et al., 1994; Barale et al., 1990; Chen and Eastmond,
1995a). At least part of this interaction is likely due to phenol enhancing the peroxidase-dependent
metabolic activation of hydroquinone.
A number of studies have evaluated the potential for phenol to cause DNA damage. These
studies tend to show that phenol can produce effects when it is metabolized to an active form, but
that inactivation is likely to predominate over activation following oral dosing.
Jansson et al. (1986) found no effect on sister chromatid exchanges (SCEs) in an in vitro
assay with human lymphocytes, although it was unclear whether sufficiently high doses were tested.
Small but statistically significant increases in SCEs in cultured human lymphocytes were reported by
Morimoto et al. (1983) and Morimoto and Wolff (1980). By contrast, Erexson et al. (1985) found a
81
-------
dose-related increase in SCEs in human lymphocytes exposed to phenol in vitro. They attributed the
difference between their results and negative results in other studies to differences in the procedure
used. In particular, Erexson and colleagues used mitogenic stimulation of the lymphocytes 24 hours
prior to the phenol exposure. This mean that the cells were blast-transformed and exposed in the Gl-
S phase (and so there was less opportunity for repair prior to DNA replication), and cytochrome P450
activity was elevated as a result of the blast transformation. Negative controls showed that the
increases were not due to the mitogenic stimulation alone. As part of a series of tests by NTP to
evaluate genotoxicity assays, phenol was tested for the induction of SCEs in CHO cells (Ivett et al.,
1989). Phenol was positive in the absence of S9 and weakly positive in the presence of S9. Cell
cycle delay was observed at all positive doses.
In an assay with Syrian hamster embryo cells, Tsutsui et al. (1997) reported that phenol
induced a slight dose-related increase in transformed colonies as well as a dose-related increase in
unscheduled DNA synthesis in the same cell line.
In an evaluation of effects on germ cells in vivo, Skare and Schrotel (1984) found no effect on
single strand-breaks in testicular cells of Sprague-Dawley rats receiving a single i.p. injection of up to
79 mg/kg or five daily i.p. injections of up to 39.5 mg/kg-day. Phenol also did not induce single
strand breaks in mouse lymphoma L5178YS cells (Pellack-Walker and Blumer, 1986) or in CHO
cells in a test up to cytotoxic concentrations (Sze et al., 1996).
Reddy et al. (1990) reported that DNA adducts were produced in cultured rat Zymbal glands
orally dosed with 750 |o,g/mL of either phenol or hydroquinone. The adducts were not chemically
characterized and their intensities were not quantified, but no spots were observed
autoradiographically in the untreated controls. By contrast, many different adducts were seen in the
analyzed tissues (bone marrow, Zymbal gland, liver, spleen) from both untreated female Sprague-
Dawley rats and from rats treated for 4 days by oral gavage with a dose of 75 mg/kg-day phenol or
150 mg/kg-day of a 1:1 mixture of phenol and hydroquinone. The adduct patterns and levels of
adducts did not differ significantly between control and treated animals.
The authors noted that endogenous adducts would interfere with the determination of
treatment-induced adducts that chromatograph similarly. To address this possibility, they compared
the chromatograms resulting from in vitro and in vivo treatments. The absence of the major in vitro
adducts of hydroquinone or benzoquinone in the in vivo samples suggested that these adducts were
not formed in the whole animal. Conversely, the primary adduct of phenol formed in vitro was also
observed in vivo, although the levels relative to controls were much higher under in vitro conditions.
82
-------
The authors suggested that the higher level of adducts following in vitro treatment versus in vivo
treatment could be attributed to detoxification of orally administered phenol, but they did not further
address the possibility that there may be a significant basal load of adducts formed by endogenously
produced phenol.
Using a fluorescence in situ hybridization approach, Chen and Eastmond (1995a) found that
treatment with phenol alone resulted in micronuclei and breaks in euchromatin, whereas
hydroquinone affected chromosome loss and chromosomal breakage, particularly in centromeric
heterochromatin. They suggested that the different pattern of effects with phenol and hydroquinone
indicates that the synergism between phenol and hydroquinone is not due solely to phenol-induced
increases in hydroquinone metabolism. Instead, they suggested, phenol or its metabolites may also
be inhibiting DNA repair. In a follow-up study, Chen and Eastmond (1995b) found that phenol alone
did not affect the DNA repair enzymes topoisomerase I or topoisomerase II in vitro. However,
mixing phenol with horseradish peroxidase to mimic the peroxidase metabolism of the bone marrow
resulted in complete inhibition of topoisomerase II; no effect was seen when glutathione was added to
the peroxidase mixture.
In an assay with NCTC 929 mouse fibroblast cells, Yang and Duerksen-Hughes (1998) found
that phenol caused a dose-related increase in levels of the p53 protein. The authors noted that cells
increased p53 levels in response to DNA damage.
4.4.3. Neurological Effects
As described above, tremors have been observed following relatively high exposures to
phenol via the oral (Dow Chemical Co., 1994; Moser et al., 1995) or inhalation (Dalin and
Kristoffersson, 1974) routes. Decreased motor activity and a statistically significant increase in
rearing post-exposure were also reported in a screening study with rats (Moser et al., 1995), and
altered balance was reported in rats exposed continuously via inhalation for 15 days (Dalin and
Kristoffersson, 1974). However, in a 13-week drinking water neurotoxicity study that included
extensive neurohistological analyses (ClinTrials BioResearch, 1998), the only observed effects were
decreased motor activity and decreased body weight (compared with the controls), which were
probably secondary to decreased water consumption as a result of poor palatability. On the basis of
the results of a short-term screening study, neurotoxic effects do not occur at lower exposures than
other systemic effects of phenol (Berman et al., 1995; Moser et al., 1995).
83
-------
4.4.4. Immunotoxicity
As described in Section 4.2, Hsieh et al. (1992) reported immune effects in CD-I mice
administered phenol in drinking water for 28 days. The reported effects included decreased antibody
response (based on the PFC assay and direct antibody quantification using ELISA), with some
evidence of decreased lymphoproliferative response and decreased mixed lymphocyte response. The
clearest response was seen in the PFC and ELISA, which are highly predictive of effects on host
resistance (Luster et al., 1992, 1993). Confidence in the study results is decreased by the somewhat
small sample size (five rather than eight per group). Berman et al. (1995) reported atrophy of the
spleen or thymus of rats gavaged with phenol under conditions that greatly enhanced toxicity in
comparison with drinking water exposure.
Although no effects on spleen weight, cellularity, or antibody-forming cells in the spleen were
observed in a two-generation study of Sprague-Dawley rats exposed to phenol in drinking water at
concentrations of up to 5000 ppm (approximately 300 mg/kg-day) (Ryan et al., 2001; reported in
unpublished form as IIT Research Institute, 1999), qualitative and quantitative differences between
rats and mice in the effects of chemicals on the immune response are not unusual (e.g., U.S. EPA,
2000c). There is also no consistent pattern between mice and rats regarding which species is more
predictive of immunotoxic effects in humans. In light of Hsieh et al. (1992) who reported
immunotoxic effects of phenol at unusually low doses, it would be useful to confirm the results of
that study in mice using a protocol compliant with EPA immunotoxicity test guidelines (U.S. EPA,
1998c).
The National Institute of Environmental Health Sciences (NIEHS) and the NTP have elected
to conduct a comprehensive series of tests to evaluate the potential of phenol to adversely affect the
immune system (verbal communication from Dori Germolec, NIEHS to Monica Barren, EPA, 2002).
Using test protocols designed to be consistent with EPA testing guidelines and GLP standards, the
first assay considered the same dosing regimen used in the Hsieh et al. (1992) study. That is, inbred
female B6C3F1 mice-rather than outbred CD-I male mice-were exposed to drinking water
concentrations of 0, 5, 20, and 100 mg/L phenol (approximately 0, 1.7, 6.7, and 33 mg/kg-day). The
study also included a positive control. Preliminary results demonstrated immunosuppression
(reduced antibody response) at all levels of exposure, confirming the Hsieh study outcome.
In order to characterize a wider range of response levels, a second assay was initiated using
drinking water concentrations of 0, 1.25, 2.5, 5.0, 20, and 40 mg/L (approximately 0, 0.4, 0.8, 1.7,
6.7, and 13.3 mg/kg-day), overlapping the previous study's range of exposures. In addition,
NIEHS/NTP has initiated host resistence studies. Depending on the final outcome of this series of
84
-------
tests, including NIEHS/NTP peer review, EPA may reconsider and, if appropriate, reopen this
assessment.
4.4.5. Other Studies
Eastmond et al. (1987) investigated the role of phenol in benzene-induced myelotoxicity. No
suppression of bone marrow cellularity was observed in male B6C3F1 mice treated intraperitoneally
with doses as high as 150 mg/kg twice daily for 12 days (daily doses up to 300 mg/kg). Only minimal
suppression was observed in mice dosed with hydroquinone at up to 100 mg/kg twice per day. By
contrast, marked statistically significant, dose-related suppression was seen in mice treated with 75
mg/kg phenol and 75 mg/kg hydroquinone under the same conditions. In further in vitro studies, the
authors showed that phenol stimulates the horseradish peroxidase-mediated metabolism of
hydroquinone, and they hypothesized that similar stimulation of local peroxidases occurs in the bone
marrow. The observation of myelotoxicity following benzene treatment-but only minimally or not at
all following phenol or hydroquinone treatment- was therefore explained by a more-than-additive
interaction between phenol and hydroquinone.
Corti and Snyder (1998) evaluated gender- and age-specific differences in cytotoxicity of
benzene metabolites in vitro. Bone marrow cells were harvested from adult unexposed male and
female Swiss Webster mice as well as from pregnant females and from fetal males and females.
Cultures of CFU-e (colony forming units-erythroid, an erythroid precursor cell particularly susceptible
to benzene toxicity) were prepared and then exposed to different concentrations of the metabolites.
Although most of the benzene metabolites caused marked cytotoxicity, only minimal toxicity (0-20%
cytolethality) of phenol was observed up to the highest concentration tested (40 • M), compared with
nearly 100% cytolethality at the same concentration of catechol or hydroquinone. The effects were
strongest in cells isolated from fetal females or from virgin adult females, but the dose-response was
inconsistent, and it appears that no statistical comparisons with the untreated control of the same life
stage were done.
Zamponi et al. (1994) studied the mechanism of phenol-induced cardiac arrhythmia, including
ventricular tachycardia. In an abbreviated report, the authors suggested that phenol caused cardiac
arrhythmia by blocking batrachotoxin-activated cardiac sodium channels. Testing conditions,
including doses tested, were not provided in the abbreviated report.
Bishop et al. (1997) investigated the effect of phenol exposure on total reproductive capacity
in mice. Groups of 26 female hybrid (SEC x C57BL6) Fl mice were given a single i.p. injection of 0
(buffer solution) or 350 mg/kg of phenol, and the females were caged individually with an untreated
male hybrid (C3H/R1 x C57BL10) Fl mouse following the day of injection for 347 days. The
85
-------
animals were observed daily for producing newborn mice; the young mice were counted and
discarded immediately after birth. Female reproductive performance was evaluated on the basis of
the total number of offspring per female and the average number of litters per female. The numbers
of offspring per female and litters per female in phenol-treated mice were comparable to those in the
controls. Thus, phenol had no measurable detrimental effect on the parameters used for evaluating
long-term reproductive effects in this study.
4.5. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS AND MODE
OF ACTION
Studies investigating the effects of orally administered phenol are summarized in Table 2.
When phenol is administered in drinking water, the most common effect is decreased water
consumption, presumably due to poor palatability. Effects seen concurrently with decreased water
consumption, and probably secondary to dehydration, include body weights lower than those of the
controls, decreased maternal body weight compared with that of developmental toxicity studies, and
decreased pup survival prior to culling. Decreased motor activity was also seen in a drinking water
neurotoxicity study, but it does not appear to be secondary to the decreased water consumption. The
data also indicate that phenol causes immune effects in mice, but not in rats.
Oral exposure to phenol also affects the kidney and liver. Kidney inflammation was observed
in the chronic drinking water study in rats (NCI, 1980). Liver and kidney pathology (tubular
degeneration, kidney necrosis, and vacuolar degeneration) in rats also has been observed in short-
term and subchronic toxicity studies using gavage dosing (Berman et al., 1995; Moser et al., 1995;
Dow Chemical Co., 1945). These results from experimental animal studies support the observations
in case studies and epidemiology studies (Shamy et al., 1994; Merliss, 1972) that the liver is a target
of phenol in humans.
A number of nervous system effects have been observed following phenol dosing. Tremors
were observed in one animal that later died (apparently of dehydration) following dosing in drinking
water (ClinTrials BioResearch, 1998). Tremors have also been observed in several gavage studies in
rats and mice (NTP, 1983a; Dow Chemical Co., 1994; Moser et al., 1995). However, in a specialized
13-week neurotoxicity study in male and female rats that included an FOB and a detailed
neurohistopathology evaluation (ClinTrials BioResearch, 1998), the only observed nervous system
effects were tremors in one animal and decreased motor activity in females. A short-term gavage
screening study (Moser et al., 1995) found that the only effect in an FOB was a marginal decrease in
motor activity and increased rearing post-exposure.
86
-------
Headaches and weakness were reported in workers exposed to a mixture of phenol and a
number of other aromatic compounds (Baj et al., 1994). Due to the mixed nature of the exposure, the
effects cannot be clearly attributed to phenol. Muscle pain in a laboratory technician who developed
phenol marasmus after being frequently exposed to phenol vapor (Merliss, 1972) may have reflected
neurological damage.
The data regarding the hematotoxic potential of phenol are conflicting. No hematological
effects were observed in rats in a well-conducted two-generation study in Sprague-Dawley rats (Ryan
et al., 2001; available in unpublished form as IIT Research Institute, 1999). By contrast, decreased
hematocrit and erythrocyte counts were seen at much lower doses in a 28-day drinking water study in
mice (Hsieh et al., 1992), although this study is limited by the use of only five males per dose. Data
from these two studies are contrasted in Table 6. The differences between the two studies cannot be
resolved by considering the results of the chronic drinking water studies conducted by NCI (1980) in
mice and rats, because no hematological evaluation was conducted in those studies.
The negative finding in rats following oral exposure are supported by the absence of
hematological effects in rhesus monkeys, male Sprague-Dawley rats, and male albino mice (strain not
further identified) exposed to an average phenol concentration of 18.2 mg/m3 continuously for 90
days (Sandage, 1961). Hemoglobin and hematocrit were also unaffected in a small study of rats
exposed to phenol in air at 100 mg/m3 continuously for 15 days (Dalin and Kristoffersson, 1974).
Neither of these inhalation studies used modern exposure protocols, and both were limited by
inadequate exposure monitoring. However, Dalin and Kristoffersson (1974) did both report systemic
effects (nervous system and liver effects) in rats at exposure levels that did not cause hematological
effects. Sandage (1961) found an indication of liver and kidney histopathology in the monkeys and
rats, although not in the mice. Assuming that all of the inhaled phenol was absorbed (see Section
3.2), the systemic dose to mice in the Sandage (1961) study can be estimated at approximately 30
mg/kg-day, based on a body weight of 0.03 kg and an inhalation rate of 0.052 mg/m3/day. This dose
is comparable to the high dose in the Hsieh et al. (1992) study.
The negative results in the inhalation studies raise further questions about the reliability of the
hematotoxicity effects seen by Hsieh et al. (1992) in mice-particularly in the light of the small sample
size-as well as the relevance of those results to humans. Because portal-of-entry conjugation is more
efficient following ingestion rather than following inhalation of phenol (see Section 3.3), it is not
surprising that the systemic toxicity (i.e., liver and kidney effects) of a given absorbed dose may be
higher for inhaled phenol. Human data on hematotoxic effects of phenol are limited. Baj et al. (1994)
reported a small but statistically significant decrease in erythrocytes in workers exposed to a mixture
87
-------
containing phenol, chlorinated benzenes, and other compounds. Due to the mixed nature of the
exposure, the effect, if any, cannot be clearly attributed to phenol.
The results of Hsieh et al. (1992) also indicate that phenol can be immunotoxic to mice. The
investigators observed a clear dose-related decrease in two related measures of antibody formation
(the PFC assay and a direct measure of antibody titer using ELISA), along with some evidence of a
decreased cell-mediated response at the high dose. Confirmation of their results in a repeat assay
conducted according to EPA test guidelines would be useful in light of the small number of animals
used and the limited number of risk assessments that have been based on in vivo/in vitro
immunotoxicity assays.
No effect on spleen weight, cellularity, or antibody-forming cells (in the PFC assay) were
observed in a two-generation study of Sprague-Dawley rats exposed to phenol in drinking water at
much higher doses (Ryan et al., 2001; reported in unpublished form as IIT Research Institute, 1999),
but qualitative and quantitative differences in effects of chemicals on the immune response of rats and
mice are not unusual (e.g., see dioxin, as described in U.S. EPA, 2000c). "Necrosis or atrophy in the
spleen or thymus" (not further described) was observed in a 14-day screening study of rats gavaged
with phenol (Berman et al., 1995), supporting the immune organs as targets of phenol.
Baj et al. (1994) reported in an epidemiology study of Polish workers that exposure to
Ksylamit® vapor resulted in immune effects, but it is unclear whether phenol is the causative agent
because Ksylamit® contains a number of different aromatic compounds. Overall, the data indicate
that phenol by itself may cause immunotoxicity in humans, but more data are needed to address this
possibility. Interaction between phenol and benzene metabolites may also cause immune effects, as
described below.
Benzene (which is metabolized to phenol) among other compounds, causes immunological
effects, including lymphopenia and leukopenia (reviewed in ATSDR, 1998). However, although
benzene is a leukemogen in humans, it has not been shown to induce leukemias in experimental
animals. For example, in the NTP gavage studies of benzene (NTP, 1986), it was carcinogenic to both
male and female F344 rats and B6C3F] mice, inducing tumors at multiple sites. There was a
statistically significant increase in lymphomas in male and female mice but not in rats of either sex.
No significant increase in leukemias was noted in either species.
One proposed mechanism by which this immunotoxicity is induced involves the interaction
between phenol and hydroquinone, in which phenol stimulates the metabolism of hydroquinone.
Eastmond et al. (1987) observed decreased bone marrow cellularity in male mice dosed
intraperitoneally with phenol and hydroquinone but not with phenol alone at doses of up to 300 mg/kg-
day for 12 days and only minimally with hydroquinone. These results appear to contradict those seen
-------
at much lower doses (Hsieh et al., 1992), although the target tissue examined by Eastmond et al. (1987)
was bone marrow, whereas Hsieh et al. examined the spleen. The former study was also conducted via
the i.p. rather than oral route, but toxicity might be expected to be higher via the i.p. route because first-
pass metabolism would be lower.
It is not known with certainty whether the toxic effects of phenol are due to the parent
compound or to its metabolites. Distinguishing between effects of each is complicated by the lack of
adequate data on concurrent blood levels of parent compound and metabolites. Phenol and metabolite
levels in blood would be expected to rise in parallel as portal-of-entry metabolism becomes saturated.
However, several lines of reasoning suggest that many of the toxic effects are due to the oxidative
metabolites of phenol. In an in vitro study of the dysmorphogenic and embryotoxic effects of benzene
and phenol and their metabolites on whole rat conceptuses, Chapman et al. (1994) found that phenol
toxicity was much higher in the presence of S9.
The target tissues of phenol toxicity (kidney, liver, lung, and possibly bone marrow) are also
those in which phenol can be oxidatively metabolized. In a 2-week inhalation study, Dalin and
Kristoffersson (1974) observed altered balance and twitches in the absence of increased amounts of
phenol in the blood, suggesting that a phenol metabolite rather than phenol itself is the toxic agent.
Alternatively, the analytical method used may not have been sensitive enough or specific enough to
detect any changes in blood levels of phenol. Phenol could produce portal-of-entry and systemic
toxicity as a result of its ability to react with and to denature protein.
A key point with regard to the evaluation of the toxicity of orally administered phenol is
whether gavage studies accurately represent the toxicity under environmental exposure conditions.
Gavage studies are typically done using a single bolus dose per day, whereas environmental exposure is
more likely to involve exposure distributed over the course of the day. Although laboratory animals
consume drinking water in a few larger doses primarily during the active period rather than in
continuous small sips, the toxicokinetics of environmental exposure are more closely modeled by a
drinking water study or a gavage study using divided dosing, than by a study using a single gavage
dose per day.
Figure 2 compares the doses and observed severity of effects in drinking water and gavage
89
-------
Figure 2. Plot of severity with dose for drinking water (DW) (open symbols) orgavage (filled-in
symbols). Values of 1, 2, 3 correspond with NOAEL, AEL, or PEL, respectively.
4 -
3 -
oDW NOAEL
oDW AEL
pDW PEL
» Gavage NOAEL
• Gavage AEL
• Gavage PEL
0)
I
U)
2 -
ooo
<»voo
10
100
Dose (mg/kg-day)
1000
10000
90
-------
studies. Of particular interest is the number of gavage studies in which death, a frank effect, was
observed, whereas drinking water studies at comparable or higher daily doses produced only tremors,
kidney inflammation, and effects secondary to decreased water intake. The gavage NOAELs that
occur at the same doses as gavage adverse effect levels are for maternal and developmental toxicity.
The sole exception to the large difference between gavage and drinking water studies in doses that
cause effects is the 28-day drinking water study by Hsieh et al. (1992), which reported hematological
and immune effects at doses comparable to NOAELs in drinking water studies. As noted above, it
would be useful to obtain independent confirmation in mice of the results reported by Hsieh et al.
(1992).
Toxicokinetic data support this difference between gavage and drinking water studies and
indicate that toxicity is correlated with peak blood concentrations rather than the area under the
curve. Dow Chemical Co. (1994) observed "phenol twitching behavior" (including tremors and eye
blinking) in rats gavaged with 150 mg/kg phenol; the behavior disappeared in less than an hour post-
dosing, as phenol blood levels declined below peak values. By contrast, no twitching behavior was
observed following a similar daily dose of phenol in drinking water. Unfortunately, blood levels of
phenol or its metabolites were not determined in the drinking water phase of the study, but they are
likely much lower than in the gavage phase, in light of the rapid blood clearance.
