HEALTH EFFECTS AND DOSE-RESPONSE ASSESSMENT
FOR HYDROGEN CHLORIDE FOLLOWING SHORT-TERM EXPOSURE
ASSIGNMENT NO. 2
FINAL REPORT
prepared for
Air Risk Information Support Center
Office of Air Quality Planning and Standards
and Office of Health Evaluation Assessment
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
EPA Work Assignment Managers: Daniel Guth and Holly Reid
prepared by
Clement International Corporation
K S Crump Division
1201 Gaines Street
Ruston, Louisiana 71270
US. Erwironrn
jvr'icn 5, il ,r
loci /JWest J'C:::
1991
-------
FOREWORD
This report has been prepared by Clement International Corporation for the U.S.
Environmental Protection Agency's Office of Air Quality, Planning and Standards,
Pollution Assessment Branch (OAQPS/PAB) in response to a request to investigate
approaches to development of air quality standards for short-term exposures to air toxics.
The project was directed by Mr. David Patrick and the principal investigators were Dr.
Annette Shipp, Dr. Richard Howe, Dr. Carl Watson, and Ms. Mary Lee Hogg.
Dr. Daniel Guth of ECAO and Ms. Holly Reid of OAQPS/PAB served as the Work
Assignment Managers. The authors wish to thank Dr. John Vandenberg and Beth
Hassett-Sipple with OAQPS/PAB, and Drs. Richard Hertzberg and Linda Knauf with
ECAO for their expert peer review.
-------
DISCLAIMER
This report has been reviewed by the Office of Air Quality, Planning and
Standards, Pollution Assessment Branch, U.S. Environmental Protection Agency, and has
been approved for distribution as received from Clement International Corporation.
Approval does not signify that the contents reflect the views and policies of the U.S.
Environmental Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
n
-------
TABLE OF CONTENTS
Page
1. Introduction 1
2. Approaches Used for Short-Term Exposure to Hydrogen Chloride 3
3. New Method for Quantitative Estimates of Risk of Noncancer Effects
from Short-Term Exposure 10
3.1 Background 10
3.2 Application of a New Method for Assessing Risk from
Short-Term Exposure to Hydrogen Chloride 15
3.2.1 Categorizing by Severity Index 16
3.2.2 Human Equivalent Concentration Estimation 25
3.2.3 Description of the Dose-Duration-Response Model 27
3.2.4 Results 30
3.3 Conclusions and Relevance to Human Health 52
4. References 57
Appendix A: Discussion of Acute Animal Toxicity Data A-1
Appendix B: Animal Bioassay Data for Input into the
Dose-Duration-Response Model B-l
Appendix C: Derivation of Maximum Likelihood Estimate Bounds
and Lower Confidence Limits C-l
Appendix D: Computer Output for Dose-Duration-Response Model D-l
111
-------
-------
LIST OF TABLES
Page
Table 1. Physiological Response to Various Concentrations
of Hydrochloric Acid Gas 5
Table 2. Various Effect Levels and Their Definition 11
Table 3. Preliminary Ranking System for Evaluation of the Acute
Inhalation Toxicity in Animals Exposed to Hydrogen Chloride 18
Table 4. Response Levels (Ranked in Order of Increasing Severity
of Toxic Effect) Considered in Deriving Inhalation Reference
Concentrations (RfCs) for Systemic Toxicants 19
Table 5. Reading Data Notation 20
Table 6. Maximum Likelihood Estimates and 95% Lower Bounds for
P(s>l | d,T) = 0.05 with Exposure to Hydrogen Chloride
with for Various Durations of Exposure 44
Table 7. Maximum Likelihood Estimates and 95% Lower Bounds for
P(s>l | d,T) = 0.01 with Exposure to Hydrogen Chloride
with for Various Durations of Exposure 46
Table 8. Relative Comparison of Several Regulatory Criteria and
Human Data with Maximum Likelihood and 95% Lower Bounds
on P(s>i | d,T)=0.01 47
Table 9. Relative Comparison of Several Regulatory Criteria and
Human Data with Maximum Likelihood and 95% Lower Bounds
on P(s>i | d,T)=0.05 48
IV
-------
-------
LIST OF FIGURES
Page
Figure 1. Effect-Dose-Duration Plot of All Relevant Human
and Animal Toxicity Data for an Example Chemical 13
Figure 2. Animal to Human Extrapolation Based on Total Lung Surface Area
Maximum Likelihood Estimate of P(s>i|d,T)=0.05 31
Figure 3. Animal to Human Extrapolation Based on Total Lung Surface Area
95% Lower Bound of Maximum Likelihood of P(s>i | d,T)=0.05 32
Figure 4. Animal to Human Extrapolation Based on Total Lung Surface Area
Maximum Likelihood Estimate of P(s>i | d,T)=0.01 33
Figure 5. Animal to Human Extrapolation Based on Total Lung Surface Area
95% Lower Bound of Maximum Likelihood of P(s>i | d,T)=0.01 34
Figure 6. Animal to Human Extrapolation Based on Total Lung and
Extrathoracic Surface Area. Maximum Likelihood Estimate of
P(s>i | d,T)=0.05 35
Figure 7. Animal to Human Extrapolation Based on Total Lung and
Extrathoracic Surface Area. 95% Lower Bound of Maximum
Likelihood of P(s>i | d,T)=0.05 36
Figure 8. Animal to Human Extrapolation Based on Total Lung and
Extrathoracic Surface Area. Maximum Likelihood Estimate of
P(s>i | d,T)=0.01 37
Figure 9. Animal to Human Extrapolation Based on Total Lung and
Extrathoracic Surface Area. 95% Lower Bound of Maximum
Likelihood of P(s>i | d,T)=0.01 38
Figure 10. Animal to Human Extrapolation Based on Administered Animal Doses
Maximum Likelihood Estimate of P(s>i | d,T)=0.05 39
Figure 11. Animal to Human Extrapolation Based on Administered Animal Doses
95% Lower Bound of Maximum Likelihood of P(s>i | d,T)=0.05 40
Figure 12. Animal to Human Extrapolation Based on Administered Animal Doses
Maximum Likelihood Estimate of P(s>i | d,T)=0.01 41
Figure 13. Animal to Human Extrapolation Based on Administered Animal Doses
95% Lower Bound of Maximum Likelihood of P(s>i | d,T)=0.01 42
-------
1. Introduction
The Environmental Protection Agency's (EPA) Office of Air Quality, Planning
and Standards (OAQPS) and Office of Health and Environmental Assessment (OHEA)
are responsible for providing technical support on health-related issues pertaining to air
toxics. This support is provided to EPA Regional Offices and to State and local air
pollution control agencies through the Air Risk Information Support Center (Air RISC),
a joint project of OAQPS and OHEA. Several requests to Air RISC for information on
the potential adverse health effects resulting from short-term exposure to hydrogen
chloride have been received.
Information concerning the toxicity of hydrogen chloride is available, and the
potential adverse effects that may occur with short-term human exposure to this chemical
are known. Criteria have been established in many states for ambient air concentrations
for environmental exposure to toxic air pollutants. Several states have established
ambient air concentrations for short-term exposure to hydrogen chloride (McCune 1990,
North Dakota State Department of Health 1990, Rhode Island Department of
Environmental Management 1990). In addition, the Occupational Safety and Health
Administration (OSHA) has established a ceiling for occupational exposure to hydrogen
chloride. However, the range of exposure concentrations of hydrogen chloride over a
range of exposure durations that is likely to be without adverse effects in exposed
populations is not well characterized. In evaluating acute exposure situations, the risk
assessor/risk manager may have to make decisions regarding the impact of ambient levels
on the exposed population. Ideally, sufficient dose-duration-response information for
-------
short-term exposure to hydrogen chloride may provide the risk assessor/risk manager with
the necessary information for characterizing potential public health risks and making
informed decisions on appropriate mitigation procedures.
The purpose of this project was to investigate the development of dose-duration-
response information for short-term exposure to hydrogen chloride. This approach is one
alternative for providing the risk assessor/risk manager a readily available guide for
decision making with respect to short-term exposure to hydrogen chloride. The
objectives of this project then were to provide a detailed analysis of the health effects
associated with short-term exposure to hydrogen chloride and to propose an approach for
evaluating the hazard to humans exposed to hydrogen chloride for short periods of time.