The higher systemic toxicity of gavaged phenol when it is administered in smaller volumes
(NTP, 1983a) also supports the idea that toxicity is related to peak blood concentrations, because
smaller dosing volumes would be expected to enhance the absorption rate. An unpublished GLP
range-finding study for maternal toxicity (International Research and Development Corp., 1993) also
found higher systemic toxicity for phenol when it was administered in smaller dosing volumes.8 Data
on the relationship between dosing volume and peak phenol blood concentrations are not available.
Data on the relationship between peak blood concentrations and effects also are not available for the
endpoints relevant to the critical effect.
The inhalation data on the effects of phenol are very limited (Table 7). Only one study
conducted according to modern toxicological methods was located (Hoffman et al., 2001; available
in unpublished form as Huntingdon, 1998), and the exposures in this study were for only 2 weeks.
8Although this unpublished study is not a primary reference for this assessment, it is presented
here because it contributes some useful information to the overall hazard identification phase of the
phenol assessment.
91
-------
Other studies ranged from 2 weeks (Dalin and Kristoffersson, 1974) to 90 days (Deichmann et al.,
1944; Sandage, 1961), but they included incomplete documentation of the study results, and they
did not use modern methods for controlling exposure levels. In addition, the authors of some of the
studies (e.g., Sandage, 1961) appear to have been looking for marked effects and thus dismissed
statistically significant incidences of organ pathology of lesser severity. Nonetheless, the studies
are fairly consistent with regard to the target organs and the effects observed. Exposure to high
concentrations produced nervous system effects, and liver, kidney, and lung pathology occurred at
lower concentrations. Rats were reported to be much less sensitive than rabbits or guinea pigs
(Deichmann et al., 1944). The systemic targets observed following inhalation exposure to phenol
are supported by data from the oral exposure route.
Information on the mode of action of inhaled phenol toxicity is also quite limited, but some
extrapolation from other routes is possible. On the basis of the irritative and corrosive effects seen
following dermal exposure to phenol, respiratory tract effects are likely due to direct contact of
phenol with the respiratory tract tissue. As noted in Section 3, phenol is extensively absorbed
following inhalation exposure. The lung can metabolize phenol prior to absorption, but the
efficiency of metabolism in the lung is lower than that for the gut or liver (Cassidy and Houston,
1984). After the inhaled phenol (and its metabolites) reaches the blood stream, the same points
described above for the oral route are relevant. In brief, it is not known whether the systemic toxic
effects of inhaled phenol are due to phenol itself or to its metabolite(s), but at least some of the
toxic effects appear to be attributable to phenol metabolite(s) (Chapman et al., 1994). Systemic
toxicity appears to be related to peak concentrations in blood rather than to total daily intake.
4.6. WEIGHT OF EVIDENCE EVALUATION AND CANCER
CHARACTERIZATION-SYNTHESIS OF HUMAN, LABORATORY ANIMAL,
AND OTHER SUPPORTING EVIDENCE, CONCLUSIONS ABOUT HUMAN
CARCINOGENICITY, AND LIKELY MODE OF ACTION
Chronic drinking water bioassays of phenol have been conducted in rats and mice (NCI,
1980). In these studies, NCI concluded that phenol was "not carcinogenic" in male or female F344
rats or B6C3F1 mice. However, the report also noted that leukemia and lymphoma were statistically
significantly increased in low-dose male rats, although there was no significant increase at the high
dose. The increases in leukemia are of particular interest in light of the leukemogenic effects of
benzene (for which phenol is a metabolite) in humans. (In experimental animals, benzene has not
92
-------
been shown to induce leukemia, although increases in lymphoma have been observed [e.g., NTP,
1986].)
The MTD was clearly reached in the rat study (NCI, 1980), on the basis of decreased body
weight compared to controls and on kidney histopathology. Although the only sign of toxicity in the
mouse study was decreased body weight (compared to the controls) secondary to decreased water
consumption, higher doses probably could not have been tested, because of the decreased water
consumption. Higher toxicity probably could have been achieved in a gavage study in mice at lower
doses. These considerations suggest that an MTD was also reached in mice, although a definitive
conclusion is difficult.
No other long-term oral carcinogenicity studies of phenol are available. No inhalation
studies of phenol were of a sufficient duration to assess phenol carcinogenicity. The only long-term
study that has assessed the carcinogenicity of phenol applied dermally without initiation was that of
Boutwell and Bosch (1959), in which increased papillomas were seen at a dose that also caused
ulceration.
In contrast with these negative results for oral carcinogenicity, dermally administered phenol
has been consistently observed to be a promoter. Several authors (Salaman and Glendenning, 1957;
Boutwell and Bosch, 1959; Wynder and Hoffmann, 1961) observed that dermally applied phenol
promoted DMBA-initiated skin tumors. These studies have generally reported significant skin
ulceration at all phenol doses tested. The exception is Wynder and Hoffman (1961), who reported
that 5% phenol promoted DMBA-initiated tumors in mice in the absence of any toxic reactions.
When the same phenol dose was administered in different volumes, higher promotion activity was
exhibited by the more concentrated solution, which also produced severe skin ulceration, suggesting
that some of the promotion activity may have been related to the rapid cell division in the repairation
of skin damage (Salaman and Glendenning, 1957). The observed response was dose-related
(Boutwell and Bosch, 1959), but marked systemic toxicity was also observed at these doses. Co-
carcinogenesis with dermally administered benzo[a]pyrene has also been observed (Wynder and
Hoffmann, 1961). Because the benzo[a]pyrene was co-administered with the phenol, this assay
cannot be classified as a true initiation/promotion assay. Production of papillomas by dermally
administered phenol (in the absence of an initiator) was observed only at a concentration that caused
ulceration, and hence was above the MTD.
Genotoxicity studies have found that phenol tends not to be mutagenic in bacteria (Pool and
Lin, 1982; Rapson et al., 1980; Haworth et al., 1983), but positive or equivocal results have been
93
-------
obtained in gene mutation assays in mammalian cells (McGregor et al., 1988a, 1988; Paschin and
Bahitova, 1982; Tsutsui et al., 1997). Increases were larger in the presence of S9 activation. Phenol
tended to induce micronuclei in mice when administered intraperitoneally (Marrazzini et al., 1994;
Chen and Eastmond,1995a; Ciranni et al., 1988b), but it produced negative (or positive only at very
high doses) results when administered orally (Ciranni et al., 1988b; Gocke et al., 1981). This
difference is likely due to the first-pass conjugation and inactivation of orally administered phenol.
Phenol was also positive in in vitro micronucleus tests with human lymphocytes (Yager et al., 1990)
and CHO cells (Miller et al., 1995), and it caused chromosome aberrations in the presence of S9
activation in CHO cells (Ivett et al., 1989). Phenol has been observed to act synergistically with
hydroquinone in the production of genotoxic effects (Marrazzini et al., 1994; Barale et al., 1990;
Chen and Eastmond, 1995a).
Epidemiology data do not shed further light on the carcinogenic potential of phenol. Some
studies (Kauppinen et al., 1986; Dosemeci et al., 1991) have reported elevated risks in phenol-
exposed workers, whereas others have observed no effect (Wilcosky et al., 1984). However, the
usefulness of each of these studies for risk assessment is limited by (depending on the study) an
absence of an effect when latency was considered, a lack of a dose-response, and potential for
confounding.
Although phenol was negative in oral bioassays conducted in rats and mice (NCI, 1980),
questions remain regarding its carcinogenic potential in light of the positive results in
initiation/promotion assays (albeit at exposures typically above the MTD), the increases in leukemia
in low-dose male rats in the oral bioassay, and the observation of gene mutations in mammalian cells
in vivo and micronuclei in vivo following i.p. dosing. No inhalation studies of sufficient duration to
assess phenol carcinogenicity have been conducted. Dermal carcinogenicity or initiation/promotion
studies with phenol at exposures below the MTD have not been conducted. The carcinogenic
potential of phenol via inhalation exposure has not been evaluated at all. Under the draft revised
Guidelines for Carcinogen Risk Assessment (U.S. EPA, 1999), the data regarding the carcinogenicity
of phenol via the oral, inhalation, and dermal exposure routes are inadequate for an assessment of
human carcinogenic potential. Under the current guidelines (U.S. EPA, 1986a), phenol falls in
Category D: not classifiable as to human carcinogenicity.
Similar conclusions have been reached by other groups in recent assessments of the
carcinogenicity of phenol. IARC (1999) concluded that there is inadequate evidence in humans
and in experimental animals for the carcinogenicity of phenol. Overall, IARC concluded that phenol
94
-------
is not classifiable as to its carcinogenicity to humans (Group 3). Phenol is not included in the 9th
Report on Carcinogens (NTP, 2000a), and it is not listed as being considered for inclusion in the 10th
Report on Carcinogens (NTP, 2000b). This report contains only chemicals and substances that have
been reviewed and classified as known human carcinogens or as reasonably anticipated to be
carcinogens.
4.7. SUSCEPTIBLE POPULATIONS
People with decreased ability to conjugate and eliminate phenol, such as those with low
activities of PST or glucuronyltransferase, may be more susceptible to phenol toxicity. If the
toxicity of phenol is due to oxidative metabolites such as hydroquinone or catechol, individuals with
increased oxidative activity would also be more sensitive to phenol toxicity. The activity of
CYP2E1, the enzyme responsible for the oxidation of phenol, can be increased by exposure to a
variety of agents, including ethanol and chlorinated solvents, so people with high exposures to these
agents may be more sensitive to phenol.
4.7.1. Possible Childhood Susceptibility
As discussed in Section 4.3, a gavage study in rats (NTP, 1983a) reported decreased fetal
body weight at a dose below that at which maternal toxicity occurred. In contrast, maternal toxicity
occurred at a lower dose than did decreased fetal body weight in a gavage developmental toxicity
study that used a divided-dosing protocol (Argus Research Laboratories, 1997). The observation of
a fetal effect at a dose as low as 120 mg/kg-day (NTP, 1983a) suggests that the developing fetus is a
possible susceptible population. The strength of this conclusion is weakened, however, by the small
magnitude of the fetal weight decrease together with the increased litter size, which led to the high
dose being identified as an equivocal LOAEL (see Section 4.3). The strength of the conclusion is
also weakened by the observation in another gavage rat study (Argus Research Laboratories, 1997)
that decreased fetal weight occurs only at doses above those that cause decreased maternal weight
gain, and the observation of decreased fetal weight in a drinking water study (Ryan et al., 2001) only
at concentrations that also resulted in reduced water consumption.
Only one study was located that specifically addressed age-related differences in the
systemic toxicity of phenol. Deichmann and Witherup (1944) compared the lethality of an oral dose
of 600 mg/kg phenol (administered as a 5% aqueous solution) in 10-day-old, 5-week-old, and adult
rats. Mortality was 90%, 30%, and 60% in the neonates, young rats, and adult rats, respectively.
Although the young and adult rats died within 1.5 hours of dosing, the neonates died 12-24 hours
95
-------
after dosing. The data from this study suggest that neonates are more sensitive than adults and
young rats may be less sensitive than adults. Alternatively, the age-related differences observed in
this study could reflect inter-individual variability that was not a consequence of age. This study has
not been replicated; therefore, definitive conclusions are not possible.
Data from humans and rats are consistent in showing very little fetal expression of CYP2E1,
which is rapidly induced shortly after birth and rises though childhood (reviewed in Hakkola et al.,
1998). The age after parturition at which CYP2E1 levels peak has been studied in laboratory
animals, with inconsistent results. Some studies suggest that peak levels are reached during
childhood, with a subsequent decrease to adult levels (Schenkman et al., 1989), whereas others have
shown a rapid rise in CYP2E1 levels during childhood to a maximum level that is sustained during
adulthood (Song et al., 1986). If the toxic moiety is a metabolite, decreased P450 metabolism could
be expected to result in decreased toxicity. However, sulfate conjugation is also lower early in life
(Iwasaki et al., 1993), so more of the dose is available for oxidative metabolism. Indeed, Heaton and
Renwick (1991) found higher production of oxidative metabolites in young rats. This difference was
higher in males, with the percentage of the administered dose recovered as hydroquinone conjugates
decreasing from 38% of the administered dose in 3-week-old males to 8.2% in 16-week-old rats.
Smaller decreases with age (from 17.8% of the administered dose in 4-week-old rats to 10.5% in 15-
week-old rats) were observed in females. These data suggest the potential for children to be more
sensitive than adults to the systemic effects of phenol.
4.7.2. Possible Gender Differences
Kenyon et al. (1995) (in mice) and Heaton and Renwick (1991) (in rats) reported higher
excretion of hydroquinone conjugates in males than in females, suggesting higher levels of
hydroquinone production in males. By contrast, Meerman et al. (1987) reported only slightly faster
metabolism in male rats. These data would tend to suggest that if hydroquinone is the toxic moiety,
phenol would be more toxic in males. However, few differences in phenol toxicity between males
and females were identified; differences in NOAELs reflect differences in water consumption per
unit weight, resulting in differences in estimated intake. Acute oral lethality data do suggest that
phenol is more toxic to males (Thompson and Gibson, 1984).
5. DOSE RESPONSE ASSESSMENTS
96
-------
5.1 Oral Reference Dose (RfD)
5.1.1. Choice of Principal Study and Critical Effect
An extensive database for the effects of orally administered phenol is available. The studies
relevant to the development of the RfD are summarized in Table 2. Two-year drinking water studies
conducted in rats and mice are available (NCI, 1980). Hematology and serum biochemical
evaluations were not included in those chronic studies, but they were included in a recent two-
generation drinking water study conducted in rats (Ryan et al., 2001; available in unpublished form
as IIT Research Institute, 1999). The only study evaluating hematological effects in mice is a 28-
day drinking water study (Hsieh et al., 1992). A specialized subchronic neurotoxicity study was
conducted with rats exposed to phenol in drinking water (ClinTrials BioResearch, 1998). A number
of developmental toxicity studies have been conducted in rats and mice, all using the gavage route
(Argus Research Laboratories, 1997;NTP, 1983a;NTP, 1983b, Narotsky andKavlock, 1995).
As shown in Table 2, the study with the lowest NOAEL/LOAEL boundary is the 28-day
drinking water study in mice by Hsieh et al. (1992). In this study, the NOAEL was 6.3 mg/kg-day,
and the LOAEL was 33.6 mg/kg-day, based on decreased antibody response, supported by decreases
in hematocrit and red blood cells. Statistically significant decreases in erythrocyte counts were
observed at the low and mid doses, but these results were considered biologically questionable in the
absence of effects on hematocrit, in the absence of data addressing whether the apparent
inconsistency was due to macrocytosis, and in light of the lack of support from other studies.
As noted in Section 4.5, this study is useful for hazard identification. However, confirmation
of the study results in an independent assay in mice would be useful before using the data for dose-
response assessment, considering the small number of animals tested and the limited number of risk
assessments that have been based on in vivo/in vitro immunotoxicity assays. In addition, although
qualitative differences between rats and mice in immune assays are not unusual (e.g., dioxin, see
U.S. EPA, 2000c), it is of interest that Hsieh et al. (1992) observed immune effects in mice at very
low doses, but the only other study evaluating similar immune parameters (Ryan et al., 2001; IIT
Research Institute, 1999) did not observe effects in rats at 10-fold higher doses. No other studies in
mice have directly evaluated effects on antibody forming cells.
In another study that evaluated immune effects of phenol on mice (Eastmond et al., 1987), no
effect on bone marrow cellularity was observed at phenol doses of up to 300 mg/kg-day in mice
dosed intraperitoneally for 12 days. Similarly, Corti and Snyder (1998) evaluated the effects of
benzene metabolites on CFU-e cells (an erythroid progenitor cell sensitive to benzene) harvested from
97
-------
mice and found that the cytotoxicity of phenol was much lower than that of other benzene metabolites.
The effects in mice and rats were not compared in that study.
In light of these issues, and in the absence of other data supporting the observed effects at such
low doses, the results of Hsieh et al. (1992) are too preliminary to be used as the basis for the phenol
RfD. However, this study does raise concerns regarding the potential of phenol to cause
hematological and immune effects, and it would be worthwhile to address these questions in a mouse
immunotoxicity study conducted according to modern methods. The uncertainties regarding these
endpoints and the use of a database uncertainty factor to address these uncertainties are further
addressed in Section 5.1.3.
The next lowest NOAEL/LOAEL combination was observed in a 14-day gavage study in rats
conducted as part of a large-scale screening study of a number of chemicals (Berman et al., 1995;
Moser et al., 1995). Tremor, kidney tubular degeneration, and increased rearing in an FOB were
observed at the high dose of 40 mg/kg-day but not at the next lower dose of 12 mg/kg-day. Although
the incidence of kidney histopathology was not statistically significant, the high dose can be
considered a LOAEL in light of the low statistical power of the study (only eight female rats per
group) and the rarity of these lesions in short-term studies. The corresponding NOAEL is 12 mg/kg-
day.
The relevance of this NOAEL to environmental exposures is questionable, however, due to the
markedly higher toxicity observed in gavage studies than in drinking water studies, as discussed in
Section 4.5, and the absence of supporting toxicity in drinking water studies of much longer duration.
In particular, drinking water studies found no kidney histopathology in rats exposed to 260 mg/kg-day
for 2 years (NCI, 1980) (although kidney inflammation was observed at higher doses), in mice
exposed to doses up to 660 mg/kg-day for 2 years (NCI, 1980), or in parental rats in a two-generation
reproduction study (Ryan et al., 2001; IIT Research Institute, 1999).
The only other study reporting kidney histopathology at low doses was a poorly documented
and unpublished 6-month gavage study in rats (Dow Chemical Co., 1945). The very small dosing
volume used by Berman et al. (1995) also may have contributed to the high toxicity, considering the
findings of NTP (1983a).
The principal study for development of the RfD is Argus Research Laboratories (1997), in
which decreased maternal weight gain was observed in rats gavaged on GD 6-15 with 120 mg/kg-day
phenol; the maternal NOAEL was 60 mg/kg-day, based on decreased body weight gain, and the
developmental NOAEL was 120 mg/kg-day. The BMDL was 93 mg/kg-day. No effect on body
98
-------
weight was reported at 120 mg/kg-day in systemic toxicity studies using drinking water as the
exposure route (NCI, 1980; ClinTrials BioResearch, 1998), but it is not surprising that maternal
toxicity following 9 days of dosing occurs at a lower dose than does chronic systemic toxicity,
considering the different physiological status during pregnancy. Although the principal study was
conducted via the gavage route, measures were taken to decrease the impact of bolus dosing by
dividing the daily dose into three administrations per day.
It is of interest that rats consume drinking water not in many little sips, but in a few larger
doses primarily associated with food consumption during the active period of the day. Therefore, the
toxicokinetic profile of the divided-dose gavage study may actually be fairly similar to the
toxicokinetic profile that would be observed with drinking water exposure. In addition, a more precise
measurement of administered dose is possible in gavage studies, because spillage can occur in
drinking water studies.
The NOAEL identified in the Argus Research Laboratories (1997) study is supported by a
developmental toxicity study (NTP, 1983a) in which decreased fetal weight was observed in CD rats
gavaged on GDs 6-15 with 120 mg/kg-day; the NOAEL was 60 mg/kg-day. The high dose of 120
mg/kg-day was considered an equivocal LOAEL for developmental effects, in light of the small
magnitude of the weight decrease, the increased litter size (which can result in decreased fetal weight),
and the absence of an effect on fetal weight at a maternally toxic dose in another gavage
developmental study in rats (Argus Research Laboratories, 1997). In the NTP (1983a) study, the
maternal toxicity NOAEL was the high dose, 120 mg/kg-day.
Because of the uncertainties regarding identification of the critical effect level for the NTP
(1983a) study, it was not considered to be an appropriate co-principal study. BMD modeling could
not be conducted on the fetal weight endpoint, because fetal weights were reported only as an average
across both sexes for all litters; no individual animal data were available. Because fetal weights of
male fetuses tend to be heavier than those of females, and because the number of fetuses per litter
affects the fetus weight, it was not appropriate to model the pooled data.
Although the decreased maternal weight gain (Argus Research Laboratories, 1997) was a
mild effect and was possibly confounded by the gavage dosing, these results are supported by a
drinking water study. Decreased motor activity was seen in female rats consuming the high
concentration of phenol (5000 ppm, corresponding to 360 mg/kg-day) in the 13-week neurotoxicity
study (ClinTrials BioResearch, 1998). The NOAEL in females was 107 mg/kg-day; no adequate fit
could be obtained using BMD modeling. As discussed in Section 4.2, the authors considered the
99
-------
decreased motor activity to be secondary to dehydration, but an analysis of the individual animal
data and comparison with the literature could not confirm this assumption.
Ryan et al. (2001) conducted a two-generation drinking water study (also available in
unpublished form as IIT Research Institute, 1999) in rats in which decreased parental and pup
weight occurred at a LOAEL of 301 mg/kg-day, with a NOAEL of 71 mg/kg-day. However, these
lower body weights, compared with control, are likely to be secondary to decreased water
consumption and not an indication of phenol toxicity.
5.1.2. Method of Analysis: Benchmark Dose
The RfD was derived by the BMD approach using BMDS Version 1.3, which downloaded
from the National Center for Environmental Assessment's web site. The BMR was defined as the
default of a change of one standard deviation (U.S. EPA, 2000d). A BMDL of 93 mg/kg-day was
derived for decreased maternal weight gain (Argus Research Laboratories, 1997) using the
polynomial model. Similar BMDL values of 125 and 129 mg/kg-day were calculated using the
power and Hill models, respectively, although the fit (based on the Akaike Information Criterion
[AIC]) was slightly better using the polynomial model, and a more conservative BMDL was
obtained using this model.
An alternative BMDL for this endpoint could be calculated using the geometric mean of the
BMDLs from all three models, 114 mg/kg-day, on the rationale that the small difference in AICs
was not meaningful. Other measures of fit (based on the goodness-of-fit p value and on visual fit)
also indicated that all three models are comparable. However, in this case the slightly more
conservative approach was used, in part as an added degree of protection because of the
uncertainties regarding immunotoxicity. Details of the model results are presented in
Appendix B.
5.1.3. RfD Derivation
The data on the within-human variability in the toxicokinetics and toxicodynamics of
ingested phenol are insufficient to adjust the default uncertainty factor for intraspecies variability
(UFA). In a sample of liver fractions from 10 people, Seaton et al. (1995) found that the kinetics of
phenol sulfation and hydroquinone conjugation varied by up to approximately threefold. Much
larger variability in CYP2E1 has been found, particularly between neonates and adults (Vieira et al.,
1996). These data on inter-individual variability in enzymatic metabolism are not adequate to move
100
-------
from the default UFH of 10, because they do not reflect potential variability in portal-of-entry
metabolism of phenol or uncertainty regarding the identity of the toxic moiety. Furthermore,
variability in CYP2E1 does not necessarily translate directly into variability in tissue dose of
because metabolism by CYP2E1 may be limited by blood flow to the liver.