The technical approach .used to accomplish these objectives consisted of several
elements: the first was to conduct a critical review of the literature to assess the relevant
experimental data and to identify the types of adverse health effects and the dose-
duration at which these effects were observed (discussed in Appendix A); secondly, the
approaches currently used by EPA and state and local air pollution programs for
assessment of human health effects from short-term exposure to hydrogen chloride were
reviewed (Chapter 2); third, a dose-duration-response method was proposed and
demonstrated (Chapter 3).
-------
2. Approaches Used for Short-Term Exposure to Hydrogen Chloride
OSHA and various state and local agencies responsible for air pollution programs
have established air concentration criteria for hydrogen chloride, some of which are
intended for a specified duration of exposure. Those approaches for which information
were available for hydrogen chloride are reviewed here. The intent of this review was to
consider the scientific basis, the assumptions maintained, and the uncertainties associated
with each approach along with the utility of the particular approach for evaluating short-
term exposures to hydrogen chloride. Current regulations or recommended standards on
hydrogen chloride exposure are discussed in the following paragraphs, including
standards for exposure in the workplace, as well as exposure to the general public. In
addition, other toxicity criteria, some of which were developed for evaluating acute
exposures under emergency response conditions, were reviewed; however, only those with
a specific value for hydrogen chloride are included in this discussion.
The American Conference of Governmental Industrial Hygienists (ACGIH) has
set a threshold limit ceiling for exposure to hydrogen chloride in air of 5 ppm
(~7 mg/m3). The threshold limit ceiling is the concentration of a chemical that should
•
not be exceeded during any part of the working exposure (ACGIH 1987). The rationale
for choosing this threshold limit value (TLV) is given in the ACGIH Documentation of
Threshold Limit Values (1980) and refers to reviews of hydrogen chloride toxicity by
Elkins (1959) and Patty (1963). Elkins (1959) reported that hydrogen chloride is
"...immediately irritating when inhaled in concentrations of 5 ppm or more...," and
proposed a maximum allowable concentration of 5 ppm (~7 mg/m3). However, the
-------
scientific basis of this conclusion was not provided. Patty (1963), citing Henderson and
Haggard (1943), reports that, in human volunteers, increasingly severe irritation of the
nose and throat occurred from exposure to 10 to 100 ppm (15 to 150 mg/m3) hydrogen
chloride for various periods of time (Table 1). Other studies cited in the ACGIH
documentation referred to individuals exposed to higher doses at which adverse effects
resulted.
OSHA has also set a Permissible Exposure Level (PEL) for hydrogen chloride in
air of 5 ppm (~7 mg/m3), apparently in accordance with ACGIH (HSDB 1991). In
addition, OSHA has established an Immediately Dangerous to Life and Health (IDLH)
value of 100 ppm (150 mg/m3) (HSDB 1991). The IDLH is established for workers in
the event of respirator failure, thereby allowing for a 30 minute escape time.
Based on the National Air Toxics Information Clearinghouse (NATICH) data
base (1988), nine states have established acceptable ambient concentration guidelines or
standards for hydrogen chloride. Agencies responsible for setting these guidelines in six
states were contacted for information on i:he basis of their standard, and four states
supplied the requested information.
State agencies of Massachusetts, North Carolina, and North Dakota based their
•guidelines on the ACGIH occupational ceiling limit of 5 ppm (~7 mg/m3), with each
state using a different uncertainty factor or other modifying factors. North Carolina
divided the ACGIH value by an uncertainty factor of 10, while North Dakota divided this
value by a factor of 100, resulting in guidance levels of 0.47 ppm (0.7 mg/m3) (15 minutes
averaging time) and 0.047 ppm (0.07 mg/m3) (1 hour averaging time), respectively
(McCune 1990, North Dakota State Department of Health 1990). The rationale for
using these uncertainty factors was not provided. The Commonwealth of Massachusetts
-------
Table 1
Physiological Response to Various
Concentrations of Hydrochloric Acid Gas
Hydrogen chloride in ppm (mg/m3)
Maximum concentration allowable
for prolonged exposure
Maximum concentration tolerable
for a few hours
Maximum concentration tolerable
for 1 hour
Dangerous for even short
exposures
10
10 to 50
50 to 100
(15)
(15 to 75)
(75 to 150)
1,000 to 2,000 (1,500 to 3,000)
SOURCE: Henderson and Haggard (1943).
-------
used a more complex approach and divided the ACGIH standard by the following
factors:
4.2 (ratio of continuous exposure/occupational exposure; 168 hrs/40 hrs)
1.75 (ratio of breathing rate and weight of child to adult)
10 (uncertainty factor for high risk groups)
10 (uncertainty factor for quality of data)
These adjustments result in what Massachusetts terms an AAL (Allowable Ambient
Level) of 0.007 ppm (0.01 mg/m3) (for 24 hours) (Commonwealth of Massachusetts
1990).
The State of Rhode Island derived its standard based on the experimental data of
Machle et al (1942) after applying dose conversion and uncertainty factors. Machle et al.
exposed three rabbits. :I;ree guinea pigs, and a monkey to 34 or 67 ppm (50 or
100 mg/m3) hydrogen chloride for 6 hours/day, 5 days/week for 4 weeks. Various effects
were reported at the higher dose; therefore, a no-observed-effect-level (NOEL) of
34 ppm (50 mg/m3) was determined. From these data the State derived a one-hour
standard under the assumption that, "...effects of hydrogen chloride exposure appear to
be more closely related to concentration rather than concentration times time" (Rhode
Island Department of Environmental Management 1990). Animal doses were first
•
converted to units of mg/kg/hr based on species-specific breathing rates and body weights.
These doses were then converted to equivalent human hourly doses (in mg/m3) by
multiplying by human body weight/hourly breathing rate (70 kg/1 m3/hr). The equivalent
human hourly doses were then converted to a surface area equivalent dose [inverse of
(human weight/animal weight)1^3] and divided by an uncertainty factor of 10 for
intraspecies variation. The adjusted doses for the three different species ranged from
-------
1.3 to 1.5 ppm (1.9 to 2.2 mg/m3). The State of Rhode Island chose to use 1.3 ppm
(2 mg/m3) as the one-hour TLV (Rhode Island Department of Environmental
Management 1990).
Other short-term limits have been proposed. Short-Term Public Limits (STPL)
and Public Emergency Limits (PEL) were recommended by the Subcommittee on
hydrogen chloride of the National Research Council in a 1971 report (as cited in
GEOMET 1981). The committee believed that the proposed limits, listed below, would
not present any health hazard but might be detectable by odor and could cause minimal
irritation. The scientific basis of these values was not presented.
Exposure time (min)
STPL
10
30
60
5 h/d, 3-4 d/mo
PEL
10
30
60
Hydrogen chloride
concentration
ij'i puiii frng/m t
4(6)
2(3)
2(3)
0.7 (1)
7(10)
3.4 (5)
3.4 (5)
The RD50 (concentration necessary to reduce respiratory rate by 50%) has been
proposed by Alarie and colleagues as a quantitative tool for the prediction of acute
human sensory irritation, i.e., a burning sensation in the eyes, nose, and throat of an
exposed individual, that can be used to establish standards under emergency (/>.,
Emergency Exposure Limits) or short-term conditions (Le., TLVs, STELs). The RD50 is
an experimentally derived value and is the concentration at which inhalation exposure
7
-------
(usually from 10 to 15 minutes) to a chemical produces a 50% reduction in the
respiratory rate of mice. Measurement of the percent decrease in respiratory rate is
utilized to evaluate the degree of sensory irritation. While the RI>5oS are based on
decreases in respiratory rate in mice, they have been reasonable predictors of whether
sensory irritation would occur in humans in 50 of 51 chemicals evaluated (Barrow et al.
1977). Using values of 1/10 RD50 and 1/100 RD50, an estimate of a TLV for hydrogen
chloride has been proposed by Barrow et al (1977) to range from 3 to 31 ppm (4.5 to
46.5 mg/m3). Based on these results, the authors suggested that at the lower limit of
3 ppm (4.5 mg/m3, 1/100 RD50), little if any sensory irritation is expected to occur in
humans, while at the upper limit of 31 ppm (46.5 mg/m3, 1/10 RD50) only slight irritation
(/.«., slight stinging/burning of the eyes, nose, and throat) is expected. Exposure to
hydrogen chloride at the RD50 (309 ppm, 465 mg/m3) was predicted to be incapacitating
to humans due to sensory irritation (Barrow et al. 1977).