The absorption, distribution, and metabolism of ingested phenol in rats and humans appear to
be generally qualitatively similar, although the data are insufficient for a quantitative comparison.
Comparison of laboratory animal and human phenol toxicokinetics is also limited by incomplete
information regarding the identity of the toxic moiety. As discussed in Section 4.5, the data suggest
that most of the toxic effects of phenol can be attributed to its oxidative metabolites, but the data are
insufficient to rule out the possibility that some effects may be attributable to phenol itself. In the
absence of adequate data on which to based a toxicokinetic or toxicodynamic comparison of rodents
and humans, the default UFA of 10 is used for interspecies extrapolation. However, it may be
possible to reduce this default value of 10 after review and evaluation of data (perhaps supplemented
by a PBPK model) that compare the toxicokinetics of phenol and its metabolites in the placenta and
fetus of rats and humans, if such data become available.
The BMDL was based on an effect of minimal severity (decreased maternal weight gain),
and a higher BMDL and NOAEL were obtained for effects on maternal weight. The BMDL is also
within 50% of the NOAEL identified for the decreased maternal weight endpoint. Therefore, no
uncertainty factor (UF) is required for extrapolation from a NOAEL to a LOAEL. No UF for
extrapolation across duration is needed because this developmental study is supported by chronic
bioassays in two species in which toxicity was observed only at higher doses.
The database for phenol by the oral route can be considered complete. It includes 2-year
drinking water studies conducted in rats and mice (NCI, 1980), a two-generation drinking water
study conducted in rats (Ryan et al., 2001; available in unpublished form as IIT Research Institute,
1999), and gavage developmental toxicity studies in rats (Argus Research Laboratories, 1997; NTP,
1983a; Narotsky and Kavlock, 1995) and mice (NTP, 1983b). However, the range of endpoints
evaluated in the chronic toxicity studies was limited and did not include hematological or serum
biochemistry evaluations. Immunological and hematological effects in mice were observed by
Hsieh et al. (1992) in a 28-day drinking water at low doses. These endpoints were evaluated, and no
significant hematological or serum biochemistry effects were observed at doses of up to >300
mg/kg-day in the two-generation rat study (IIT Research Institute, 1999; Ryan et al., 2001). The
difference in these results suggest species differences between mice and rats, but confirmation of the
101
-------
immunological and hematological effects in an assay done according to modern test methods would
be useful.
An i.p. study of the effects of phenol on bone marrow cellularity in mice at doses of up to
300 mg/kg-day (Eastmond et al., 1987) and an in vitro study with mouse bone marrow cells (Corti
and Snyder, 1998) also do not indicate that mouse blood cells are highly susceptible to effects of
phenol. However, these studies did not evaluate the same parameter measured by Hsieh et al.
(1992), and significant interspecies differences in immunotoxicity are not unusual. It is of interest
that the endpoints affected in the Hsieh et al. (1992) study (two measures of effects on antibody-
forming cells, PFC and ELISA) are the immune endpoints most highly predictive of effects on host
resistance (Luster et al., 1992; Luster et al., 1993). Therefore, to account for the uncertainties
regarding the immunological and hematological effects in mice, a database uncertainty factor of 3 is
used.
An additional degree of public health protection may also be provided by the use of a gavage
study rather than the more environmentally relevant route of drinking water. This is because gavage
administration results in a higher peak blood level-presumably even using a divided-dosing
protocol-than does ingestion of the same daily dose in drinking water. Because at least some toxic
effects of phenol are related to peak blood levels rather than to total intake, toxicity would be
expected to be higher following gavage exposure than drinking water exposure.
A composite UF of 300 results. No modifying factor is applied because the existing
uncertainties have been addressed with the standard uncertainty factors.
RfD = 93 mg/kg-day 7300 = 0.3 mg/kg-day, or 3E-1 mg/kg-day.
Note that this RfD is applied to ingested phenol in addition to the normal daily endogenous
production of phenol, as discussed in further detail in Section 6.1.2.
An additional uncertainty factor for sensitive populations such as infants and children is not
needed for phenol because sufficient studies of reproductive and developmental toxicity have been
performed.
5.2. Inhalation Reference Concentration (RfC)
5.2.1. Choice of Principal Study and Critical Effect
102
-------
The minimal database needed for the development of an RfC is a well-conducted subchronic
inhalation study that has adequately evaluated a comprehensive array of endpoints, including the
respiratory tract, and established a NOAEL and a LOAEL (U.S. EPA, 1994b). This criterion was
not met for phenol. Neither of the two available subchronic studies (Deichmann et al., 1944;
Sandage, 1961) are adequate for exposure-response assessment, because neither included adequate
documentation of the histopathology results, and neither used modern methods for generating or
monitoring exposure levels. These studies can, however, be used for hazard identification, and they
identify the respiratory tract, liver, and kidney as targets of inhalation exposure to phenol.
The phenol database also includes a well-conducted 2-week inhalation study with rats that
used modern exposure methods, evaluated a wide array of endpoints, and included a thorough
histopathology evaluation of the respiratory tract (Hoffman et al., 2001; the full unpublished study
report is available as Huntingdon, 1998). The only treatment-related effect observed was a red nasal
discharge in male rats, which was observed with a statistically significant duration-related and
concentration-related incidence in the mid- and high-concentration groups. However, because the
red nasal discharge was likely due to a nonspecific response to stress, this response is not considered
adverse.
In the absence of an inhalation study of sufficient duration, no RfC for phenol can be
derived. A route-to-route extrapolation is not appropriate, because phenol can be a direct-contact
irritant, and so portal of entry effects are a potential concern.
5.2.2. RfC Derivation
No RfC could be derived, due to insufficiencies of the database.
5.3. CANCER ASSESSMENT
As discussed in Section 4.6, the data regarding the carcinogenicity of phenol are
inadequate for assessment of human carcinogenic potential. Phenol was negative in oral
carcinogenicity studies in rats and mice, but questions remain regarding increased leukemia in
male rats in the bioassay as well as the positive gene mutation data and the positive results in
dermal initiation/promotion studies at doses at or above the MTD. No inhalation studies of an
appropriate duration exist. Therefore, no quantitative assessment of carcinogenic potential via
any route is possible.
103
-------
6. MAJOR CONCLUSIONS IN CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE
6.1. HUMAN HAZARD POTENTIAL
6.1.1. Oral Noncancer
In most studies of phenol administered in drinking water, water consumption was markedly
decreased at the highest dose, presumably due to poor palatability. A number of toxic effects
secondary to the decreased water consumption have been observed, including decreased
body weight compared to controls, decreased pup weight, and decreased pup survival pre-culling.
Other effects that may not have been secondary to decreased water consumption were kidney
inflammation (NCI, 1980) and decreased motor activity (ClinTrials BioResearch, 1998). Gavage
studies found more severe effects and reported these effects at lower doses. Observed effects
included lung, liver, and kidney pathology; tremors and other nervous system effects; and, at
sufficiently high doses, death.
These data suggest that the toxicity of phenol is higher via gavage dosing than via
administration in drinking water. The suggestion is supported by the finding that a series of
behaviors termed "phenol twitching behavior" correlate with peak blood levels rather than area
under the curve (Dow Chemical Co., 1994). For a given daily dose, peak blood levels would be
much higher following gavage dosing than following continuous administration in water. A direct
comparison of the toxicity of phenol when administered via these two routes could determine
definitively whether route-specific differences exist. Nonetheless, the data supporting the higher
toxicity of phenol administered by gavage were considered sufficiently strong to consider it
inappropriate to use the Berman et al. (1995) study (which also used small dosing volumes) as the
principal study.
Developmental toxicity studies have been conducted only via the gavage route. In the
principal study (Argus Research Laboratories, 1997), decreased maternal weight gain was observed
in rats administered 120 mg/kg-day in a divided-dosing protocol. The BMDL for this study was 93
mg/kg-day and the NOAEL was 60 mg/kg-day. Although exposure in this study was for only 9 days,
comparison with the entire database for phenol via the oral route indicates that this study
appropriately identifies the critical effect. Because a maternal effect (decreased maternal body
weight gain during gestation) is considered the critical effect for phenol (i.e., the first adverse effect
or its known precursor that occurs to the most sensitive species as the dose rate of an agent increases),
104
-------
protection from this effect would also be protective from systemic effects of chronic exposure, which
occur at higher doses. In light of the different physiological status during pregnancy, it is not
unreasonable for maternal toxicity following 9 days of dosing to occur at a lower dose than does
chronic systemic toxicity.
Although a NOAEL of 60 mg/kg-day was identified for decreased fetal weight in the NTP
(1983a) study, this study was not considered adequate to be a co-principal study. The high dose of
120 mg/kg-day was considered an equivocal LOAEL for developmental effects, in light of the small
magnitude of the weight decrease and the absence of an effect on fetal weight at a maternally toxic
dose in another gavage developmental study in rats (Argus Research Laboratories, 1997). In
addition, although the observed decrease in fetal weight was small (but statistically significant),
increased litter size was also seen at this dose. It is possible, therefore, that the dams were near the
limit of what they could carry in terms of pup burden (total fetal weight).
The NOAEL was supported quantitatively by the NOAEL of 107 mg/kg-day for decreased
motor activity in a 90-day drinking water neurotoxicity study (ClinTrials BioResarch, 1998). A
NOAEL of 71 mg/kg-day for decreased parental and pup body weights was also identified in a
drinking water two-generation reproduction study (Ryan et al., 2001; available in unpublished form
as IIT Research Institute, 1999), although these effects are likely secondary to decreased water
consumption. There is, however, some uncertainty in the identification of the NOAEL for this latter
study. As described in Section 4.3, a statistically significant decrease in uterine weight was observed
at all three doses in this study. The decrease was not considered adverse for a number of reasons,
including the absence of a dose-response and the small number of animals outside the control range.
This consideration, however, is rather subjective and is based on considerable professional
judgement.
A key uncertainty in the development of the RfD is the interpretation of the study by Hsieh et
al. (1992). Immunotoxicity (decreased response of antibody-forming cells) and hematotoxicity
(decreased red blood cells and hematocrit) were observed in this 28-day drinking water study in mice
at doses much lower than the doses than produced toxicity in other studies. No immunological or
hematotoxic effects were seen at much higher doses in a two-generation drinking water study in rats
(Ryan et al., 2001; IIT Research Institute, 1999). These differing results suggest species differences
between mice and rats, but confirmation of the immunological and hematological effects in an assay
done in mice according to modern test methods would be needed before using the data for dose-
response assessment, considering the small number of animals tested and the limited number of risk
105
-------
assessments that have been based on in vivo/in vitro immunotoxicity assays.
Similarly, in the absence of hematotoxicity in monkeys, rats, and mice following inhalation
exposure to phenol levels resulting in comparable or higher systemic doses of phenol (Sandage, 1961;
Dalin and Kristoffersson, 1974), confirmation of the reported hematological effects would also be
useful. A database uncertainty factor of 3 was used to account for the uncertainty regarding
immunotoxicity. This factor could be removed if an immunotoxicity study conducted according to
U.S. EPA (1998c) guidelines became available.
Although it does not directly affect the determination of the RfD, uncertainty also exists
regarding whether the decreased motor activity in females reported by ClinTrials BioResearch,
(1998) was due to dehydration only or whether phenol exposure also contributed to the effect. The
NOAEL from this study was used as supporting data for the principal study. The study authors
attributed the decreased motor activity to dehydration, because of the marked decrease in water intake
and the absence of supporting changes in the FOB. By contrast, this assessment concluded that
phenol at least contributed to the effect, because there was no clear correlation between individual
animals with dehydration and those with decreased activity and because the limited literature on the
topic reports no effect on motor activity of water deprivation for several days. A neurotoxicity study
in which the controls were allowed only limited access to drinking water would also address this
issue.
6.1.2. Inhalation Noncancer
The database for inhalation toxicity of phenol is very limited. A well-conducted 2-week study
is available (Hoffman et al., 2001; available in unpublished form as Huntingdon, 1998), but the
duration is less than that appropriate for serving as the basis for the RfC. Longer-term studies have
been conducted (Deichmann et al., 1944; Sandage, 1961), but they are limited by inadequate control
of exposure levels, unclear sensitivity of the evaluation, and limited reporting.
However, the inhalation toxicity studies are sufficient however, to identify the respiratory
tract, liver, kidney, and nervous system as targets of inhaled phenol toxicity. A significant
uncertainty exists regarding which species is the most appropriate for extrapolation to humans.
Deichmann et al. (1944) reported marked systemic toxicity in rabbits and deaths in guinea pigs at
exposure concentrations that caused no histopathology in rats. No inhalation studies in guinea pigs or
rabbits have been conducted to confirm these findings. In addition, it is unclear which of these
species is most like humans.
106
-------
The primary data need for developing an RfC is a 90-day inhalation study that includes a
thorough examination of the respiratory tract. Pharmacokinetic studies of inhaled phenol would also
aid in the extrapolation from experimental animals to humans.
6.1.3. Cancer
Several epidemiology studies have evaluated the carcinogenesis of phenol, but they have not
found a consistent dose-related association. Because all of the subjects were also exposed to other
chemicals and there was no correction for smoking, these studies are not adequate to reach
conclusions on the carcinogenic potential of phenol.
Phenol was negative in drinking water bioassays with rats and mice (NCI, 1980), although an
increased incidence of leukemias was observed in low-dose male rats. No inhalation studies of
sufficient duration to assess carcinogenicity were found. In short-term dermal assays, tumorigenicity
(production of papillomas in the absence of treatment by an initiating agent) was observed only at a
dose/concentration combination that produced ulceration and thus was well above the MTD (Salaman
and Glendenning, 1957). Similarly, although phenol was a promoter when tested in
initiation/promotion studies, the doses tested typically caused ulceration (Salaman and Glendenning,
1957; Boutwell and Bosch, 1959) and death (Boutwell and Bosch, 1959). There were two
exceptions. First, the low concentration tested by Salaman and Glendenning) caused promotion as
well as "transient light crusting." Because the site of the weekly treatment was rotated across four
sites on the body, it is unclear whether more severe effects would have been observed if the same site
had been treated for the entire study. The second exception was that the low concentration tested by
Wynder and Hoffmann (1961) was reported as causing no toxicity, although the sensitivity of the
evaluation is unclear. On the basis of the high observed toxicity, it is not clear whether the promoting
activity observed for phenol in several studies was secondary to the repeated injury and healing of the
skin. From these considerations, the data regarding the carcinogenic potential of phenol are
inadequate for an assessment of human carcinogenic potential.
6.2. DOSE-RESPONSE
No human data that are adequate for the derivation of a phenol RfD were located. Therefore,
laboratory animal data were used.
The RfD of 0.3 mg/kg-day was based on a BMDL of 93 mg/kg-day for decreased maternal
body weight gain in a gavage rat developmental toxicity study that used a divided-dosing protocol
107
-------
(Argus Research Laboratories, 1997). There was a corresponding NOAEL of 60 mg/kg-day and a
LOAEL for maternal toxicity of 120 mg/kg-day. A composite UF of 300 was used. This factor is
based on a default factor of 10 for extrapolation from laboratory animals to humans, a default factor
of 10 to account for intrahuman variability, and a factor of 3 to account for database insufficiencies.
Although the database for phenol can be considered "complete," there are uncertainties
regarding the immunotoxicity potential of phenol in light of the immunotoxicity (decreased antibody
forming cells) reported by Hsieh et al. (1992). The database factor may be reconsidered with results
from an immunotoxicity study in mice that is compliant with EPA immunotoxicity test guidelines
(U.S. EPA, 1998c). This RfD is at least twice the endogenous rate of phenol formation in humans,
estimated as 0.014-0.14 mg/kg-day (Bone et al., 1976; Lawrie and Renwick, 1987; Renwick et al.,
1988), based on total phenol (free plus conjugated) levels in urine. This means that endogenous
production is approximately 5-50% of the RfD.
Note also that the RfD is meant to apply to ingested phenol in addition to the endogenous
formation of phenol. Endogenous phenol is produced by bacteria in the gut, so endogenous phenol
and ingested phenol would have similar toxicokinetics. Both humans and laboratory animals
efficiently conjugate and excrete phenol at low doses, resulting in only a small degree of systemic
exposure to free phenol (or any of its oxidative metabolites) at these low levels. The primary
difference between endogenous and exogenous phenol would result, because endogenous phenol is
formed in the intestines and some phenol may reach the colon and rectum, where some will escape
the hepatic portal circulation and be absorbed directly without conjugation. By contrast, a smaller
amount of ingested phenol would be expected to reach the colon and rectum.
The data are insufficient to determine the degree of conjugation of endogenously formed
phenol in humans because the available data are based on analysis of daily urinary excretion of total
phenol (i.e., phenol conjugates plus any trace amounts of free phenol) (Lawrie and Renwick, 1987;
Renwick et al., 1988); oxidative metabolites of phenol were not measured. The phenol conjugation
capacity of the liver is an important determinant of the ingested dose that would result in toxicity, but
there is no information on the degree of phenol conjugation by humans at doses in the range of the
RfD. Human variability exists in both the levels of endogenous phenol production and in the
conjugative capacity of the liver.
In the absence of more detailed information, it is reasonable to assume that humans have
adapted by having adequate conjugation capacity for the range of endogenous phenol production.
Therefore, the default total uncertainty factor of 10 for human variability in toxicokinetics and
108
-------
toxicodynamics is considered adequate. Determining whether oxidative metabolites are formed in
individuals who have high endogenous levels of phenol formation would enhance the confidence in
determining the intraspecies uncertainty factor.
The principal study (Argus Research Laboratories, 1997) used an adequate number of animals
and evaluated an appropriate array of endpoints for a developmental toxicity study. Although gavage
dosing was used, the divided-dosing protocol provided a significant enhancement that made the
gavage dosing more closely resemble an environmentally relevant route of exposure. The principal
study is judged to have medium confidence. Although the use of gavage dosing lowers the
confidence in the study, the dosing frequency may be fairly similar to that in drinking water studies,
in which rodents typically consumed water in a few larger doses and often in association with food
consumption.
Confidence in the supporting database is medium to high. Although the oral toxicity database
meets the minimal criteria for a high-confidence database (chronic studies in two species,
developmental toxicity studies in two species, and a multigeneration reproduction study), the chronic
studies did not evaluate a sufficient array of endpoints. In particular, the chronic mouse study (NCI,
1980) did not evaluate hematological and immunological effects, making interpretation of the results
of the Hsieh et al. (1992) study difficult. Consideration of the above issues results in medium to high
confidence in the RfD.
The RfD developed in this document can be compared with other limits on phenol exposure,
partially as a test of the reasonableness of the RfD. Phenol is used in a number of industrial products,
as well as in over-the-counter medicines such as cough drops, throat sprays, and mouthwashes (e.g.,
Cepastat® and Chloraseptic® brands). Use of these consumer products can result in short-term, high-
level phenol exposures, but prolonged exposure (more than a week) at these levels is not
recommended. The short duration of exposures to the cough medicines suggests that safe exposure
levels would be higher than those for lifetime exposure. On the other hand, at least some aspects of
phenol toxicity appear to be related to peak concentrations in blood, and higher peak blood
concentrations could result from the consumer product exposure.
ATSDR (1998) estimated that intake of the maximum recommended dosage of 300 mg
phenol/day would result in an approximate dose of 4-8 mg/kg-day. No documentation of this
maximum recommended dosage could be located. Use of these products on a daily basis over the
course of a lifetime would result in a dose approximately 10-fold higher than the RfD derived in this
document; however, these products are not intended for use over a prolonged period. Evaluation of
109
-------
potential health effects in individuals who do consume these products in large amounts or over long
periods of time could provide additional information about human health effects of phenol and safe
exposure levels.
The use of a higher dose of phenol than the RfD in over-the-counter medicines suggests that
this RfD provides an adequate degree of public health protection. However, the maximum
recommended dosage may not be protective of pregnant women and fetuses, as consumers are
advised "as with any drug, if you are pregnant or nursing a baby, seek the advice of a medical
professional before using this drug."
Although a substantial amount of data on phenol toxicokinetics are available, they are not
sufficient to move away from the default UFs for interspecies extrapolation and intraspecies
variability (IPCS, 2001). Data on how blood levels of phenol and its metabolites relate to doses
in rats and humans would be useful in addressing the interspecies UF, as would data on the potential
for phenol to cross the placenta. Similarly, data on how differences in enzyme activities relate to
phenol and metabolite blood levels would be useful in addressing intrahuman variability. Finally, a
drinking water study compliant with EPA test guidelines (U.S. EPA, 1998c) that evaluated
hematological effects and immunological effects in mice could address the uncertainties associated
with the Hsieh et al. (1992) study and lead to reconsideration for the need for a database UF.
The available data are inadequate to derive an RfC. As noted above, a 90-day inhalation
study that evaluated the respiratory tract would be necessary for development of an RfC.
Because the data were considered inadequate to assess the carcinogenicity of phenol, no
quantitative assessment was conducted.
110
-------
7. REFERENCES
ACGIH. 1991. American Conference of Governmental Industrial Hygienists, Inc. Documentation of
the Threshold Limit Values and biological exposure indices. Cincinnati, OH.
Argus Research Laboratories, Inc. 1997. Oral (gavage) developmental toxicity study of phenol in
rats. Horsham, Pennsylvania.. Protocol number: 916-011.
Agency for Toxic Substances and Disease Registry (ATSDR). 1998. Toxicological profile for
phenol. Update. Atlanta, GA.
Baj, Z., E. Majewska, K. Zeman, et al. 1994. The effect of chronic exposure to formaldehyde,
phenol and organic chlorohydrocarbons on peripheral blood cells and the immune system in humans.
Journal of Investigational Allergology & Clinical Immunology. 4(4): 186-191.
Baker, E.L., P.J. Landrigan, P.E. Bertozzi, et al. 1978. Phenol poisoning due to contaminated
drinking water. Arch Environ Health. 33:89-94.
Ballinger, L.N., S.E. Cross and M.S. Roberts. 1995. Availability and mean transit times of phenol
and its metabolites in the isolated perfused rat liver: normal and retrograde studies using tracer
concentrations of phenol. J Pharm Pharmacol. 47(11): 949-56.
Barale, R., A. Marrazzini, C. Betti, et al. 1990. Genotoxicity of two metabolites of benzene: Phenol
and hydroquinone show strong synergistic effects in vivo. Mutat Res. 244(1): 15-20.
Baranowska-Dutliewicz, B. 1981. Skin absorption of phenol from aqueous solutions in men. Int
Arch Occup Environ Health. 49: 99-104.
Beckmann, J.D., J.F. Spurzem, and S.I. Rennard. 1995. Phenol sulfotransferase expression in the
airways: Enzymological and immunohistochemical demonstration. Biochem Biophys Res Commun.
213: 104-111.
Bentur, Y., O. Shoshani, A. Tabak, et al. 1998. Prolonged elimination half-life of phenol after
dermal exposure. J Toxicol Clin Toxicol. 36(7): 707-11.
Berman, E., M. Schlicht, V.C. Moser, et al. 1995. A multi-disciplinary approach to toxicological
screening: I. Systemic toxicity. J Toxicol Environ Health. 45: 127-143.
Bishop, J.B., R.W. Morris, J.C. Seely et al. 1997. Alterations in the reproductive patterns of female
mice exposed to xenobiotics. Fund, and Appl. Tox. 40: 191-20
Bohrman, J.S., J.R. Burg, E. Elmore, et al. 1988. Interlaboratory Studies With the Chinese Hamster
V79 Cell Metabolic Cooperation Assay to Detect Tumor-Promoting Agents. Environ Mol Mutagen.
12:33-51.
Bois, F.Y., M.T. Smith, and R.C. Spear. 1991. Mechanisms of benzene carcinogenesis: Application
of a physiological model of benzene pharmacokinetics and metabolism. Toxicol Lett. 56: 283-298.
Ill
-------
Bone, E., A. Tamm and M. Hill. 1976. The production of urinary phenols by gut bacteria and their
possible role in the causation of large bowel cancer. Am. J. Clin. Nutr. 29:1448-1454.
Boutwell, R.K. and D.K. Bosch. 1959. The tumor-promoting action of phenol and related compounds
for mouse skin. Cancer Res. 19: 413-424.
Bruce, R.M., J. Santodonato and M.W. Neal. 1987. Summary review of the health effects associated
with phenol. Toxicol Ind Health. 3(4): 535-568.
Bulsiewicz, H. 1977. The influence of phenol on chromosomes of mice (Mus musculus) in the
process of spermatogenesis. Folia Morphol(Warsaw). 36:13-22.
Campbell, B.A. and G.A. Cicala. 1962. Studies of water deprivation in rats as a function of age. J.
Comp. Physiol. Psychol. 55: 763-768.
Campbell N., J.A. Van Loon, and R.M. Weinshilboum. 1987. Human liver phenol sulfotransferase:
Assay conditions, biochemical properties and partial purification of isozymes of the thermostable
form. Biochem. Pharmacol. 36: 1435-1446.
Capel, I.D., M.R. French, P. Millburn, et al. 1972. Fate of C-14-phenol in various species.
Xenobiotica. 2: 25-34.
Carpenter, S.P., J.M. Lasker, and J.L. Raucy. 1996. Expression, induction, and catalytic activity of
the ethanol-inducible cytochrome P450 (CYP2E1) in human fetal liver and hepatocytes. Mol.
Pharmacol. 49: 260-268.
Cassidy, M.K., and J.B. Houston. 1984. In vivo capacity of hepatic and extra hepatic enzymes to
conjugate phenol. Drug Metab Dispos. 12: 619-624.
Chapman, D.E., M.J. Namkung and M.R. Juchau. 1994. Benzene and benzene metabolites as
embryotoxic agents: effects on cultured rat embryos. Toxicol Appl Pharmacol. 128(1): 129-37.
Charles River Laboratories. 1986. Technical bulletin: Baseline hematology and clinical chemistry
values as a function of sex and age for Charles River Outbred Mice: Crl: CD-1(ICR)BR, Crl: CF-1
BR. Summer 1986.
Charles River Laboratories. 1988. Embryo and fetal development toxicity (teratology) control data
in the Charles River Crl:CD BR rat. Feb. 1988.
Charles River Laboratories. 1996. Historical control data (1992-1994) for developmental and
reproductive toxicity studies using the Crl:CD (SD)BR rat. Downloaded: December 26, 2001.
Available online at http://www.criver.com/techdocs/toxdata.hctab03.html.
Chen, H. and D.A. Eastmond. 1995a. Synergistic increase in chromosomal breakage within the
euchromatin induced by an interaction of the benzene metabolites phenol and hydroquinone in mice.
Carcinogenesis. 16(8): 1963-9.
112
-------
Chen, H. and D.A. Eastmond. 1995b. Topoisomerase inhibition by phenolic metabolites: A potential
mechanism for benzene's clastogenic effects. Carcinogenesis. 16: 2301-2307.
Cirranni, R., R. Barale, and A. Marrazzini. 1988a. Benzene and the genotoxicity of its metabolites. I.
Transplacental activity in mouse fetuses and in their dams. Mutat. Res. 208: 61-67.
Ciranni, R., R. Barale, G. Ghelardini, et al. 1988b. Benzene and the genotoxicity of its metabolites.
II. The effect of the route of administration on the micronuclei and bone marrow depression in mouse
bone marrow cells. Mutat Res. 209(1-2): 23-8.
ClinTrials BioResearch. 1998. A 13-week neurotoxicity study of phenol administered in the
drinking water to the rat. Volumes 1 and 2. Senneville, Quebec, Canada. Project I.D.: 97439. P.
Beyrouty, study director.
Conning, D.M. and M.J. Hayes. 1970. The dermal toxicity of phenol: an investigation of the most
effective first-aid measures. Br J Ind Med. 27:155-159. As cited in ATSDR, 1998
Corti, M. and C.A. Snyder. 1998. Gender- and age- specific cytotoxic susceptibility to benzene
metabolites in vitro. Toxicol Sci. 41:42-48.
Dalin, N.M., and R. Kristoffersson. 1974. Physiological effects of a sublethal concentration of
inhaled phenol on the rat. Ann Zool Fenn. 11: 193-199.
De Ceaurriz, J.C., J.C. Micillino, P. Bonnet, et al. 1981. Sensory irritation caused by various
industrial airborne chemicals. Toxicol Lett (Amst). 9:137-144.
Deichmann , W.B. and M.L. Klepinger. 1981. Phenols and phenolic compounds. In: Clayton, G.D.
and F.E. Clayton, eds. Pattys' industrial hygiene and toxicology, 3rd ed. New York, NY: John Wiley
and Sons, Inc, 2567-2627. As cited in ATSDR, 1998
Deichmann, W.B and B. Witherup. 1944. The acute and comparative toxicity of phenol and o-, m-,
and p-creosols for experimental animals. J. Pharmacol. Exp. Ther. 80: 233-240.
Deichmann, W.B., K.V. Kitzmiller, and B.S. Witherup. 1944. Phenol Studies. VII. Chronic phenol
poisoning, with special reference to the effects upon experimental animals of the inhalation of phenol
vapor. Am J Clin Pathol. 14:273-277.
Dosemeci, M., A. Blair, P.A. Stewart, et al. 1991. Mortality among industrial workers exposed to
phenol. Epidemiology. 2(3): 188-93.
Dow Chemical Co. 1994. Pharmacokinetics, metabolism, and distribution of C 14 phenol in Fischer
344 rats after gavage, drinking water, and inhalation exposure, with cover letter dated 07/13/1994.
U.S. EPA/OPTS Public Files Fiche # OTS0557473; Doc#: 86940001296. Hiser, M.F., B.E.
Kropscott, R.J. McGuirk, and J.S. Bus, authors.
Dow Chemical Co. 1945. The toxicity of phenol. Biochem. Res. Lab. Unpublished report dated
04/12/1945. U.S. EPA/OPTS Public Files Fiche # OTS 0517006; Doc#: 86870002216.
113
-------
Eastmond, D.A., M.T. Smith and R.D. Irons. 1987. An interaction of benzene metabolites
reproduces the myelotoxicity observed with benzene exposure. Toxicol Appl Pharmacol. 91(1): 85-
95.
Eastmond, D.A., M.T. Smith, L.O. Ruzo, et al. 1986. Metabolic activation of phenol by human
myeloperoxidase and horseradish peroxidase. Mol Pharmacol. 30: 674-679.
Edwards, V.T., B.C. Jones, D.H. Hutson. 1986. A comparison of the metabolic fate of phenol,
phenyl glucoside and phenyl 6-o-malonyl-glucoside in the rat. Xenobiotica 16: 801-807.
Erexson, G.L., J.L. Wilmer, and A.D. Kligerman. 1985. Sister chromatid exchange induction in
human lymphocytes exposed to benzene and its metabolites in vitro. Cancer Res. 45: 2471-2477.
Flickinger, C.W. 1976. The benzenediols: catechol, resorcinol, and hydroquinone - a review of the
industrial toxicology and current industrial exposure limits. Am Ind Hyg Assoc J. 37:596-506.
Gad-el-Karim, M.M., V.M. Ramanujam, A.E. Ahmed, et al. 1985. Benzene myeloclastogenicity: a
function of its metabolism. American Journal Of Industrial Medicine. 7:5-6.
Ghantous, H., and B.R. Danielsson. 1986. Placental transfer and distribution of toluene, xylene and
benzene, and their metabolites during gestation in mice. Biol Res Pregnancy Perinatol. 7: 98-105.
Gilmour, S.K., G.F. Kalf and R. Snyder. 1986. Comparison of the metabolism of benzene and its
metabolite phenol in rat liver microsomes. Adv Exp Med Biol. 197(35): 223-35.
Gocke, E., M.-T. King, K. Eckhardt and D. Wild. 1981. Mutagenicity of cosmetics ingredients
licensed by the European communities. Mutat. Res. 90: 91-109.
Hakkola, J., O. Pelkonen, M. Pasanen and H. Raunio. 1998. Xenobiotic-metabolizing cytochrome
P450 enzymes in the human feto-placental unit: role in intrauterine toxicity. Crit Rev Toxicol
28(l):35-72.
Haworth, S., T. Lawlor, K. Mortelmans, W. Speck, and E. Zeiger. 1983. Salmonella mutagenicity
test results for 250 chemicals. Environ. Mutagen. 5(Suppl 1): 3-142.
Heaton, G.D., and A.G. Renwick. 1991. The effects of high dietary concentrations of sodium
saccharin on in vivo metabolism of xenobiotics in rats. Fund Chem Toxic. 29: 305-312.
Hogg, S.I., C.G. Curtis, D.G. Upshall, et al. 1981. Conjugation of phenol by rat lung. Biochem
Pharmacol. 30(12): 1551-5.
Hoffmann, M.J., S. Ji, C.C. Hedli, and R. Snyder. 1999. Metabolism of [14C]phenol in the isolated
perfused mouse liver. Toxicol. Sci. 49:40-47
Hoffman, G.M., B.J. Dunn, C.R. Morris, J.H. Butala, S.S. Dimond, R.Gingell, and J.M. Waechter, Jr.
2001. Two-week (ten-day) inhalation toxicity and two-week recovery study of phenol vapor in the
rat. Int. J. Toxicol. 20:45-52
114
-------
Hotchkiss, S.A.M., P. Hewitt, and J. Caldwell. 1992. Percutaneous absorption of nicotinic acid,
phenol, benzoic acid and triclopyr butoxyethyl ester through rat and human skin in vitro: Further
validation of an in-vitro data. Food Chem. Toxicol. 30: 891-899.
HSDB. 1999. Hazardous Substances Data Bank. National Library of Medicine, National
Toxicology Information Program, Bethesda, MD.
Hsieh, G.C., R.P. Sharma, R.D. Parker, et al. 1992. Immunological and neurobiochemical alterations
induced by repeated oral exposure of phenol in mice. European Journal Of Pharmacology. 228: 107-
114.
Hughes, M.F. and L.L. Hall. 1995. Disposition of phenol in rat after oral, dermal, intravenous, and
intratracheal administration. Xenobiotica. 25(8): 873-83.
Humphrey, M.J., C.W. Filer, D.J. Jeffery, et al. 1980. The availability of carfecillin and its phenol
moiety in rat and dog. Xenobiotica. 10:771-778.
Huntingdon. 1998. Two-week (ten day) inhalation toxicity and two-week recovery study of phenol
vapor in the rat. Chemical Manufacturers Association. CMA Reference No. PHL-4.0-INHAL-HLS.
U.S. EPA/OPTS Public Files Fiche #: OTS0559328; Doc#: 40-980000008.
IARC. 1999. Monographs on Evaluation of Carcinogenic Risks to Humans: Re-evaluation of some
organic chemicals, hydrazine, and hydrogen peroxide. Vol. 71. WHO International Agency for
Research on Cancer. Lyon France.
IIT Research Institute. 1999. Two-generation oral (drinking water) reproductive toxicity study of
phenol in rats. Volume 1. Chicago, IL. IITRI Project No. L08657, Study No. 2.
International Research and Development Corporation. 1993. Range-finding maternal toxicity study
with phenol in rats. US EPA/OTS#: OTS0537777. Doc#: 86-930000341.
IPCS. 2001. Guidance document for the use of chemical-specific adjustment factors (CSAFs) for
interspecies differences and human variability in dose/concentration-response assessment. Available
at: http://www.ipcsharmonize.org.
Ivett, J.L., B.M. Brown, C. Rodgers, B.E. Anderson, M.A. Resnick, and E. Zeiger. 1989.
Chromosomal aberration and sister chromatid exchange tests in Chinese hamster ovary cells in vitro
IV: Results for 15 chemicals. Environ. Molec. Mutagen. 14: 165-187.
Iwasaki, K., T. Shiraga, K. Takeshita, et al. 1993. Perinatal development of amine, alcohol and
phenol sulfoconjugations in the rat. Res Commun Chem Pathol Pharmacol. 81(2): 183-90.
Jansson, T., M. Curvall, A. Hedin, et al. 1986. In vitro studies of biological effects of cigarette
smoke condensate. II. Induction of sister chromatid exchanges in human lymphocytes by weakly
acidic, semivolatile constituents. Mut. Res. 169: 129-139.
Jarvis, S.N., R.C. Straube, A.L.J. Williams, et al. 1985. Illness Associated With Contamination of
115
-------
Drinking Water Supplies With Phenol. Br. Med J. 290: 1800-1802.
Kalf, G., R. Shurina, J. Renz, et al. 1990. The role of hepatic metabolites of benzene in bone marrow
peroxidase-mediated myelo- and genotoxicity. Adv. Exp. Med. Biol. 283(55): 443-55.
Kao, J., J.W. Bridges, and J.K. Faulkner. 1979. Metabolism of C-14-phenol by sheep, pig, rat.
Xenobiotica. 9: 141-147.
Kauppinen, T.P., T.J. Partanen, N.M. Nurminen. 1986. Respiratory cancers and chemical exposures
in the wood industry: A nested case-control study. Br. J. Ind. Med.. 43: 84-90.
Kawamoto, T., M. Koga, T. Oyama, et al. 1996. Habitual and genetic factors that affect urinary
background levels of biomarkers for organic solvent exposure. Arch Environ Contam Toxicol 30:
114-120.
Kenyon, E.M., M.E. Seeley, D. Janszen, et al. 1995. Dose-, route-, and sex-dependent urinary
excretion of phenol metabolites in B6C3F1 mice. J Toxicol Environ Health. 44(2): 219-33.
Kenyon, E.M., M.J. Seaton, M.W. Himmelstein, et al. 1998. Influence of gender and acetone
pretreatment on benzene metabolism in mice exposed by nose-only inhalation. J Toxicol Environ
Health. 55(6): 421-43.
Kim, D.H., S.K. Lee, B.Y. Chun, et al. 1995. Illness associated with contamination of drinking
water supplies with phenol. Journal Of Korean Medical Science. 9(3): 218-23.
Kolachana, P., V.V. Subrahmanyam, K.B. Meyer, et al. 1993. Benzene and its phenolic metabolites
produce oxidative DNA damage in HL60 cells in vitro and in the bone marrow in vivo. Cancer
Research 53(5): 1023-6.
Koop, D.R., C.L. Laethem and G.G. Schnier. 1989. Identification of ethanol-inducible P450
isozyme 3a (P450IIE1) as a benzene and phenol hydroxylase. Toxicol Appl Pharmacol. 98(2): 278-
Koster, H., I. Halsema, E. Scholtens, et al. 1981. Dose-dependent shifts in the sulfation and
glucuronidation of phenolic compounds in the rat in vivo and in isolated hepatocytes. the role of
saturation of phenolsulfotransferase. Biochem Pharmacol. 30: 2569-2575.
Lawrie, C.A. and A.G. Renwick. 1987. The effect of saccharin ingestion on the excretion of
microbial amino acid metabolites in rat and man. Toxicol. Appl. Pharmacol. 91(3): 415-428.
Legathe, A., B.A. Hoener and T.N. Tozer. 1994. Pharmacokinetic interaction between benzene
metabolites, phenol and hydroquinone, in B6C3F1 mice. Toxicol Appl Pharmacol. 124(1): 131-8.
Liao, T.F. and F.W. Oehme. 1981. Tissue distribution and plasma protein binding of carbon-14-
labeled phenol in rats. Toxicol Appl Pharmacol. 57(2): 220-225.
Lide, D.R., ed. 1993. CRC Handbook of Chemistry and Physics. 73rd edition. CRC Press.
116
-------
Lunte, S.M. and P.T. Kissinger. 1983. Detection and identification of sulfhydryl conjugates ofrho-
benzoquinone in microsomal incubations of benzene and phenol. Chem Biol Interact. 47(2): 195-
212.
Luster, M.I., C. Portier, D.G. Pait, K.L. White, C. Gennings, A.E. Munson and G.J. Rosenthal. 1992.
Risk assessment in immunotoxicology. I. Sensitivity and predictability of immune tests. Fund. Appl.
Toxicol. 18:200-210.
Luster, M.I., C. Portier, D.G. Pait, G.J. Rosenthal, D.R. Germolec, E. Corsini, B.L. Blaylock, P.
Pollack, Y. Kouchi, W. Craig, K.L. White, A.E. Munson and C.E. Comment. 1993. Risk assessment
in immunotoxicology. II. Relationships between immune and host resistance tests. Fund. Appl.
Toxicol. 21:71-82.
MacPhail, R.C., E. Berman, J.A. Elder, et al. 1995. A multidisciplinary approach to toxicological
screening: IV. Comparison of results. Journal Of Toxicology And Environmental Health. 45(2):
211-20.
Marrazzini, A., L. Chelotti, I. Barrai, et al. 1994. In vivo genotoxic interactions among three
phenolic benzene metabolites. Mutation Research. 341(1): 29-46.
McGregor, D.B., A. Brown, P. Cattanach, et al. 1988a. Responses of the L5178Y TK+/TK- Mouse
Lymphoma Cell Forward Mutation Assay. 3. 72 Coded Chemicals. Environ Mol Mutagen. 12:85-
154.
McGregor, D.B, C.G. Rlach, A. Brown, et al. 1988b. Reactivity of catecholamines and related
substances in the mouse lymphoma L5178Y cell assay. Environ Molec Mutagen. 11: 523-544.
Medinsky, M.A., E.M. Kenyon and P.M. Schlosser. 1995. Benzene: a case study in parent chemical
and metabolite interactions. Toxicology. 105(2-3): 225-33.
Meerman, J.H.N., C. Nijland and G.J. Mulder. 1987. Sex differences in sulfation and glucuronidation
of phenol, 4-nitrophenol and N-hyrdoxy-2-acetylaminofiourene in the rat in vivo. Biochem
Phamacol. 36(16):2605-2608.
Mehta, R., P.C. Hirom, and P. Millburn. 1978. The influence of dose on the pattern of conjugation
of phenol and 1-naphtol in nonhuman primates. Xenobiotica . 8: 445-452.
Merliss, R.R. 1972. Phenol marasmus. Occup Med. 14:55-56.
Miller, B.M., E. Pujadas and E. Gocke. 1995. Evaluation of the micronucleus test in vitro using
Chinese hamster cells: results of four chemicals weakly positive in the in vivo micronucleus test.
Environmental and Molecular Mutagenesis. 26(3): 240-7.
Miyagawa, M., H. Takasawa, A. Sugiyama, et al. 1995. The in vivo-in vitro replicative DNA
synthesis (RDS) test with hepatocytes prepared from male B6C3F1 mice as an early prediction assay
for putative nongenotoxic (Ames'-negative) mouse hepatocarcinogens. MutRes. 343: 157-183.
Morimoto, K., and S. Wolff. 1980. Increase of sister chromatid exchanges and perturbations of cell
117
-------
division kinetics in human lymphocytes by benzene metabolites. Cancer Res. 40: 1180-1193.
Morimoto, K., S. Wolff, and A. Koizumi. 1983. Induction of sister-chromatid exchanges in human
lymphocytes by microsomal activation of benzene metabolites. Mutat Res. 119: 355-360.
Morrison, J.E., Jr., D. Matthews, R. Washington, et al. 1991. Phenol motor point blocks in children:
plasma concentrations and cardiac dysrhythmias. Anesthesiology. 75(2): 359-62.
Moser, V.C., B.M. Cheek and R.C. MacPhail. 1995. A multidisciplinary approach to toxicological
screening: III. Neurobehavioral toxicity. J Toxicol Environ Health. 45(2): 173-210.
Narotsky, M.G. and R.J. Kavlock. 1995. A multidisciplinary approach to toxicological screening: II.
Developmental toxicity. J Toxicol Environ Health. 45(2): 145-71.
NCI (National Cancer Institute). 1980. Bioassay of phenol for possible carcinogenicity in F344 rats
and B6C3F1 mice. Prepared by the National Cancer Institute, Bethesda, MD for the National
Toxicology Program, Research Triangle Park, NC. NCI-CG-TR-203, DHHS/PUB/NIH80-1759.
National Research Council (NRC). 1983. Risk Assessment in the Federal Government: Managing the
Process. National Academy Press.
NTP (National Toxicology Program). 1983a. Teratologic evaluation of phenol in CD rats. Report
prepared by Research Triangle Institute, Research Triangle Park, NC. NTIS PB83-247726.