As a comparison to the other hydrogen chloride criteria developed, EPA's chronic
reference concentration (RfC) is included in this discussion. In general, the RfC is an
estimate of a daily exposure to the human population (including sensitive subgroups) that
is likely to be without an appreciable risk of deleterious effects during a lifetime. An
inhalation RfC for hydrogen chloride was calculated based on the study by Sellakumar
et al. (1985) (EPA 1991). In this lifetime study, rats were exposed to hydrogen chloride
for 6 hours/day for 4.7 days/week for life and a lowest-observed-adverse-effect level
(LOAEL) of 15 mg/m3 (10 ppm) (the only dose tested) was reported based on the
occurrence of hyperplasia of the nasal mucosa, larynx, and trachea at that dose. The
adjusted LOAEL (adjusted by 6/24 and 4.7/7) was approximately 2.5 mg/m3 (1.7 ppm).
8
-------
The LOAEL (adj) of 2.5 mg/m3 (1.7 ppm) was further converted to a human equivalent
concentration (LOAELHEC) by use of the following conversion:
2.5 mg/m3 x
(BRA/SAA)
(BRH/SAH)
= 6.5 mg/m3
where,
BRA = animal breathing rate of 0.5 m3/day
SAA = regional surface area (cm2) of respiratory region (extrathoracic and
tracheobronchial) affected in animals (49.2)
BRH = human breathing rate of 20 m3/day,
SAH = regional surface area (cm2) of respiratory region (extrathoracic and
tracheobronchial) affected in humans (5213)
The LOAEL (HEC) was then divided by 1000 (factors of 10 for intraspecies variability,
interspecies extrapolation, and use of a LOAEL instead of a NOAEL), and rounded to
one significant figure. The resulting inhalation RfC is 7xlO"3 mg/m3 (5.0xlO'3 ppm).
-------
3. New Method for Quantitative Estimates of Risk of Noncancer Effects
from Short-Term Exposure
3.1 Background
The evaluation of toxicity data for noncarcinogens is complicated by the
multiplicity of possible endpoints, and the variation both in the severity of effect and in
the response rate. As stated by Hertzberg and Knauf (1989) standard dose-response
models most often assume a fixed severity (e.g., lethality) and endpoint (e.g., cancer), and
would then need to be generalized into a multivariate form to be applicable to
noncarcinogenic effects. Since response rates for noncancer endpoints are often poorly
reported, multivariate dose-response models are seldom developed, and "dose-response"
analysis usually relates dose only to the severity of the observed effects, often in a
qualitative way (Hertzberg and Knauf 1989, Guth et al. 1991).
Dourson et al. (1985) presented an approach, described in the following
paragraphs, which assigns a severity description to ordered categories and graphs the
"dose-category" relationship. In this approach, each experimental group in a bioassay is
summarized by five variables: human equivalent exposure concentration (mg/m3), human
equivalent exposure durations (years), effect levels (NOEL, NOAEL, NOFEL, etc., see
Table 2), target organ, and study usefulness. Usefulness is based on quantitative
accuracy (e.g., sample size, measurement error, whether certain values are reported or
must be assumed) and qualitative characteristics of the experiment (e.g., extent of
pathological workup, existence of widespread adverse health effects in control group,
unnatural route of exposure). These five parameters are graphed in the following way.
10
-------
Table 2
Various Effect Levels and Their Definition
Effect
Levela
Symbol
Definition5
PEL * Frank-effect level. That exposure level which produces
unmistakable adverse effects, such as irreversible functional
impairment or mortality, at a statistically or biologically
significant increase in frequency of severity between an exposed
population and its appropriate control.
AEL • Adverse-effect level. That exposure level at which there are
statistically or biologically significant increases in frequency or
severity of effects between the exposed population and its
appropriate control.
NOFEL A No-observed-frank-effect level. That exposure level at which
there are no statistically or biologically significant increases in
frequency or severity of frank effects between an exposed
population and its appropriate control. Experimenters may or
may not have looked for other adverse effects.