NTP (National Toxicology Program). 1983b. Teratologic evaluation of phenol in CD-I mice. Report
prepared by Research Triangle Institute, Research Triangle Park, NC. NTIS PB85104461.
NTP (National Toxicology Program). 1986. Toxicology and carcinogenesis studies of benzene (CAS
No. 71-43-2) in F344/N rats and B6C3F1 mice (gavage studies). NTIS PB86-216967/AS
NTP (National Toxicology Program). 2000a. 9th Report on Carcinogens. 1998 Summary. U.S.
Department of Health and Human Services, Public Health Service.
NTP (National Toxicology Program). 2000b. Information available on-line; downloaded February,
2000 from http://ntp-server.niehs.nih.gov/NewHomeRoc/AboutRoC.html
Ogata, M., Y. Yamasaki and T. Kawai. 1986. Significance of urinary phenyl sulfate and phenyl
glucuronide as indices of exposure to phenol. International Archives of Occupational and
Environmental Health. 58(3): 197-202.
Ohtsuji, H., and M. Ikeda. 1972. Quantitative relation between atmospheric phenol vapor and vapor
in the urine of workers in Bakelite factories. Br J Ind Med. 29: 70-73.
Paschin Yu, V. and L.M. Bahitova. 1982. Mutagenicity of benzo[a]pyrene and the antioxidant
phenol at the HGPRT locus of V79 Chinese hamster cells. Mutat. Res. 104(6): 389-93.
Patrick, E., H.I. Maibach and A. Burkhalter. 1985. Mechanisms of chemically induced skin irritation.
118
-------
Toxicol Appl Pharmacol. 81:476-490.
Pellack-Walker, P. and J.L. Blumer. 1986. DNA damage in L5178YS cells following exposure to
benzene metabolites. Molecular Pharmacology. 30(1): 42-7.
Piotrowski, J.K. 1971. Evaluation of exposure to phenol: Absorption of phenol vapor in the lungs
through the skin and excretion of phenol in urine. BrJIndMed. 28: 172-178.
Pool, B.L. and P.Z. Lin. 1982. Mutagenicity testing in the Salmonella typhimurium assay of phenolic
compounds and phenolic fractions obtained from smokehouse smoke condensates. Food Chem.
Toxicol. 20:383-391.
Post, G., R. Snyder and G.F. Kalf. 1986. Metabolism of benzene and phenol in macrophages in vitro
and the inhibition of RNA synthesis by benzene metabolites. Cell Biol Toxicol. 2(2): 231-46.
Powell, G.M., J.J. Miller, A.H. Olavesen, et al. 1974. Liver as a major organ of phenol
detoxification. Nature(London). 252: 234-235.
Pullin, T.G., M.N. Pinkerton, R. V. Johnston, and D.J. Kilian. 1978. Decontaminationi of the skin of
swine following phenol exposure: a comparison of the relative efficiency of water versus
polyethylene glycol/industrial methylated spirits. Toxicol Appl Pharmacol. 43:199-206.
Rapson, W.H., M.A. Nazar, and V. Butsky. 1980. Mutagenicity produced by aqueous chlorination
of organic compounds. Bull. Environ. Contam. Toxicol 24:590-596.
Reddy, M.V., W.T. Bleicher, G.R. Blackburn, et al. 1990. DNA adduction by phenol, hydroquinone,
or benzoquinone in vitro but not in vivo: nuclease PI-enhanced 32P-postlabeling of adducts as
labeled nucleoside bisphosphates, dinucleotides and nucleoside monophosphates. Carcinogenesis.
11(8): 1349-57.
Renwick, A.G., A. Thakrar, C.A. Lawrie and C.F. George. 1988. Microbial amino acid metabolites
and bladder cancer: No evidence of promoting activity in man. Human Toxicol. 7:267-272.
Ryan, B.M., R. Selby, R. Gingell, J.M. Waechter, Jr., J.H. Butala, S.S. Dimond, B.J. Dunn, R. House,
and R. Morrissey. 2001. Two-generation reproduction study and immunotoxicity screen in rats
dosed with phenol via the drinking water. Inter. J. Toxicol. 20:121-142.
Salaman, M.H. and O.M. Glendenning. 1957. Tumor promotion in mouse skin by sclerosing agents.
Br. J. Cancer. 11:434-444.
Sandage, C. 1961. Tolerance criteria for continuous inhalation exposure to toxic material. I. effects
on animals of 90-day exposure to phenol, CC14, and a mixture of indole, skatole, hydrogen sulfide,
and methyl mercaptan. Wright Patterson Air Force Base, HO. U.S Air Force systems command,
Aeronautical Systems Division, ASD technical report 61-519(1).
Sawahata, T. and R.A. Neal. 1983. Biotransformation of phenol to hydroquinone and catechol by rat
liver microsomes. Mol Pharmacol. 23(2): 453-60.
119
-------
Schenkman, J.B., K.E. Thummel, and L.V. Favreau. 1989. Physiological and patho-physiological
alterations in rat hepatic cytochrome P-450. Drug Met. Rev. 20(2-4): 557-584.
Schlicht, M.P., V.C. Moser, B.M. Sumrell, et al. 1992. Systemic and neurotoxic effects of acute and
repeated phenol administration (Abstract No. 1047). Toxicologist. 12:274.
Schlosser, P.M., J.A. Bond and M.A. Medinsky. 1993. Benzene and phenol metabolism by mouse
and rat liver microsomes. Carcinogenesis. 14(12): 2477-86.
Seaton, M.J., P. Schlosser and M.A. Medinsky. 1995. In vitro conjugation of benzene metabolites
by human liver: potential influence of interindividual variability on benzene toxicity.
Carcinogenesis. 16(7): 1519-27.
Selgrade, M.K. 1999. Use of immunotoxicity data in health risk assessments: Uncertainties and
research to improve the process. Toxicology 133:59-72.
Shamy, M.Y., R.M. El Gazzar, M.A. El Sayed, et al. 1994. Study of some biochemical changes
among workers occupationally exposed to phenol, alone or in combination with other organic
solvents. Industrial Health. 32(4): 207-214.
Skare, J.A. and K.R. Schrotel. 1984. Alkaline elution of rat testicular DNA: Detection of DNA
strand breaks after in vivo treatment with chemical mutagens. Mutat. Res. 130: 283-294.
Smart, R.C. and V.G. Zannoni. 1984. DT-diaphorase and peroxidase influence the covalent binding
of the metabolites of phenol, the major metabolite of benzene. Mol Pharmacol 26: 105-111.
Snyder, R., T. Chepiga, C.S. Yang, et al. 1993. Benzene metabolism by reconstituted cytochromes
P450 2B1 and 2E1 and its modulation by cytochrome b5, microsomal epoxide hydrolase, and
glutathione transferases: Evidence for an important role of microsomal epoxide hydrolase in the
formation of hydroquinone. Toxicol Appl Pharmacol 122(2): 172-81.
Song, B-J, H.V. Gelboin, S-S. Park, C.S. Yang, and F.J. Gonzalez. 1986. Complementary DNA and
protein sequences of ethanol-inducible rat and human cytochrome P-450s. J. Biol. Chem. 261(35):
16689-16697.
Spalding, J.W., J. Momma, M.R. Elwell, et al. 1993. Chemically induced skin Carcinogenesis in a
transgenic mouse line (TG.AC) carrying a v-Ha-ras gene. Carcinogenesis 14(7): 1335-41.
Stenius, U., M. Warholm, A. Rannug, et al. 1989. The role of GSH depletion and toxicity in
hydroquinone-induced development of enzyme-altered foci. Carcinogenesis. 10(3): 593-599.
Sturtevant, P.M. 1952. Studies on the mutagenicity of phenol in Drosophila melanogaster. J Hered.
43: 217-220.
Subrahmanyam, V.V. and P.J. O'Brien. 1985. Phenol oxidation product(s), formed by a peroxidase
reaction, that bind to DNA. Xenobiotica. 15(10): 873-85.
120
-------
Sze, C-C., C-Y Shi, and C-N Ong. 1996. Cytotoxicity and DNA strand breaks induced by benzene
and its metabolites in Chinease Hamster ovary cells. J Appl Toxicol. 16: 259-264.
Tanaka, T., K. Kasai, T. Kita, et al. 1998. Distribution of phenol in a fatal poisoning case
determined by gas chromatography/mass spectrometry. J Forensic Sci. 43(5): 1086-8.
Thompson, E.D., and D.P. Gibson. 1984. A method for determining the maximum tolerated dose for
acute in vivo cytogenic studies. Food Chem Toxicol. 22(8): 665-676.
Tremaine, L.M., G.L. Diamond, and A.J. Quebbemann. 1984. In vivo quantification of renal
glucuronide and sulfate conjugation of 1-naphtol and p-nitrophenlyphenol in the rat. Biochem
Pharmacol. 33:419-427.
Tsutsui, T., N. Hayashi, H. Maizumi, et al. 1997. Benzene-, catechol-, hydroquinone- and phenol-
induced cell transformation, gene mutations, chromosome aberrations, aneuploidy, sister chromatid
exchanges and unscheduled DNA synthesis in Syrian hamster embryo cells. Mutation Research.
373(1): 113-23.
UBTL. Utah Biomedical Test Laboratory. 1991. A Utah biomedical test laboratory report on NIOSH
sponsored inhalation study for IDLH values (final report). Study sponsor: Shell Oil company. Fiche
No. OTS 0534605. TSCATS 88-920000156.
Ueno, T., and F.J. Gonzalez. 1990. Transcriptional control of the rat hepatic CYP2E1 gene.
Molecular and Cellular Biology. 10(9): 4495-4505.
Umeno, M., B.J. Song, C. Korzak, et al. 1988. The rat P-450IIE1 gene: Complete intron and exon
sequence, chromosome mapping, and correlation of developmental expression with specific 5'
cytosine demethylation. J. Biol. Chem. 263(10): 4956-4962.
U.S. EPA. 1986a. Guidelines for carcinogen risk assessment. Federal Register 51(185):33992-
34003.
U.S. EPA. 1986b. Guidelines for the health risk assessment of chemical mixtures. Federal Register
51(185):34014-34025.
U.S. EPA. 1986c. Guidelines for mutagenicity risk assessment. Federal Register 51(185):34006-
34012.
U.S. EPA. 1986d. Summary review of the health effects associated with phenol: Health issue
Assessment. Office of Health and Environmental Assessment, Environmental Criteria and
Assessment Office, Research Triangle Park, NC. EPA/600/8-86/003F
U.S. EPA. 1987. Risk Assessment Guidelines of 1986 (EPA/600/8-87/045, dated August, 1987)
U.S. EPA. 1988. Recommendations for and Documentation of Biological Values for use in Risk
Assessment. EPA 600/6-87/008, NTIS PB88-179874/AS, February, 1988.
121
-------
U.S. EPA 1991. Guidelines for developmental toxicity risk assessment. Federal Register
56(234):63798-63826.
U.S. EPA. 1994a. Interim policy for particle size and limit concentration issues in inhalation
toxicity: notice of availability. Federal Register 59(206):53799.
U.S. EPA. 1994b. Methods for derivation of inhalation reference concentrations and application of
inhalation dosimetry. EPA/600/8-90/066F.
U.S. EPA. 1994c. Peer review and peer involvement at the U.S. Environmental Protection Agency.
Signed by the U.S. EPA Administrator Carol M. Browner, dated June 7, 1994.
U.S. EPA. 1995. Use of the benchmark dose approach in health risk assessment. EPA/630/R-
94/007.
U.S. EPA. 1996a. Proposed guidelines for carcinogen risk assessment. Federal Register
61(79):17960-18011.
U.S. EPA. 1996b. Guidelines for reproductive toxicity risk assessment. Federal Register
61(212):56274-56322.
U.S. EPA. 1998a. Guidelines for neurotoxicity risk assessment. Federal Register 63(93):26926-
26954.
U.S. EPA. 1998b. Science policy council handbook: peer review. Prepared by the Office of Science
Policy, Office of Research and Development, Washington, DC. EPA 100-B-98-001.
U.S. EPA. 1998c. U.S. Environmental Protection Agency, Office of Prevention, Pesticides and
Toxic Substances. Health effects test guidelines. OPPTS 870.7800 Immunotoxicity. Washington,
DC: U.S. Environmental Protection Agency, Office of Prevention Pesticides and Toxic Substances.
U.S. EPA. 1999. Guidelines for carcinogen risk assessment. Review Draft, NCEA-F-0644, July
1999. Risk Assessment Forum.
U.S. EPA. 2000a. Science policy council handbook: peer review. Second edition. Prepared by the
Office of Science Policy, Office of Research and Development, Washington, DC. EPA 100-B-OO-
001.
U.S. EPA. 2000b. Science policy council handbook: risk characterization. Prepared by the Office of
Science Policy, Office of Research and Development, Washington, DC. EPA 100-B-00-002.
U.S. EPA. 2000c. Exposure and human health reassessment of 2,3,7,8-Tetrachlorodibenzo-p-dioxin
(TCD) and related compounds. Downloaded May 16, 2002. Available at
http://www.epa. gov/NCE A/pdfs/dioxin/index.htm.
U.S. EPA. 2000d. Benchmark dose technical guidance document. October 2000, Risk Assessment
Forum, Washington, D.C. External review draft.
122
-------
Van Duuren,BL and BM. Goldschmidt. 1976. Cocarcinogenic and tumor-promoting agents in
tobacco carcinogenesis. J. Natl. Cancer Inst. 56(6): 1237-1242.
VanDuuren, B.L., T. Blaze, B.M. Goldshmidt, C. Katz, S. Melchione, and A. Sivak. 1971.
Cocarcinogenesis studies on mouse skin and inhibition of tumor induction. J. Natl. Cancer Inst. 46:
1039-1044.
Van Duuren, B.L., C. Katz, and BM. Goldschimdt. 1973. Cocarcinogenic agents in tobacco
carcinogenesis. J. Natl. Cancer Inst. 51:703-705.
Vieira, I., M. Sonnier, and T. Cresteil. 1996. Developmental expression of CYP2E1 in the human
liver: Hypermethylation control of gene expression during the neonatal period. Eur J Biochem. 238:
476-483.
Von Oettingen, WF. and NE. Sharpless. 1946. The toxicity and toxic manifestation of 2,2-bis(p-
chlorophneyl)-l,l,l-trichloroethane (DDT) as influenced by chemical changes in the molecules. J
Pharmacol Exp Ther 88:400-413. As cited in ATSDR, 1998.
Weitering, J.G., K.R. Kriggsheld, and G.J. Mulder. 1979. The availability of inorganic sulfate as a
rate limiting factor in the sulfate conjugation xenobiotics in the rat. Biochem Pharmacol. 28: 757-
762.
Woodruff, R.C., J.M. Mason, R. Valencia, et al. 1985. Chemical mutagenesis testing in drosophila.
5.Results of 53 coded compounds tested for the national toxicology program. Environ Mutagen.
7:677-702.
WHO. 1994. World Health Organization. IPCS Environmental Health Criteria 161. Phenol. 161:
92-4.
Wilcosky, T.C., H. Checkoway, E.G. Marshall, et al. 1984. Cancer mortality and solvent exposures
in the rubber industry. Am. Ind. Hyg. Assoc. J. 45(12): 809-811.
Wynder, E.L. and D. Hoffmann. 1961. A study of tobacco carcinogenesis. VIII. The role of the
acidic fractions as promoters. Cancer. 14(6): 1306-1315.
Yager, J.W., D.A. Eastmond, M.L. Robertson, et al. 1990. Characterization of micronuclei induced
in human lymphocytes by benzene metabolites. Cancer Research. 50(2): 393-9.
Yang, J. and P. Duerksen-Hughes. 1998. A new approach to identifying genotoxic carcinogens: p53
induction as an indicator of genotoxic damage. Carcinogenesis. 19(6): 1117-25.
Zamponi, G.W., D. Ing, and R.J. French. 1994. Arrhythmias during phenol therapies: A specific
action on cardiac sodium channels? Circulation. 89: 914.
123
-------
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW
COMMENTS AND DISPOSITION
The support document and IRIS summary for phenol have undergone both internal peer
review by scientists within EPA and a more formal external peer review by scientists outside EPA in
accordance with EPA guidance on peer review (U.S. EPA, 1998b, 2000a). Comments made by the
internal reviewers were addressed prior to submitting the documents for external peer review and are
not part of this appendix. The three external peer reviewers were tasked with providing written
answers to general questions on the overall assessment and on chemical-specific questions in areas of
scientific controversy or uncertainty. A summary of significant comments made by the external
reviewers and EPA's response to these comments follows.
(1) General Comments
A. Data Presentation
All three reviewers agreed that the document is well organized. Two reviewers recommended
specific changes to increase the clarity of certain sections.
Response to comments: The specific changes requested by the reviewers to clarify the text
were made.
B. Are there additional data/studies that should be included?
One reviewer was not aware of any other data/studies that should be included. Another
reviewer recommended a developmental toxicity study but noted that it uses an unconventional study
design. A third reviewer criticized EPA for superficially exploring human health literature and
relying heavily on animal experiments. This third reviewer also suggested that EPA expand its
discussion of studies looking at phenol usage in over-the-counter drugs.
Response to comments: We reviewed the developmental toxicity study (Minor and Becker,
1971) recommended by one reviewer and concluded that it should not be included in the document
because it would not contribute any significant information to the database. The specific reasons for
the exclusion are: (1) this study used intraperitoneal (i.p.) dosing, a route of exposure of limited
relevance for human environmental exposure; because several well-conducted oral developmental
toxicity studies of phenol are available, it is not useful or necessary to supplement the database with
an i.p. study; (2) phenol was administered on gestation days 9-11 or days 12-14; such studies of
short windows of exposure can be useful for mechanistic purposes, but they are inadequate to fully
assess developmental toxicity; (3) this study is presented only as an abstract published 30 years ago,
and the full study was never published, raising questions about the reliability of the results, and in
addition, the abstract provides insufficient information for evaluation of the study.
In response to the request for additional human data, additional reviews of literature databases
and secondary review articles were conducted, but no additional data were identified. Trade
associations were also contacted, with the same results. However, information relating the effects
observed in animals to effects observed in the available human studies was added to Section 4.5. Due
to limitations such as confounding exposures and uncertain exposure estimation, the available human
124
-------
data are useful for hazard identification but not for dose-response assessment.
Manufacturers of over-the-counter drugs containing phenol, the relevant trade associations
were contacted, and reference books were consulted, in order to obtain information on the basis for
the recommended maximal dosage, but the information was not available.
C. For the RfD, has the most appropriate critical effect (decreased fetal body weight in the
NTP (1983a) study and decreased maternal weight gain in the Argus Research Laboratories
(1997) study) been chosen?
One reviewer agreed with the selection of decreased fetal body weight as the critical effect,
with a NOAEL of 60 mg/kg-day. This reviewer noted that although this is a relatively nonspecific
toxic effect, it could be appropriately considered adverse. In contrast, a second reviewer stated that
the reduction in body weight was a weak basis for the RfD because it was nonspecific and because of
other changes affecting body weight (e.g., increased litter size). On the basis of these considerations,
the reviewer did not consider the observed reduction in body weights to be an adverse effect and
suggested that the corresponding dose might be considered a NOAEL. Neither of these reviewers
commented on the appropriateness of the co-critical effect of decreased maternal weight gain
endpoint, also with a NOAEL of 60 mg/kg-day (and a BMDL of 93 mg/kg-day).
A third reviewer asserted that the critical effect occurs at a lower dose and that the RfD should
be based on a NOAEL lower than 60 mg/kg-day. The reviewer stated that the data from the NTP
(1983a) study are consistent with a LOAEL of 30 mg/kg-day, based on statistically significant
increases in the number of litters with nonlive fetuses in the low- and mid-dose groups and a trend
comparison of the fetal weight data. The reviewer also stated that data from the Argus Research
Laboratories (1997) study support the conclusion that 60 mg/kg-day is a LOAEL, based on an
increase in the percent litters with fetuses with alterations. The reviewer also recommended that the
LOAEL in the IIT Research Institute (1999) study should be 20 mg/kg-day (with no NOAEL
identified), based on decreased uterine weight and decreased prostate weight, rather than the NOAEL
of 70.0 mg/kg-day identified in the Toxicological Review. The reviewer asked for additional
discussion of the interpretation of these results but did not recommend a specific critical effect. This
reviewer also recognized the uncertainties in the Hsieh et al. (1992) study but suggested that
macrocytosis could explain the inconsistency in that study between erythrocyte count and hematocrit.
Response to comments: The relationship between litter size and fetal weight was noted, and
the analysis of the NTP (1983a) study notes that 120 mg/kg-day is an equivocal LOAEL, with a
NOAEL of 60 mg/kg-day. Due to the uncertainties in the identification of the NOAEL/LOAEL
boundary for this study, the NTP (1983a) study was then judged not appropriate as a co-principal
study. If the NOAEL in the NTP (1983a) study were changed to 120 mg/kg-day, there would not be
a significant effect on the RfD because the Argus Research Laboratories (1997) study identified a
NOAEL of 60 mg/kg-day and a LOAEL of 120 mg/kg-day, with a BMDL of 93 mg/kg-day. In
addition, the NOAEL identified in the principal study (Argus Research Laboratories, 1997) is
supported by a NOAEL of 107 mg/kg-day in the ClinTrials BioResearch (1998) study, a value that is
very close to the BMDL used to derive the RfD.
125
-------
Additional information about the litters with nonlive fetuses, decreased uterine weight, and
other findings mentioned by the third reviewer were added to the document (Section 4.3). The
former endpoint was not considered treatment-related due to the absence of a dose-response. The
data on percent litters with fetuses with alterations were independently analyzed using the chi-square
test, and no significant effect was observed. Text was also added to Section 6.1.1 regarding
uncertainties in the interpretation of the uterine weight data.
D. Has the noncancer assessment been based on the most appropriate studies?
Two of the reviewers agreed that the noncancer assessment is based on the most appropriate
studies, although, as indicated in the previous question, one of the reviewers raised some concerns
about the interpretation of those studies. Two of the reviewers also agreed with EPA's assessment of
the deficiencies of the study by Hsieh et al. (1992). One reviewer agreed with the selection of
significant endpoints in the phenol document; however, this reviewer disagreed with the use of
gavage studies for the derivation of the RfD. This reviewer asserted that EPA should have used the
two-generation drinking water study by IIT Research Institute (1999) to derive the RfD.
Response to comments: As described in the Toxicological Review, the effects at the LOAEL
(the high dose) of the IIT Research Institute (1999) study appear to be secondary to decreased water
consumption due to poor palatability, and so do not appropriately reflect phenol toxicity. Text was
added to the Toxicological Review noting that rats consume water only intermittently during the day,
and so a divided-dose gavage protocol is fairly similar to drinking water consumption.
E. For the noncancer (RfD) assessment, are there other data that should be considered in
developing uncertainty factors or the modifying factor? Do the data support the use of
different values from those proposed?
All three reviewers agreed with the uncertainty factors presented in the phenol document and
EPA's rationale for selecting these factors.
Response to comments: None.
F. Do the confidence and weight-of-evidence statements present a clear rationale and
accurately reflect the utility of the studies chosen, the relevancy of the effects (cancer and
noncancer) to humans, and the comprehensiveness of the database? Do these statements make
sufficiently apparent all the underlying assumptions and limitations of these assessments?
The reviewers agreed that the confidence statements are carefully reasoned and clearly stated.
Two of the reviewers agreed that the weight-of-evidence statements are appropriate, and a third
reviewer noted that no weight-of-evidence statement is used for noncarcinogenicity.
Response to comments: None needed.
(2) Chemical-Specific Comments
A. When endogenously produced phenol is taken into account, can the RfD be supported?
Note that the RfD is applied to ingested phenol in addition to the normal daily endogenously
126
-------
produced phenol. Are there differences in endogenous phenol production between rats and
humans that should be taken into account in the development of the RfD?
The reviewers agreed that the RfD applied to ingested phenol in addition to the normal daily
endogenously produced phenol is appropriate. One reviewer noted that endogenous production is a
relatively small fraction (-2-20%) of the RfD. (This fraction became 5-50% after addition of the
database uncertainty factor during consensus review.) Another reviewer noted that both the
experimental animals and humans would have similar baseline levels of endogenous phenol
production.
Response to comments: Additional supporting information provided by the reviewers was
incorporated into the text. Toxicokinetic considerations regarding quantitatively accounting for
endogenous production were also incorporated into the text.
B. Do you agree/disagree with the recommendation that there are not sufficient data to
generate a scientifically defensible RfC and cancer slope factor?
Two reviewers agreed that data are insufficient to generate an RfC and cancer slope factor.
One of these reviewers also commented on statements made by EPA related to red nasal discharge
identified in study animals. This reviewer stated that the secretion/discharge is not directly
suggestive of, or a precursor to, a nasal or ocular lesion but is simply the animal's response to stress.
A third reviewer did not comment on this question because it was beyond this person's area of
expertise.
Response to comments: The information on red nasal discharge provided by the reviewer
was incorporated into the text.
C. Was the interpretation of the decreased fetal body weight in rats in the National Toxicology
Program (NTP) study (NTP, 1983a) appropriate?
One reviewer raised some issues regarding the adversity of the effect but generally agreed
with EPA's decision to designate the decreased fetal body weight finding at 120 mg/kg-day as an
equivocal LOAEL, resulting in a NOAEL of 60 mg/kg-day. A second reviewer disagreed that
decreased fetal body weight is an adverse effect and asserted that 120 mg/kg-day should be classified
as a NOAEL instead. A third reviewer supported interpreting the data using trend analysis rather
than pairwise comparisons, resulting in a LOAEL of 30 mg/kg-day for fetal body weight reduction.
Response to comments: On the basis of the weight of evidence, the weight of the reviewers'
comments, and the supporting data from the Argus Research Laboratories (1997) study, the high dose
in the NTP (1983a) study was retained as an equivocal LOAEL, but the study was removed from
being a co-principal study for the derivation of the RfD. Although there was a dose-response trend at
the low and mid doses, the decreases in the fetal body weight at these doses were marginal (l%-2%)
and were not considered biologically significant. Only the response (7%) at the high dose was
significantly different from the control. A meaningful benchmark dose could not be calculated for
these data, because fetal body weight by sex were not in the published study and not available in
127
-------
NTP's archives.
D. Please comment on the choice of gavage developmental toxicity studies as the co-critical
studies in light of the differences between phenol toxicity when administered in drinking water
and by gavage.
One reviewer agreed with EPA's willingness to use gavage studies but suggested that the IIT
Research Institute (1999) drinking water study represents a more relevant exposure scenario and
supports a lower LOAEL, based on decreased absolute and relative uterine weight. A second
reviewer asserted that the drinking water study is more appropriate because a divided gavage dose is
not equivalent to more extended intake during ad libitum water consumption. A third reviewer
supported the use of the gavage study and asserted that the divided dose administered in the Argus
Research Laboratories study (1997) is not unrealistic because animals in drinking water studies tend
to drink when they eat, not continuously throughout the day. This reviewer believed that the NOAEL
and the LOAEL are not overly conservative.
Response to comments: Additional information about the interpretation of the decreased
uterine weight in the IIT Research Institute (1999) drinking water study and associated uncertainties
was added Section 4.3. In particular, even though the decrease in relative uterine weight was
statistically significant at all dose levels, there was no dose-response. Information was also added to
the text noting that rats drink water in a small number of periods during the day rather than
continuously through the day, so a divided-dose gavage study is fairly similar to drinking water
exposure.
E. Was the interpretation of decreased motor activity in the 13-week oral neurotoxicity study
appropriate?
One reviewer agreed with EPA's interpretation of decreased motor activity in the 13-week
oral neurotoxicity study. A second reviewer found the interpretation difficult to comment on, given
the confounded results of the study, and a third reviewer did not respond to this question because it
was beyond this person's area of expertise.
Response to comments: The issues potentially confounding the neurotoxicity were noted in
the document.
OVERALL RECOMMENDATION
All three reviewers stated that the document is acceptable with revisions.
New Reference:
Minor, JL; Becker, BA. (1971). A comparison of the teratogenic properties of sodium salicylate,
sodium benzoate, and phenol. Toxicol Appl Pharmacol 19:373.
128
-------
Appendix B. Benchmark Dose Modeling Results
Benchmark dose (BMD) modeling was performed to identify potential critical effect levels for
derivation of the RfD for phenol. The modeling was conducted according to draft EPA guidelines
(U.S. EPA, 2000d) using Benchmark Dose Software Version 1.3 (BMDS), which is available from
EPA (U.S. EPA, 2001). The BMD modeling results are summarized in Table B-l, and the output is
attached as Appendix C. A brief discussion of the modeling results for each endpoint is presented
below.
Because all the following endpoints are continuous variables, the continuous models available
with BMDS (power, polynomial, and Hill models) were used. The hybrid model was not used,
because the hybrid model software in BMDS is still undergoing Beta-testing and was not considered
sufficiently validated to provide a BMDL as the basis for the quantitative dose-response assessment.
(The hybrid modeling approach defines the benchmark response [BMR] directly in terms of risk,
whereas the standard approach, defines the BMR in terms of a change in the mean.) For all of the
modeling conducted, the BMR was defined as a 1.0 SD change in the mean because this is the default
measure recommended by the EPA (U.S. EPA, 2000d) in the absence of a clear biological rationale
for selecting an alternative response level.
Argus Research Laboratories, 1997
Two endpoints were modeled from this study: decreased maternal body weight gain and the
related endpoint of decreased maternal body weight. The decrease in weight gain was the more
sensitive endpoint, with a NOAEL of 60 mg/kg/day and a LOAEL of 120 mg/kg/day.
As summarized in Table B-l, the BMD and BMDL estimates for the endpoint of decreased
maternal body weight gain was similar for all three models. The model fit was also generally similar.
A visual analysis of the data fit in the regions of the BMDLs indicated that the data fitting from the
three model was adequate and comparable across all models. The goodness-of-fit/>-values calculated
for the power and polynomial models were very similar, but no/>-value could be computed for the
Hill model because there were no degrees of freedom for the calculation. (This was because the
number of parameters included in the model was equal to the number of data points.) The Akaike
Information Criterion (AIC), a measure of goodness of fit that takes into account the number of
degrees of freedom, was very similar for all three models but was marginally better (i.e., lower) for
the polynomial model. The polynomial model was chosen as the basis for the BMDL for this
endpoint, based on the slightly better fit and as a slightly more health-protective value. An
alternative BMDL for this endpoint could be calculated using the geometric mean of the BMDLs
from all three models, 114 mg/kg-day, based on the rationale that the small difference in AIC
observed was not meaningful.
For the endpoint of decreased maternal body weight, all three continuous models gave a
similar BMD estimate and provided adequate data fits, with goodness-of-fit/>-values larger than 0.1.
However, the best data fit was obtained with the polynomial model, which had a/>-value of 0.92,
compared with a/>-value of 0.30 obtained with the power model. The/>-value for the Hill model was
not computed due to insufficient degrees of freedom. A visual analysis of the data fit in the region of
129
-------
the BMDLs also indicated that good fit was obtained with all three models, with the best fit obtained
using the polynomial model. In addition, the AIC analysis also indicated that the best data fit was
obtained with the polynomial model. Comparable BMDLs of 143 and 147 mg/kg/day were obtained
using the polynomial and Hill models, respectively. Thus, the BMDL chosen for this endpoint was
143 mg/kg-day, obtained with the polynomial model, with a corresponding BMD of 345 mg/kg. This
BMDL is higher than the BMDL for maternal body weight gain, the other endpoint modeled for this
study.
ClinTrials BioResearch (1998)
Only one endpoint from this study was modeled: decreased total activity counts in a motor
activity assay in females at week 4. Acceptable fits (/?=0.35 and/>=0.17) were obtained with the
polynomial and power models, respectively. The visual fit for both models in the region of the
BMDL was acceptable but not as good as would be desired. Much better fit was obtained with the
Hill model, based on visual fit, but no BMDL could be calculated for this model. The same BMDL
of 219 mg/kg-day was calculated with both the power and polynomical models and was chosen as the
study BMDL.
Hsieh et al. (1992)
Three related endpoints were modeled for this study: plaque-forming cells, plaque-forming
cells/106 spleen cells, and antibody titer. The study NOAEL was 6.2 mg/kg-day, with a LOAEL of
33.6 mg/kg-day.
As summarized in Table B-l, unacceptable fits were obtained with the power and polynomial
models for the endpoint of plaque-forming cells/106 spleen cells. No/>-value could be obtained for
the Hill model due to insufficient degrees of freedom, but a visual analysis of the results indicated
that the Hill model provided an acceptable fit. Based on the Hill model, the BMD and BMDL for
decreased plaque-forming cells/106 spleen cells were 1.26 mg/kg-day and 0.38 mg/kg-day,
respectively.
Similar results were obtained for plaque forming cells/spleen. Inadequate fits were obtained
with the power and polynomial models, and the Hill model provided no/>-value at all due to
insufficient degrees of freedom. A visual analysis of results indicated that the Hill model provided an
overall adequate fit, but not a very good fit in the range of the BMD. In addition, this model failed to
estimate BMD or BMDL. Therefore, no BMD and BMDL can be identified for this particular
endpoint.
For the endpoint of decreased antibody titer, the power and polynomial models both had
marginal fit, based on the goodness-of-fit/>-values; visual inspection of the data indicated that these
models had inadequate fit. No p-value could be calculated for the Hill model, due to insufficient
degrees of freedom, but a visual analysis of the model results indicated an adequate fit. Therefore,
based on the Hill model, the BMD and BMDL for decreased antibody titer were 3.51 mg/kg-day and
0.73 mg/kg-day, respectively.
The lowest BMDL for this study was 0.38 mg/kg-day, calculated for plaque-forming cells/106
130
-------
spleen cells. However, this BMDL is not used for risk assessment due to uncertainties in the
appropriate BMR for this in vivo/in vitro study design.
Table B-l. Benchmark Dose Modeling Results for Phenol
Model |p-value
AIC
BMDa
BMDLa
Argus Research Laboratories (1997)
Maternal Body Weight Gain
Power
Polynomial
Hill
0.3165
0.3191
N/A
545
543
545
152
157
151
125
93
129
Maternal Body Weight
Power
Polynomial
Hill
0.3013
0.9188
N/Ab
731
729
733
354
345
345
244
143
147
ClinTrials BioResearch Ltd. (1998)
Motor Activity - Total Activity Counts in Females in Week 4
Power
Polynomial
Hill
0.1701
0.3477
N/A
629
625
630
337
336
246
219
219
—
Hsieh et al., 1992
Plaque-Forming Cells
Power
Polynomial
Hill
Plaque-Forming Cells/Total Spleen
Power
Polynomial
Hill
0.008
0.003
NA
0.054
O.0001
N/A
236
227
229
469
509
467
15.7
14.3
1.26
23
23.4
6.14
10.5
9.7
0.38
15.3
5.87
C
131
-------
Antibody Titer
Power
0.102
-66.2
21.4
13.1
Polynomial
0.102
-70.2
21.4
13.1
Hill
N/A
-66.9
3.51
0.73
a. BMD and BMDL are based on benchmark response of 1.0 SD. Results are presented in units of mg/kg/day.
b. NA: the information is not available because there are insufficient degrees of freedom for the test
c. —: failed to estimated this value.
132
-------
Appendix C. Benchmark Dose Modeling Output
133
-------
Ma.out
maternal body weight gain
Power Model. $Revision: 2.1 $ $Date: 2000/10/11 20-57-36 $
Input Data File: F:\BMDS\MA.(d)
Gnuplot Plotting File: F:\BMDS\MA.plt
Fri May 10 11:56:51 2002
:============================================-=
BMDS MODEL RUN
-~" 1 ,-*_-«__. ,_ — -—**«*,^*^^.
The form of the response function is:
Yfdose] = Control ••• slope * dose^power
Dependent variable = MEAN
Independent variable = dose
rho is set to 0
The power is restricted to be greater than or equal to 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations =,250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 101.864
rho = 0 Specified
control = 64
slope = -43.1882
power = -0.0984037
Asymptotic Correlation Matrix of Parameter Estimates
alpha
rho
control
slope
power
alpha
1
-1
0.01
-0.049
-0.053
rho
-1
1
-0.011
0.049
0.054
control
0.01
-0.011
1
-0.77
-0.74
slope
-0.049
0.049
-0.77
1
1
power
-0.053
0.054
-0.74
1
1
Variable
alpha
Parameter Estimates
Estimate
98.6453
Std. Err
328.851
Page 1
-------
Ma.out
rho
control
slope
power
0
63.4328
-0.0653203
1
0.834088
2.12406
0.11847
0.30005
Table of Data and Estimated Values of Interest
Do»« H Obs Mean Obs Std Dev Est Mean
0
60
120
360
23
25
23 .
25
Est Std Dev Chi^2 Res.
64
58
56.8
39.8
10.7
9.4
10.8
9.5
63.4
59.5
55.6
39.9
9.93
9.93
9.93
9.93
0.0571
-0,152
0.121
-0.0118
Model Descriptions for likelihoods calculated
Model Al: Yij
Var{e(ij)}
Model A2 Yij
Var{e(ij)}
Model R:
•Var{e(i)}
Mu(i) +
Sigma"2
Mu(i) +
Mu
SigmaA2
Test 1:
Test 2:
Test 3:
Likelihoods of Interest
Model Log(likelihood) DF AIC
Al -267.891852 5 545.783705
A2 -267.502264 8 551.004528
fitted -268.393473 4 544.786945
R -297.755244 2 599.510487
Does response and/or variances differ among dose levels
(A2 vs. R)
Are Variances Homogeneous (Al vs A2)
Does the Model for the Mean Fit (Al vs. fitted)
Tests of Interest
Test -2*log(Likelihood Ratio)
Test 1
Test 2
Test 3
60.506
0.779177
1.00324
df
6
3
1
p-value
<.00001
0.8544
0.3165
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .05
model appears to be appropriate here
A homogeneous variance
Page 2
-------
Ma.out
The p-value for Test 3 is greater than .05. The model chosen appears
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 152.051
BMDL = 124.591
Page 3
-------
Power Model with 0.95 Confidence Level
CO
ffl
CC.
(0
I
45
40
35
0
300
350
11:5605/102002
dose
-------
maternal body weight gain
Polynomial Model. $Revision: 2.1 $ $Date: 2000/10/11 17:51:39 $
Input Data File: F:\BMDS\MA.(d)
Gnuplot Plotting File: F:\BMDS\MA.plt
Fri May 10 11:55:32 2002
BMDS MODEL RUN
The form of the response function is:
Ytdose] « b*ta_0 + beta_l*dose + beta_2*doseA2
Dependent variable = MEAN
Independent variable = dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 101.864
, rho = 0 Specified
beta_0 = 63.3326
beta_l = -0.0615798
beta 2 = -1.00086e-005
Parameter Estimates
Variable Estimate Std. Err
alpha 98.6343 14.2366
beta_0 63.2966 1.94612
beta_l -0.0618515 0.0343059
beta_2 -8.99169e-006 8.68663e-005
Asymptotic Correlation Matrix of Parameter Estimates
alpha beta_0 beta_l beta_2
alpha 1 -1.2e-006 1.9e-006 -2e-006
beta_0 -1.2e-006 .1 -0.77 0.68
beta_l 1.9e-006 -0.77 1 -0.98
beta_2 -2e-006 0.68 -0.98 1
Table of Data and Estimated Values of Interest
Page 1
-------
Ma.out
Dose
Res.
0
60
120
360
N
23
25
23
25
Obs Mean
64
58
56.8
39.8
Obs Std De-v
10.7
9.4
10.8
9.5
r Est Mean
63.3
59.6
55.7
39.9
Est Std Dev
9.93
9.93
9.93
9.93
Cni~2
1.63
-3.91
2.44
-0.163
Model Descriptions for likelihoods calculated
Model Al Yij = Mu(i) +
Var{e(ij)) = SigmaA2
Model A2 Yij = Mu(i) +
Var{«(ij)} = Sigma(i)"2
Model R Yi = Mu + e(i)
Var{e(i)} -= SigmaA2
Likelihoods of Interest
Test
levels
Test
Test
1
2
3
Test
Test
Test
Test
Model
Al
-A2
fitted
R
Log (likelihood)
-267.891852
-267.502264
-268.388117
-297.755244
DF
5
8
3
2
AIC
545.783705
551.004528
542.776234
599.510487
Does response and/or variances differ among dose
(A2 vs. R)
Are Variances Homogeneous (Al vs A2)
Does the Model for the Mean Fit (Al vs. fitted)
Tests of Interest
-2*log(Likelihood Ratio) Test
1
2
3
60.506
0.779177
0.992529
6
3
1
df
P-
value
<.0001
0.8544
0.3191
The p-value for Test 1 is less than .05. There appears
to be a .
difference between response and/or variances among the
dose levels.
It seems appropriate to model the data
The p-value for Test 2 is greater than 05. A
homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than 05 The model
Page 2
-------
chosen appears
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type * Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 156.987
*,
BMDL = 92.9272
Page 3
-------
Polynomial Model with 0.95 Confidence Level
70
65
60
-------
Ma.out
maternal body weight gain
. BH i '•-; «•". ! '4 ;
It:; F • '. t:!-'D,;,:',Mn "~"! r
Fri May 10 11:57:36 2002
BMDS MODEL RUN
The form of the response function is
Y[dose] = intercept + v*dose*n/(k^n + dose^n)
Dependent variable = MEAN
independent variable = dose
rho is set to 0
Power parameter restricted to be greater than
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
.Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 101.585
, rho = 0 Specified
intercept = 64
v = -24.2
n = 1.42687
k = 189.176
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -n
have been estimated at a boundary point, or have been specified
by the user,
and do not appear in the correlation matrix )
alpha
rho
intercept
V
k
alpha
1
0
0
0
0
rho
0
1
0
0
0
intercept
0
0
1
0
0
V
0
0
0
1
0
k
0
0
0
0
1
Page 1
-------
Ha.out
Variable
alpha
rho
intercept
v
n
k
Parameter Estimates
Estimate Std Err
98.6445 1
0 1
63.4595 1
-2388.7 1
1 NA
36185.3
NA - Indicates that this parameter has hit a bound
implied by some'inequality constraint and thus
has no standard error.
Table of Data and Estimated Values of Interest
Dose N Obs Mean Obs J>td Dav Est Mean
0
60
120
360
23
25
23
25
Est Std Dev ChiA2 Res.
64
58
56.8
39.8
10.7
9.4
10.8
9.5
63.5
59.5
55.6
39.9
9.93
9.93
9.93
9.93
0.0544
-0.152
0.124
-0.013
Model Descriptions for likelihoods calculated
Model Al: Yi.j
Var{e(ij)}
Model A2: Yij
Var{e(ij)}
Model R: Yi
Var{e(i)>
Mu(i) +
Sigma "^
Mu(i) +
Sigma (i)"2
Mu -i- e(i)
Sigma"2
Degrees of freedom for Test Al vs fitted <= 0
Likelihoods of Interest
Model Log(1ikelihood) DP
Al -267.891852 5
A2 -267.502264 8
fitted -268.395812 4
R -297.755244 2
AIC
545.783705
551.004528
544.791623
599.510487
Test 1: Does response and/or variances differ among dose levels
(A2 vs. R)
Teat 2: Are Variances Homogeneous (Al vs A2)
Teat 3: Does the Model for the Mean Fit (Al vs. fitted)
Tests of Interest
Page 2
-------
Ma.out
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 60.506 6 -c.OOOl
Test 2 0.779177 3 0.8544
Test 3 1.00792 0 NA
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .05. A homogeneous variance
model appears to be appropriate here
NA - Degrees' of freedom for Test 3 are less than or equal to 0 The Chi-Square
test for fit is not valid
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 151.083
BMDL = >. 129.177
Warning optimum may not have been found Bad completion code in Optimization r
outine.