NOAEL O No-observed-adverse-effect level. That exposure level at which
there are no statistically or biologically significant increases in
frequency or severity of adverse effects between the exposed
population and its appropriate control. Effects are produced
at this level, but they are not considered to be adverse.
NOEL o No-observed-effect level. That exposure level at which there
are no statistically or biologically significant increases in
frequency or severity of effects between the exposed
population and its appropriate control.
a Listed in order of decreasing severity.
b Adverse effects are considered to be functional impairment or pathological lesions that
may affect the performance of the whole organism, or that reduce an organism's
ability to respond to an additional challenge (EPA 1980).
SOURCE: Dourson et aL 1985.
11
-------
The first two parameters (exposure concentration and duration) are continuous variables
which are usually defined by the experimental protocol; thus, these are used as
independent axis variables. The remaining three are categorical variables, which are
denoted by symbols and placed on the graph at the corresponding "dose-duration"
location.
The categorical variables (effect severity, study usefulness, and target organ) are
represented by the shape, size or labeling of graphics symbols (see Figure 1). Each
symbol's shape denotes one of the "effect level" designations used for toxicants as defined
in Table 2. These designations represent a simple ranking scheme for the severity of the
effect. The size of the symbol denotes the relative usefulness of the study for
quantitative health risk estimation, with the larger symbol denoting the more appropriate
study.
By partitioning the dose rate axis into areas expected to cause: (a) gross toxicity
and death, (b) adverse effects to various organ systems, (c) no adverse effects, and (d)
levels which are now considered "safe", /.&, the RfD (see Figure 1 and Table 2), an
overview of the expected toxicity in a group of individuals with a known or estimated
exposure pattern (both time and dose) may be accomplished. Lines are drawn to
approximate the boundary between effect levels.
This approach could similarly be related to the RfC concept and could be
extended to exposures shorter than the subchronic or chronic studies used for RfC
derivation. Operationally, RfCs are calculated by dividing the human equivalent
concentration of a NOEL. NOAEL, or the lowest observed AEL (LOAEL) derived from
human or animal toxicity studies by required uncertainty factors. One possible approach
is to interpolate along the NOAEL curve, defined as the line beneath which no AEL is
12
-------
Figure 1
100.000 -
10.000
PEL
1.000 -
EQUIVALENT
HUMAN
DOSE
Img/d)
0.07
0.7 7.0
EQUIVALENT HUMAN DURATION (yoiiu)
Oo,e
«v.l. irrflc.ted by .yrt.1. .re
- K , eonw'rled •* • ^^ •«•«•«• •«• «««" «<> •PPTMIMU the equlv.lent liu** do.e.
.r, di« ted by ih..pproprl.te •pecUt Illetpan to yl.ld . (r.ctlon. which. Nhen ^ItlplUd by 70 yt.n (th. ...uMd -vcr.fl. h«Mn
ll.ip*n) glv.. th* corrttpoodlog position on th. ..«,!.. study uselulnti. I. dtMted by .y^>ol .Ife/ f.rl.t «r««na .r. iv (liver) »
r»pro«k« v. .r|M>. GR (.^th reduction), .nd IP (.pl.en). The do.. ..I, |. divided |j,,.\r... t.pect«l to c.^ro.! u.|e H 2nd de.th (A)
cflectt !•), nor,-»dv*rM effect* (C). or no effect! (0). (Sourc*: oourton 19Bo) tomciiy ana aeith (A).
-------
observed, to the desired duration, and then divide the corresponding dose rate by an
uncertainty factor(s) to obtain the estimate analogous to the RfC for that duration.
Since these lines are "sight" fitted and somewhat arbitrary in nature, it is easy to
see that two experimenters could get lines which differed by a large amount. To remove
the objection of arbitrarily drawn lines, categorical regression has been suggested.
Hertzberg (1989) suggests that the form of categorical regression for risk estimation
should be the logistic model. The numerical score assigned to each category (defined
below) only serves as an index of the ordering of categories, e.g., category 2 is not
interpreted as twice as severe as category 1, only more severe.