*
BMDL computation failed for one or more point on the BMDL curve
The BMDL curve will not be plotted
Page 3
-------
Hill Model with 0.95 Confidence Level
W
0>
DC
50
45
40
35
350
11:5705/102002
-------
BMR=1.0 SD
DAMWEIGHT.OUT
Power Model. $Revision: 2.1 $ $Date: 2000/10/11 20:57:36 $
Input Data File: C:\DOCUMENTS AND SETTINGS\JZHAO\MY
DOCUMENTS\PHENOL\DAMWEIGHT.(d)
Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\3ZHAO\MY
DOCUMENTS\PHENOL\DAMWEIGHT.pit
Thu Jan 10 16:36:20 2002
BMDS MODEL RUN
The form of'-the response function is:
Y[dose] = control + slope * doseApower
Dependent variable = MEAN
independent variable = dose
rho is set to 0
The power is restricted to be greater than or equal to 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default initial Parameter values
alpha
rho =
control =
slope =
power =
779.391
0
440.4
-357.319
-0.432011
Speci fi ed
Asymptotic Correlation Matrix of Parameter Estimates
alpha
rho
control
slope
power
alpha
1
-1
-0.095
0.16
0.16
rho
-1
1
0.094
-0.16
-0.16
control
-0.095
0.094
1
-0.78
-0.76
slope
0.16
-0.16
-0.78
1
1
power
0.16
-0.16
-0.76
1
1
Parameter Estimates
variable
alpha
rho
control
Esti mate
747.101
0
439.654
Std. Err
25868.4
5.71247
6.06278
Page 1
-------
slope
power
DAMWEIGHT.OUT
-0.0770251 0.338099
1 0.724264
Table of Data and Estimated values of interest
Dose N obs Mean Obs std Dev Est Mean
Est Std Dev ChiA2 Res.
0
60
120
360
23
25
23
24
440
435
429
412
29
28.2
25.1
29.1
440
435
430
412
27.3
27.3
27.3
27.3
0.0273
0.00247
-0.0443
0.0137
Model Descriptions for likelihoods calculated
Model Al:
Model A2:
Model R:
Yii
var{e(ij)}
Yii
var{e(ij)}
Yi
var{e(i)}
Mu(i) +
Si gmaA2
Mu(i)
Sigma(i)A2
Mu H
Si gmaA2
warning: Likelihood for model Al larger than the Likelihood for model A2.
Likelihoods of interest
Model . Log(likelihood) DF AIC
Al -361.236054 5 732.472108
A2 -361.411149 8 738.822299
fitted -361.769541 4 731.539082
R -368.360689 2 740.721377
Test 1:
Test 2:
Test 3:
Does response and/or variances differ among dose levels
(A2 vs. R)
Are variances Homogeneous (Al vs A2)
Does the Model for the Mean Fit (Al vs. fitted)
Test
Test 1
Test 2
Test 3
Tests of interest
-2*log(Likelihood Ratio)
13.8991
-0.350191
1.06697
df
6
3
1
p-value
0.003046
<.00001
0.3016
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels.
it seems appropriate to model the data
The p-value for Test 2 is less than .05.
non-homogeneous variance model
Consider running a
The p-value for Test 3 is greater than .05. The model chosen appears
to adequately describe the data
Page 2
-------
DAMWEIGHT.OUT
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 354.861
BMDL = 244.116
Page 3
-------
Power Model with 0.95 Confidence Level
0
50
100
150 200
250
300
350
16:3701/102002
dose
-------
Polynomial Model. $Revision: 2.1 $ $Date: 2000/10/11 17:51:39 $
Input Data File: F:\BMDS\Argus maternal bw.(d)
Gnuplot Plotting File: F:\BMDS\UNSAVEDl.plt
Fri Hay 10 10:18:09 2002
BMDS MODEL RUN
The form of the response function is
Yfdose] = <,beta_0 + beta_l*dose + b*ta_2*dose/N2
Dependent variable = MEAN
Independent variable = dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 779.391
rho = 0 Specified
beta_0 = 440.188
beta_l = -0.0971452
beta_2 = 5.21988e-005
Parameter Estimates
Variable Estimate Std. Err.
alpha 746.657 108.336
beta_0 440.199 5.35447
beta_l -0.0970673 0.0943987
beta_2 5.19026e-005 0.000239328
Asymptotic Correlation Matrix of Parameter Estimates
alpha beta_0 beta_l beta^_2
alpha 1 -5.5e-008 7.6e-008 -7.5e-008
beta_0 -5.5e-008 1 -0.77 0.68
beta_l 7.6e-008 -0.77 1 -0.98
beta_2 -7.5e-008 0.68 -0.98 1
Table of Data and Estimated Values of Interest
Page 1
-------
Dose
Res.
0
60
120
360
N
23
25
23
24
Obs Mean
440
435
429
412
Obs Std Dev
29
28.2
25.1
29.1
Est Mean
440
435
429
412
Est Std Dev
27.3
27.3
27.3
27.3
Chi" 2
-0.167
0.401
-0.251
0.0167
Model Descriptions for likelihoods calculated
Model Al: Yij
Var{e(ij)}
Model A2:
Model R:
Mu(i)
Yij
Var{e(ij)}
Yi
Var{e(i)}
Model
Al
- A2
fitted
R
= Mu(i) + e(ij)
= Sigma (i) "2
= Mu + e(i
= Sigma/v2
Likelihoods of
Log (likelihood)
-361.736054
-361.411149
-361.741255
-368.365774
Interest
DP
-5
8
3
2
AIC
733.472108
738.822299
729.482510
740.731548
Test 1 Does respqnse and/or variances differ among dose
levels
(A2 vs. R)
Test 2: Are Variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Test
Test 1
Test 2
Test 3
Tests of Interest
-2*log(Likelihood Ratio) Test df
13.9092
0.649809
0.010402
6
3
1
p-value
0.003031
0.8849
0.9188
The p-value for Test 1 is less than .05. There appears
to be a
difference between response and/or variances among the
dose levels.
It seems appropriate to model the data
The p-value for Test 2 is greater than .05. A
homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than 05 The model
Page 2
-------
Unsavedl.out
chosen appears
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence larval » 0.95
HMD = 345.237
BMDL = 142.794
Page 3
-------
Polynomial Model with 0.95 Confidence Level
450
10:1805/102002
-------
DAMWEIGHT.OUT
BMR=1.0 SO
mil Model. $Revision: 2.1 $ Soate: 2000/10/11 21:21:23 $
input Data File: C:\DOCUMENTS AND SETTINGS\JZHAO\MY
DOCUMENTS\PHENOL\DAMWEIGHT.(d)
Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\JZHAO\MY
DOCUMENTS\PHENOL\DAMWEIGHT.pit
Thu Jan 1C] 16r 38: 54
BMDS MODEL RUN
The form of^the response function is:
Y[dose] = intercept + v*doseAn/(kAn + doseAn)
Dependent variable = MEAN
independent variable = dose
rho is set to 0
Power parameter restricted to be greater than 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default initial Parameter values
alpha = 772.871
rho « 0 Specified
intercept = 440.4
v = -28.1
n = 1.44931
k = 160.473
Asymptotic Correlation Matrix of Parameter Estimates
k
alpha
0
rho
0
intercept
0
V
0
n
0
k
alpha
1
0
0
0
0
0
rho
0
1
0
0
0
0
intercept
0
0
1
0
0
0
V
0
0
0
1
0
0
n
0
0
0
0
1
0
Page 1
-------
DAMWEIGHT.OUT
Variable
alpha
rho
intercept
v
n
k
Parameter Estimates
Estimate
746.575
0
440.4
-57.271
1.25328
370.906
Std
Err.
1
1
1
1
1
1
Table of Data and Estimated values of interest
Dose N obs Mean obs Std Dev Est Mean Est Std Dev chiA2 Res
0
60
120
360
23
25
23
24
440
435
429
412
29
28.2
25.1
29.1
440
435
429
412
27.3
27.3
27.3
27.3
5.28e-008
3 . 13e-007
-3.6e-007
9.886-007
Model Descriptions f^>r likelihoods calculated
Model Al: Yii = Mu(i) + e(ij)
var{e(ij)} = sigmaA2
Model A2: Yii •= Mu(i) + e(ij)
var{e(ij)> = Sigma(i)A2
Model R: Yi = MU + e(i)
var{e(i)} = sigmaA2
warning: Likelihood for fitted model larger than the Likelihood for model Al
Likelihoods of interest
Model Log(likelihood) DF Aic
Al -361.736054 5 733.472108
A2 -361.411149 8 738.822299
fitted -361.736054 5 733.472107
R -368.860689 2 741.721377
Test 1: Does response and/or variances differ among dose levels
(A2 vs. R)
Test 2: Are variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Test
Test 1
Test 2
Test 3
Tests of interest
-2*1og(Likelihood Ratio) Test df
14.8991
0.649809
1.01313e-007
6
3
0
Page 2
p-value
0.001W5
0,88-19
-------
DAMWEIGHT.OUT
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels.
It seems appropriate to model the data
The p-value for Test 2 is greater than .05. A homogeneous variance
model appears to be appropriate here
NA - Degrees of freedom for Test 3 are less than or equal to 0. The Chi-Square
test for fit is not valid
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 344.737
BMDL = 147.428
Page 3
-------
Hill Model with 0.95 Confidence Level
450
440
|
8. 430
OT
I
-------
Unsavedl.out
Clintrials motor activity. BMR=1 SO
Power Model. $Revision: 2.1 $ $Date: 2000/10/11 20:57:36 $
Input Data File: F:\BMDS\UNSAVED1.(d)
Gnuplot Plotting File: F:\BMDS\UNSAVEDl.plt
Fri May 17 13:46:34 2002
BMDS MODEL RUN
The form of the response function is:
Y[dose] =*control + slope * dose^power
Dependent variable = MEAN
Independent variable = COLUMN1
rho is set to 0
The power is restricted to be greater than or equal to 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values • 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 14556.4
rho = 0 Specified
control = 468
slope = -2101.71
power = -0.471502
Asymptotic Correlation Matrix of Parameter Estimates
alpha rho control slope power
alpha 1 -1 NA NA NA
rho -1 1 NA NA NA
control NA NA NA NA NA
Slope NA NA NA NA NA
power NA NA NA NA NA
NA - This parameter's variance has been estimated at zero
Page 1
-------
Unsavedl.out
Variable
alpha
rho
control
slope
power
Parameter Estimates
Estimate Std. Err.
13773.9 81444
0 0.995906
455.234 58.2246
-0.348708 5.10257
1 2.40548
Table of Data and Estimated Values of Interest
Dose N Obs Mean Obs Std Dev Est Mean
Est Std Dev Chi"2 Res.
0
24.6
107
360
15
15
15
14
468
451
394
337
118
149
78
127
455
447
418
330
117
117
117
117
0.109
0.037
-0.204
0.0622
Model Descriptions for likelihoods calculated
Model Al:
Var{e(ij)> =
Model A2: Yij
Var{e(ij)}
Model
Mu(i) +
Sigma"2
Mu(i) +
Sigma(i)"2
Yi
Var{e(i)>
Mu
SigmaA2
Likelihoods of Interest
Model Log(likelihood) DP AIC
Al -309.709699 5 629.419399
A2 -307.217620 8 630.435240
fitted -310.650690 4 629.301380
R -315.381614 2 634.763229
Test 1: Does response and/or variances differ among dose levels
(A2 vs. R)
Test 2: Are Variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Tests of Interest
Test -2*log(Likelihood Ratio) df p-value
Test 1 16.328 6 0.0009712
Test 2 4.98416 3 0.173
Test 3 1.88198 1 0.1701
Page 2
-------
Unsavedl.out
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than 05. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .05. The model chosen appears
to adequately describe the data
Benchmark Dose Computation
Specified effect =
Risk Type *
Confidence level =
Estimated standard deviations from the control mean
0.95
336.564
219.138
Page 3
-------
Power Model with 0.95 Confidence Level
-------
Clintrials motor activity in females at 4 weeks, BMR = 1 SD
Polynomial Model. $Revision: 2.1 $ $Date: 2000/10/11 17:51:39 $
Input Data File: F:\BMDS\UNSAVEDl.(d)
Gnuplot Plotting File: F:\BMDS\UNSAVEDl.plt
Fri May 17 13:53:20 2002
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta_0 + beta_l*dose + beta_2*doseA2
Dependent variable = MEAN
Independent variable = COLUMN 1
rho is set to 0
The polynomial coefficients are restricted to be negative
A constant^variance^hodel is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of* iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 1
rho = 0 Specified
beta_0= 469.177
beta_l = -0.83776
beta 2 = 0
Parameter Estimates
Variable Estimate Std. Err.
alpha 13763.4 2532.12
beta 0 455.234 20.0191
-------
beta_l -0.348708 0.108862
beta_2 0 NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Asymptotic Correlation Matrix of Parameter Estimates
alpha beta_0 beta_l
alpha 1 7.1e-010 -2.3e-01
beta_0 7.1e-010 1 -0.65
beta_l -2.3e-011 -0.65 1
The following parameter(s) have been estimated at a
boundary
point or have been specified. Correlations are not
computed:
beta 2
Table of Data and Estimated Values of Interest
Dose N ObsMean ObsStdDev EstMean EstStdDev ChiA2
Res.
0 15
24.6 15
107 15
360 14
468
451
394
337
118
149
78
127
455
447
418
330
117
117
117
117
1.63
0.555
-3.06
0.871
Model Descriptions for likelihoods calculated
Model Al: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
-------
Model R: Yi = Mu + e(i)
Var{e(i)}=SigmaA2
Likelihoods of Interest
Model Log(likelihood) DF AIC
Al -310.209699 5 630.419399
A2 -307.217620 8 630.435240
fitted -310.650699 2 625.301398
R . -315.381614 2 634.763229
Test 1: Does response and/or variances differ among dose
levels
(A2 vs. R)
Test 2: Are Variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Tests of Interest
Test -2*log(Likelihood Ratio) Testdf p-value
Testl 16.328 6 0.0009712
Test 2 5.98416 3 0.1124
Test 3 0.881999 1 0.3477
The p-value for Test 1 is less than .05. There appears
to be a
difference between response and/or variances among the
dose levels.
It seems appropriate to model the data
The p-value for Test 2 is greater than .05. A
homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .05. The model
chosen appears
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
-------
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 336.435
BMDL= 219.137
BMDL computation failed for one or more point on the BMDL curve.
The BMDL curve will not be plotted
-------
Polynomial Model with 0.95 Confidence Level
550
250
0
50 100 150 200
dose
250
300
350
13:5305/172002
-------
Unsavedl. out
Clintrials motor activity BMR=1 SD
Hill Model. $Revision: 2.1 $ $Date: 2000/10/11 21:21:23 $
Input Data File: F:\BMDS\UNSAVED1.(d)
Gnuplot Plotting File: F:\BMDS\UNSAVEDl.plt
Fri May 17 13:48:20 20O2
HMDS MODEL RUN
The form of the response function is:
Yfdose] =.sintercept + v*dose"W (k^n + dose^n)
Dependent variable = MEAN
Independent variable = COLUMN1
rho is set to 0
Power parameter restricted to be greater than 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 13197.2
rho = 0 Specified
intercept = 468
v = -131
n = 1.77566
k = 94.7123
Asymptotic Correlation Matrix of Parameter Estimates
alpha rho intercept v n
k
alpha 10000
0
rho 01000
0
intercept 00100
0
v 0 0 0 1 0
0
n 0 0 0 0 1
0
Page 1
-------
Unsavedl.out
Variable
alpha
rho
intercept
v
n
•• k
Parameter Estimates
Estimate Std Err
13569.6
468 1
-161.338 1
1.34226 1
121.061 1
Table of Data and Estimated Values of Interest
Dose N Obs Mean Obs Std Dev Est Mean
Est Std Dev Chi^2 Res.
u
24.6
107
360
15
15
15
14
468
451
394
337
118
149
78
127
468
451
394
337
116
116
116
116
1.17e-007
-3.09e-007
-2.44e-007
7.32e-007
Model Descriptions for likelihoods calculated
Model Al: Yij = Mu(i)
Var{e(ij)} =
Model A2 Yij
Var{e(ij)}
Model R Yi
Var{e(i)}
Mu(i)
Sigma(i)A2
Mu
Sigma ^
Degrees of freedom for Test Al vs fitted <= 0
Likelihoods of Interest
Model Log(likelihood) DP
Al -310.209699 5
A2 -307.217620 8
fitted -310.209699 5
R -315.881614 2
AIC
630.419399
630.435240
630.419399
635.763229
Test 1: Does response and/or variances differ among dose levels
(A2 vs. R)
Test 2 Are Variances Homogeneous (Al vs A2)
Test 3 Does the Model for the Mean Fit (Al vs. fitted)
Tests of Interest
Page 2
-------
Unsavedl.out
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 17.328 6 0.000605
Test 2 5.98416 3 0.1124
Test 3 2.93312e-011 0 NA
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than 05 A homogeneous variance
model appears to be appropriate here
NA - Degrees of freedom for Test 3 are less than or equal to 0. The Chi-Square
test for fit is not valid
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 246.508
Warning optimum itey not have been found Bad completion code in Optimization r
outine.
BMDL computation failed
Page 3
-------
Hi Mrxtol
550
01
H
Of
250
50
I MO
tSQ
;;;K:I
250 300
3:48 05/1 2002
dose
-------
CONT.OUT
RFC response, BMR=1.0 SD
Power Model. $Revision: 2.1 $ $Date: 2000/10/11 20:57:36 $
input Data File: C:\BMDS\DATA\CONT.(d)
Gnuplot Plotting File: C:\BMDS\DATA\CONT.plt
rue May 14 14:01:03 2002
BMDS MODEL RUN
The form of the response function is:
Y[dose] = control + slope * doseApower
Dependent variable = MEAN
independent variable = Dose
rho is set to 0
The power is restricted to be greater than or equal to 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter convergence has been set to: le-008
Default initial Parameter values
alpha = 28373.3
rho = 0 Specified
• control = 1123
slope = -1807.87
power = -0.361754
Asymptotic Correlation Matrix of Parameter Estimates
alpha
rho
control
slope
alpha
1
-1
0.065
NA
rho
-1
1
-0.056
NA
control
0.065
-0.056
1
NA
slope
NA
NA
NA
NA
power
NA
NA
NA
NA
power NA NA NA NA NA
NA - This parameter's variance has been estimated at zero.
Parameter Estimates
variable Estimate std. Err
Page 1
-------
alpha
rho
control
slope
power
32160.5
0
976.333
-11.4263
1
CONT.OUT
346106
1.59091
37.4625
18.1954
0.423031
Table of Data and Estimated values of interest
Dose N obs Mean Obs std Dev Est Mean
Est Std Dev ChiA2 Res
0 5 1.12e+003
1.8 5 896
6.2 5 795
33.6 5 616
221
134
110
186
976
956
905
592
179
179
179
179
0.818
-0.333
-0.616
0.132
Model Descriptions for likelihoods calculated
Model Al:
Model A2:
Model R:
Yii
var{e(ij)}
Yii
var{e(ij)}
>.
Yf
var{e(i)}
Mu(i) +
Si gmaA2
Mu(i)
Sigma(i)A2
Mu
Si gmaA2
Likelihoods of Interest
Model Log(likelihood) DF AIC
Al -110.300585 5 230.601171
A2 -108.879712 8 233.759424
fitted -113.784950 4 235.569899
R -119.888342 2 243.776683
Test 1: Does response and/or variances differ among dose levels
(A2 vs. R)
Test 2: Are Variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Tests of interest
Test -2*1og(Likelihood Ratio)
Test 1
Test 2
Test 3
22.0173
2.84175
6.96873
df
6
3
1
p-value
.469e-005
0.4167
0.008295
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels.
it seems appropriate to model the data
The p-value for Test 2 is greater than .05. A homogeneous
model appears to be appropriate here
variance
The p-value for Test 3 is less than .05. You may want to try a
Page 2
-------
CONT.OUT
different model
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 15.6948
BMDL = 10.4544
Page 3
-------
Power Mtadel witti 0,95 Confidence Level
UOO
1200
1000
900
600
400
14:01 OS/t4 ^002
0QS9
•
-------
Ma.out
Polynomial Model. $Revision: 2.1 $ $Date: 2000/10/11 17:51:39 $
Input Data File: F:\BMDS\MA.(d)
Gnuplot Plotting File: F:\BMDS\MA.plt
Tue May 14 14:20:26 2002
HMDS MODEL RUN
The form of the response function is:
Y[dose] sObeta_0 + beta_l*dose + beta_2*dose~2
Dependent variable = MEAN
Independent variable = dose
rho is set to 0
The polynomial coefficients are restricted to be negative
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 1
• rho = 0 Specified
beta_0 = 1071.13
beta_l = -56.7007
beta_2 = 0
Parameter Estimates
Variable Estimate Std. Err
alpha 26722.6 8451.07
beta_0 976.316 46.0371
beta_l -11.4246 2.69108
beta_2 0 NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Asymptotic Correlation Matrix of Parameter Estimates
alpha beta_0 beta_l
alpha 1 -0,00012 0.00019
beta_0 -0.00012 1 -0.61
beta 1 0.00019 -0.61 1
Page 1
-------
Ma.out
The following parameter(s) have been estimated at a
boundary
point or have been specified. Correlations are not
computed:
beta_2
Dose
Res.
Table of Data and Estimated Values of Interest
N Obs Mean Obs Std Dev Est Mean Est Std Dev Chi
0 5 1.12e+003
1.8 5 896
6.2 5 795
33.6 5 616
221
134
110
86
976
956
905
592
163
163
163
163
4.49
-1.83
-3.38
0.72
Model Descriptions for likelihoods calculated
Model Al: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2 "&J = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model R: Yi = Mu + e(i
Var{e(i)} = SigmaA2
Likelihoods of Interest
Test 1:
levels
Test 2:
Test 3:
Model Log (likelihood) DF
Al -107.560658 5
A2 -105.022715 8
fitted -111.931872 2
R -118.871191 2
Does response and/or variances differ
(A2 vs. R)
Are Variances Homogeneous (Al vs A2)
AIC
225.121317
226.045431
227.863744
241.742383
among dose
Does the Model for the Mean Fit (Al vs. fitted)
Test
Test 1
Test 2
Test 3
Tests of Interest
-2*log(Likelihood Ratio) Test df
27.697
5.07589
8.74243
6
3
1
p-value
<.0001
0.1663
0.003109
The p-value for Test 1 is less than 05. There appears
to be a
Page 2
-------
Ma.out
difference between response and/or variances among the
dose levels.
It seems appropriate to model the data
The p-value for Test 2 is greater than 05 A
homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is less than .05. You may want
to try a
different model
Benchmark Dose Computation
Specified effect = 1
Risk Type * Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 14.3086
BMDL = 9.73181
Page 3
-------
Polynomial Model with 0.95 Confidence Level
(0
I
I
1400
1200
1000
800
600
Polynomial
BMD Lower Bound
10
dose
30
35
14:2005/142002
-------
CONT.OUT
RFC response, BMR=1.0 SD
Hill Model. SRevision: 2.1 $ $Date: 2000/10/11 21:21:23 $
input Data File: C:\BMDS\DATA\CONT.(d)
Gnuplot Plotting File: C:\BMDS\DATA\CQNT.plt
Tue May 14 14:02:22 2002
————— — — ~ =3a——————J=^s—._— — ——.—__—aS=aS====S===-sa..—^^j—^—•.—.———.••-•.•••====^
BMDS MODEL RUN
The form of the response function is:
Y[dose] = ihtercept + v*doseAn/(kAn + doseAn)
Dependent variable = MEAN
independent variable = Dose
rho is set to 0
Power parameter restricted to be greater than 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
5.
Default initial Parameter values
alpha = 24615.1
rho = 0 specified
•intercept = 1123
v = -507
n = 0.62026
k = 2.95446
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -n
the us r haVe bee" estimated at a boundary point, or have been specified by
and do not appear in the correlation matrix )
v k
0 0
0 0
0 0
1 0
0 1
Parameter Estimates
Page 1
alpha
rho
intercept
v
k
alpha
1
0
0
0
0
rho
0
1
0
0
0
intercept
0
0
1
0
0
-------
221
134
110
186
1.12e+003
923
763
629
152
152
152
152
0.048
-0.175
0.21
-0.0825
CONT.OUT
variable Estimate std. Err.
alpha 23185.3 1
rho 0 1
intercept 1115.7 1
v -533.126 1
n 1 NA
k 3.17243 1
NA - indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Table of Data and Estimated values of interest
0056 * Obs Mean obs std Dev Est Mean Est std Dev chiA2 Res.
0 5 1.12e+003
1.8 5 896
6.2 5 795
33.6 5 616
Model Descriptions for likelihoods calculated
Model Al: „ Yif = Mu(i) + e(ij)
var{e(ij)> = sigmaA2
Model A2: Yii = Mu(i) + e(ij)
var{e(ij)} = Sigma(i)A2
Model R: Yi = MU + e(i)
var{e(i)} = sigmaA2
Degrees of freedom for Test Al vs fitted <= 0
Likelihoods of interest
Model Log(likelihood) OF AIC
Al -110.300585 5 230.601171
A2 -108.879712 8 233.759424
fitted -110.512739 4 229.025477
R -119.888342 2 243.776683
Test 1: Does response and/or variances differ among dose levels
(A2 vs. R)
Test 2: Are Variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Tests of interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 22.0173 6 <.0001
Test 2 2.84175 3 0.4167
Test 3 0.424307 0 NA
The p-value for Test 1 is less than .05. There appears to be a
Page 2
-------
...« L CONT.OUT
difference between response and/or variances among the dose levels.
it seems appropriate to model the data
The p-value for Test 2 is greater than .05. A homogeneous variance
model appears to be appropriate here
NA - Degrees of freedom for Test 3 are less than or equal to 0. The Chi-Square
test for fit is not valid
Benchmark Dose Computation
Specified effect = , 1
Risk Type " = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1.26834
BMDL = 0.383495
Page 3
-------
Hill Model with 0.95 Confidence Level
1400
1200
1000
800
600
400
Hill
BMD Lower Bound
dose
14:0205/142002
-------
CONT.OUT
RFC/total spleen
Power Model. SRevision: 2.1 $ $Date: 2000/10/11 20:57:36 $
input Data File: c:\BMDS\DATA\CONT.(d)
Gnuplot Plotting File: C:\BMDS\DATA\CONT.plt
Tue May 14 15:50:01 2002
BMOS MODEL RUN
The form of the response function is:
Y[dose] - control + slope * doseApower
Dependent variable = MEAN
independent variable = Dose
The power is restricted to be greater than or equal to 1
The variance is to be modeled as var(i) = alpha*mean(i)Arho
Total number of dose groups » 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default initial Parameter values
alpha = 4.76625e+009
rho - 0
control = 265947
slope - -173704
power = -0.0701236
Asymptotic Correlation Matrix of Parameter Estimates
alpha
rho
control
slope
alpha
1
-1
0.39
NA
rho
-1
i
-0.4
NA
control
0.39
-0.4
1
NA
slope
NA
NA
NA
NA
power
NA
NA
NA
NA
power NA NA NA NA NA
NA - This parameter's variance has been estimated at zero.
Parameter Estimates
variable Estimate Std. Err.
alpha 0.00015149 0.0025216
Page 1
-------
rho
control
slope
power
2.52122
237600
-3207.96
1
CONT.OUT
1.36567
17339.2
6498.98
0.554928
Table of Data and Estimated values of interest
Dose N obs Mean obs Std Dev Est Mean Est Std Dev
0 52.66e+005
1.8 5 2.15e+005
6.2 5 2>08e+005
33.6 5 1.3e+005
ChiA2 Res.
1.196+005
3.91e+004
4.22e+004
4.076+004
2.38e+005
2 . 32e+005
2.186+005
1.36+005
7.366+004
7.146+004
6.59e+004
3.446+004
0.385
-0.24
-0.152
0.0108
Model Descriptions for likelihoods calculated
Model Al: Yii = Mu(i) + e(ij)
var{e(ij)} = SigmaA2
Model A2:
Model A3:
Model R:
Yii
var{e(ij)}
Mu(i) +
Sigma(i)A2
Yii = Mu(i)
e(ii)
u(i))
var{e(ij)J>= alpha* (Mu(i))Arho
Yi = MU + e(i)
var{e(i)} = sigmaA2
Likelihoods of Interest
Model Log(likelihood) OF AIC
Al -230.616816 5 471.233632
A2 -225.384337 8 466.768674
A3 -227.759631 6 467.519262
fitted -229.618859 5 469.237718
R -235.936079 2 475.872158
Test 1:
Test 2:
Test 3:
Test 4:
Explanation of Tests
Does response and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (Al vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
Tests of interest
Test -2*log(Likelihood Ratio) d.f
Test 1
Test 2
Test 3
Test 4
21.1035
10.465
4.75059
3.71846
p-value
0.001758
0.015
0.09299
0.05381
Test 4 3.71846 1 0.05381
The p-value for Test 1 is less than .05. There appears to be a
Page 2
-------
_,.„ CONT.OUT
difference between response and/or variances among the dose levels
it seems appropriate to model the data
The p-value for Test 2 is less than .05. A non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is greater than .05. The modeled variance appears
to be appropriate here
The p-value for Test 4 is greater than .05. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
22.9511
15.3302
Page 3
-------
Power Model with 0.95 Confidence Level
400000
350000
300000
o 250000
ce
200000
150000
100000
50000
15:5005/142002
35
-------
Ma.out
PFC/total spleen
Polynomial Model. $Revision: 2.1 $ $Date 2000/10/11 17:51:39 $
Input Data File: F:\BMDS\MA.(d)
Gnuplot Plotting File: F:\BMDS\MA.plt
Tue May 14 16:19:41 2002
HMDS MODEL RUN
The form of the response function is
Y[dose] = -beta_0 + beta_l*dose + beta_2*doseA2
Dependent variable - MEAN
Independent variable = dose
The polynomial coefficients are restricted to be negative
The variance is to be modeled as Var(i) = alpha*mean(i)Arho
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations * 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha - 1
rho * 0
beta_0 = 251869
beta_l * -9322.72
beta_2 = 0
Parameter Estimates
Variable Estimate Std. Err
alpha 216454 344531
rho 0.982068 0.138912
beta_0 188775 63404
beta_l -7724.65 20131.8
beta_2 -1.63675 555.248
Asymptotic Correlation Matrix of Parameter Estimates
alpha rho beta^.0 b«ta_l beta_2
alpha 1 -1 0.14 -0.027 0.017
rho -1 1 -0.053 0.021 0.0082
beta_0 0.14 -0.053 1 -0.79 0.71
beta_l -0.027 0.021 -0.79 1 -1
beta_2 0.017 0.0082 0.71 -1 1
Page 1
-------
Dos*
Re».
Ma.out
Table of Data and Estimated Values of Interest
N Obs Mean Obs Std Dev Est Mean Est Std Dev ChiA2
0 5 2.66e+005
1.8 5 2.15e+005
6.2 5 2.08e+005
33.6 5 1.3e+005
1.19e+005
3.91e+004
4.22e+OQ4
4.07e+004
l.H9e+005
l."5e+005
1.' le+005
-7.:;6e+004
l.Sle+005
1.75e+005
1.57e+005
1.13e+005
2.13
1.14
2.13
8.94
Model Descriptions for likelihoods calculated
t,
Model Al: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2 Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigraa(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = alpha*(Mu(i))~rho
Model R: Yi = Mu + e(i
Var{e(i)} = Sigroa~2
Likelihoods of Interest
Model Log(likelihood) DF AIC
Al -230.616816 5 471.233632
A2 -225.384337 8 466.768674
A3 -227.759622 6 467.519244
fitted -249.295307 5 508.590613
R -235.936079 2 475.872158
Test 1:
levels?
Test 2:
Test 3:
Test 4:
Explanation of Tests
Does response and/or variances differ among Dose
(A2 vs. R)
Are Variances Homogeneous? (Al vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
Tests of Interest
T«»t -2*log(Likelihood Ratio) Test df p-value
Test 1
Test 2
Test 3
Test 4
The p-value for Test 1 is less than 05 There appears
to be a
21.1035
10.465
4.75057
43.0714
6
3
2
1
0.001758
0.015
0.09299
<.0001
Page 2
-------
Ma.out
difference between response and/or variances among the
dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .05 A
non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is greater than 05. The
modeled variance appears
to be appropriate here
The p-value for Test 4 is less than 05 You may want
to try a different
model .*•-
Benchmark Dose Computation
Specified effect = 1
Risk Type « Estimated standard deviations from the control mean
Confidence level = 0.95
HMD = 23.3523
BMDL = > 5.86859
Page
-------
RFC/total spleen
CONT.OUT
Hill Model. SRevision: 2.1 $ $Date: 2000/10/U 21:21:23 $
input Data File: C:\BMDS\DATA\CONT.(d)
Gnuplot Plotting File: C:\BMDS\DATA\CONT.plt
Tue May 14 15:50:56 2002
BMDS MODEL RUN
The form of the response function is:
Y[dose] = intercept + v*doseAn/(kAn -i- doseAn)
Dependent variable = MEAN
independent variable = Dose
Power parameter restricted to be greater than 1
The variance is to be modeled as Var(i) - alpha * mean(i) A rho
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default initial Parameter values
alpha
rho
intercept
v
n
k
1
2.30623
265947
-135762
0.475261
9.59273
the user,
alpha
rho
i ntercept
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -n
have been estimated at a boundary point, or have been specified by
and do not appear in the correlation matrix )
alpha rho Intercept v k
1 -1 NA -0.99
-1 1 NA 0.99
NA NA NA NA NA
V
k
-0.99
0.21
0.99 NA
-0.22 NA
1
-0.62
Parameter Estimates
Page 1
-------
variable
alpha
rho
intercept
v
n
k
Estimate
9.6535e-006
2.73759
265131
-136181
1
4.00666
CONT.OUT
Std. Err.
1.0€893e-008
8.76021e-005
1.22396
2.5363
MA
8.616366-005
NA - indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Table of Bata and Estimated values of interest
Dose N Obs Mean Obs Std Dev Est Mean Est Std Dev chiA2 Res.
0 5 2.66e+005
1.8 5 2.156+005
6.2 5 2.086+005
33.6 5 1.3e+005
1.19e+005
3.916+004
4.226+004
4.076+004
2.656+005
2.236+005
1.826+005
1.436+005
8.24e+004
6.56+004
4.94e+004
3 . 566+004
0.0099
-0.127
0.511
-0.373
Model Descriptions for likelihoods calculated
Model Al: vit-= Mu(i) + e(ij)
vaf{e(ij)> = sigmaA2
Model A2: Yii = Mu(i) + e(ij)
var{e(ij)> = Sigma(i)A2
Model A3:
Model R:
Yii
var{e(ij)}
Yi
var{e(i)}
Mu(i) + e(ii)
alpha*(Mu(i))Arho
Mu
Si gmaA2
Degrees of freedom for Test A3 vs fitted <- 0
Likelihoods of interest
Model Log (1 i kel i hood) DF
Al -230.616816 5
A2 -225.384337 8
A3 -227.759620 6
fitted -228.438245 5
R -235.936079 2
AIC
471.233632
466.768674
467.519240
466.876489
475.872158
Explanation of Tests
Test 1: Does response and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are variances Homogeneous? CAl vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
Tests of interest
Page 2
-------
CONT.OUT
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 21.1035 6 0.001758
Test 2 10.465 3 0.015
Test 3 4.75057 2 0.09299
Test 4 1.35725 0 NA
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
it seems appropriate to model the data
The p-value for Test 2 is less than .05. A non-homogeneous variance
model appears to be appropriate
s
The p-value for Test 3 is greater than .05. The modeled variance appears
to be appropriate here
NA - Degrees of freedom for Test 4 are less than or equal to 0. The Chi-square
test for fit is not valid
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = *• 6.13974
warning: optimum may not have been found. Bad completion code in Optimization
routine.
•
BMDL computation failed.
Page 3
-------
Hill Model
CD
CO
I
8
cc
i
100000
50000
10
15:5005/142002
dose
-------
CONT.WT
Antibody titer
Power Model. $Revision: 2.1 $ SDate: 2000/10/11 20:57:36 $
input Data File: C:\BMDS\DATA\CONT.(d)
Gnuplot Plotting File: C:\BMDS\DATA\CONT.plt
Tue May 14 15:52:46 2002
BMOS MODEL RUN
The form of the response function is:
Y[dose] - cdntrol + slope * doseApower
Dependent variable = MEAN
independent variable = Dose
rho is set to 0
The power is restricted to be greater than or equal to 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
parameter convergence has been set to: le-008
Default initial Parameter Values
alpha = 0.0098315
rho = 0 Specified
control = 0.446
slope = -1.46784
power = -0.592416
Asymptotic Correlation Matrix of Parameter Estimates
alpha
rho
control
slope
alpha
1
0.99
0.091
NA
rho
0.99
1
0.041
NA
control
0.091
0.041
1
NA
slope
NA
NA
NA
NA
power
NA
NA
NA
NA
power NA NA NA NA NA
NA - This parameter's variance has been estimated at zero
Parameter Estimates
variable Estimate Std. Err
Paae 1
-------
CONT.OUT
alpha
rho
control
slope
power
0.00898754
0
0.402624
-0.00443496
1
0.018154
1.9925
0.0174706
0.0109976
0.658352
Table of Data and Estimated values of interest
Dose N obs Mean obs Std Dev Est Mean
Est Std Dev chiA2 Res.
0
1.8
6.2
33.6
5
5
5
5
0.446*
0.392
0.325
0.263
0.087
0.152
0.042
0.083
0.403
0.395
0.375
0.254
0.0948
0.0948
0.0948
0.0948
0.458
-0.0279
-0.529
0.0991
Model Descriptions for likelihoods calculated
Model Al: vii
var{e(ij)}
Model A2: Yii
var{e(ij)>
Model R:
• var{e(i)} =
Mu(i) + e(ij)
sigmaA2
Mu(i) + e(ij)
sigma(i)A2
Mu + e(i)
sigmaA2
Likelihoods of interest
Model Log(1i keli hood) DF
Al 38.453073 5
A2 42.155047 8
fitted 37.119156 4
R 33.214605 2
AIC
-66.906146
-68.310094
-66.238312
-62.429209
Test 1: Does response and/or variances differ among dose levels
(A2 vs. R)
Test 2: Are Variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Tests of interest
Test -2*log(|.ikel Ratio) df
Test 1 17,8809 6
Test 7 7.40395 3
~o« 3 2..€€783 i
p v a 1 u •
.rOG'li.-.'i 1
L.rrrn.5
The p-value for Test 1 is less than ,05, Ther-p .iti.pr-CL, ic be a
difference between variances amonq ~ie d IM> levels
rt appropriate to the data
The p-value for Tust 2 is greater
mode) appears to be appropriate here
A homogeneous va r i a n ce
The p-value for Test 3 is greater than .05. The model chosen appears
Page 2
-------
CONT.OUT
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 21.3762
BMBL = 13.0626
Page 3
-------
Power Model with 0.95 Confidence Level
ft
c
0.6
0.5
0.4
0.3
0.2
0.1
15:5205/142002'
BMOlj
30 35
-------
Ma.out
antibody titer
Polynomial Model. $Revision: 2.1 $ $Date: 2000/10/11 17:51:39 $
Input Data File: F:\BMDS\MA.(d)
Gnuplot Plotting File: F:\BMDS\MA.plt
Tue May 14 16:22:39 2002
HMDS MODEL RUN
The form of the response function is:
Y[dose] = «-beta_0 + beta_l*dose + beta_2*doseA2 +
Dependent variable = MEAN
Independent variable = dose
rho is set to 0
The polynomial coefficients are restricted to be negative
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence .has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.0098315
rho = 0 Specified
beta_0 = 0.439645
beta_l = -0.0221168
beta_2 = 0
Variable
alpha
beta_0
beta_l
beta_2
Parameter Estimates
Estimate
0.00898754
0.402624
-0.00443496
0
Std. Err
0.00284211
0.0266986
0.00156066
NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Asymptotic Correlation Matrix of Parameter Estimates
alpha
beta_0
beta_l
alpha
1
-2e-008
5.4e-009
beta_0
-2e-008
1
-0.61
beta_l
5.4e-009
-0.61
1
Page 1
-------
Ma.out
The following parameter(s) have been estimated at a
boundary
point or have been specified. Correlations are not
computed:
beta.
Table of Data and Estimated Values of Interest
Dose
Res.
0
1.8
6.2
33.6
N
5
5
5
5
Obs Mean
0.446
0.392
0.325
0.263
Obs Std Dev
0.087
0.152
0.042
0.083
Est Mean
0.403
0.395
0.375
0.254
Est Std Dev
0.0948
0.0948
0.0948
0.0948
Chi-2
2.29
-0.139
-2.64
0.495
Model Descriptions for likelihoods calculated
Model Al Yij
Var{e(ij)>
Model A2
var{e(ij)>
Model R: Yi
Var{e(i)}
Mu(i)
Sigma
Mu(i)
Mu
Sigma.^2
Likelihoods of Interest
Model Log(likelihood) DF AIC
Al 38.453073 5 -66.906146
A2 42.155047 8 -68.310094
fitted 37.119156 2 -70.238312
R 33.214605 2 -62.429209
Test 1 Does response and/or variances differ among dose
levels
(A2 vs. R)
Test 2: Are Variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs fitted)
Test
Test 1
Test 2
Test 3
Tests of Interest
-2*log(Likelihood Ratid) Test df
17.8809
7.40395
2.66783
6
3
1
p-value
0.0004654
0.06008
0.1024
The p-value for Test 1 is less than .05. There appears
to be a
Page 2
-------
Ma.out
difference between response and/or variances among the
dose levels.
It seems appropriate to model the data
The p-value for Test 2 is greater than 05. A
homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .05 The model
chosen appears
to adequately describe the data
Benchmark Dose Computation
Specified effect = i
Risk Type = Estimated standard deviations from the control mean
Confidence level » 0.95
BMD = 21.3762
BMDL = 13.0626
Page 3
-------
Polynomial Model with 0.95 Confidence Level
0)
<0
I
-------
CONT.OUT
Antibody titer
Hill Model. SRevision: 2.1 $ $Date: 2000/10/11 21:21:23 $
input Data File: C:\BMDS\DATA\CONT.(d)
Gnuplot Plotting File: C:\BMDS\DATA\cONT.plt
Tue May 14 15:53:39 2002
BMDS MODEL RUN
The form of the response function is:
Y[dose] = intercept + v*doseAn/(kAn + doseAn)
Dependent variable = MEAN
independent variable = Dose
rho is set to 0
Power parameter restricted to be greater than 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default initial Parameter values
alpha = 0.00678961
rho = 0 Specified
.intercept = 0.446
v = -0.183
n = 1.39743
k = 4.26269
Asymptotic Correlation Matrix of Parameter Estimates
k
alpha
0
rho
0
intercept
0
V
0
n
0
k
1
alpha
1
0
0
0
0
0
rho
0
1
0
0
0
0
i ntercept
0
0
1
0
0
0
V
0
0
0
1
0
0
n
0
0
0
0
1
0
Page 1
-------
CONT.OUT
Variable
alpha
rho
intercept
v
n
k
Parameter Estimates
Estimate
0.0078652
0
0.446
-0.200218
1.14792
4.2868
Std.
Err.
1
1
1
1
1
1
Table of .Data and Estimated values of interest
Dose N obs Mean obs Std Dev Est Mean
Est Std Dev chiA2 Res.
0
1.8
6.2
33.6
5
5
5
5
0.446
0.392
0.325
0.263
0.087
0.152
0.042
0.083
0.446
0.392
0.325
0.263
0.0887
0.0887
0.0887
0.0887
-1.73e-008
3.45e-007
1.51e-007
-1.34e-007
Model Descriptions for likelihoods calculated
Model Al: YiJ-= Mu(i) + e(ii)
vaf{e(ij)} = sigmaA2
Model A2:
: Yii - Mu(i) + e(ij
var{e(ij)} = sigma(i)A2 ,
Model R: Yi = MU + e(i)
var{e(i)} = sigmaA2
Degrees of freedom for Test Al vs fitted <= 0
Likelihoods of Interest
Model Log(likelihood) DF
Al 38.453073 5
A2 42.155047 8
fitted 38.453073 5
R 33.214605 2
AIC
-66.906146
-68.310094
-66.906146
-62.429209
Test 1: Does response and/or variances differ among dose levels
(A2 vs. R)
Test 2: Are variances Homogeneous (Al vs A2)
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Test
Test 1
Test 2
Test 3
Tests of interest
-2*1og(Likelihood Ratio) Test df
17.8809
7.40395
5.46761e-007
6
3
0
p-value
0.0004654
0.06008
NA
The p-value for Test 1 is less than .05. There appears to be a
Page 2
-------
CONT.OUT
difference between response and/or variances among the dose levels
it seems appropriate to model the data
The p-value for Test 2 is greater than .05. A homogeneous variance
model appears to be appropriate here
NA - Degrees of freedom for Test 3 are less than or equal to 0. The Chi-square
test for fit is not valid
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 3.51089
BMDL = 0.726794
Page 3
-------
Hill Model with 0.95 Confidence Level
o 0.4
8
cc
BMD,Lower Bound
15:5305/142002
------- |