The following model is assumed:
P(s>i|X) = l/(l+exp[-0i - X*/J])
where:
s = severity
i = severity category 1, 2,...,L-1
L = number of severity levels
X = an independent variable such as log-dose (can be a vector representing
more than one independent variable)
0 = parameter estimate associated with an independent variable (can be a
vector representing more than one parameter estimate)
oij = parameter estimate associated with response category i.
The logistic dose-risk model is then L equations that are coupled by the common
slope parameter (vector of parameters), p. Note that all other probabilities are linearly
related to the above L-l probabilities. That is,
P(s=l) =
P(s=i) =
P(s=L) =
Some difficulties arise because of the data that are available for analysis. Almost
all of these data are from animal bioassays, so the risk estimates that are produced by
the regression analyses are animal risk estimates that must be extrapolated to produce
14
-------
human risk estimates. Also, many times the data available represent information from an
entire dose group, not from an individual animal. The interpretation of the risk
estimates depends, then, upon the unit of input: animal versus human data and
individual versus dose group information.
Guth et al. (1991) applied this approach to data for acute inhalation exposure for
methyl isocyanate, tetrachloroethylene, dichloromethane, and propylene oxide. Data sets
were developed and analyzed using categorical regression using information at the
experimental group level. In addition to analysis of the complete: data sets for acute
exposure (all available studies), experimental data for specific endpoints, exposure
duration ranges, and species (human or experimental animal) were analyzed and the
results were compared with the results obtained using concentration response modeling
(benchmark approach) or a LOAEL/NOAEL approach.
The methodology presented below continues development of the model proposed
by Hertzberg (1989) and Hertzberg and Knauf (1989) as previously discussed. The
model is expanded to allow for more extensive use of the experimental data (see section
3.2.3). Individual animal data for all of the dose groups can be considered in the analysis
to develop maximum likelihood estimates, (MLEs) and statistical lower bounds for the
prediction of various levels of severity for a range of time and dose values.
3.2 Application of a New Method for Assessing Risk from Short-Term Exposure
to Hydrogen Chloride
This approach to developing quantitative estimates of risk incorporated several
elements. The first was to conduct a thorough literature search for all the available
acute inhalation toxicity data on hydrogen chloride. Then, a critical review of the studies
15
-------
was made to select relevant experimental data and to identify the types of adverse health
effects and the exposures (concentration-duration) at which these effects were observed
(discussed in Appendix A). Next, the adverse effects were assigned a severity index
ranking using a rank order scale (section 3.2.1). Finally, animal exposure concentrations,
described in section 3.2.2 and Appendix A, were converted to human equivalent
concentrations based on the methodology developed for determining inhalation RfCs
(EPA 1990). Once input data (including size of dose group, dose, duration, and severity
of effect and number of animals with that effect in each dose group) were developed
from the relevant studies, these data were analyzed by the proposed model (section
3.2.3). The results of this analysis are discussed in section 3.2.3 and model output values
given in Appendix D. Conclusions and recommendations are given in section 3.3.
3.2.1 Categorizing by Severity Index
The experimental data indicated that the response to hydrogen chloride exposure
is dependent on both the concentration and the duration of exposure. As either the time
or the concentration increased, either the number of animals exhibiting a specified effect
or the severity of that effect increased, or both. In studies in which alterations in
respiratory function or respiratory histopathology were evaluated, the results were usually
reported as a continuum of effects of increasing severity rather than as an increasing
number of animals with any one discrete effect. Usually in lethality studies only the
mortality incidence was reported without discussion of the presence of nonlethal effects.
Given the variety of the level of detail and the quality of the data reported,
ranking by use of a complex severity index, such as that proposed by Hartung (1986), was
not feasible. For the purpose of this assessment, a simplified severity scale (given in
16
-------
Table 3) was utilized similar to those developed by Dourson et al. (1985) and Hertzberg
and Miller (1985) (see Table 4). Table 3 lists a rank order scale in that the difference
between severity categories is not an arithmetic assignment, i.e., an index of 2 is not twice
as severe as an index of 1.
The four response levels listed in Table 3 are ranked in order of increasing
severity of the toxic effect. An index of 1 is equivalent to a NOAEL at which there are
no statistically or biologically significant effects observed. Exposure levels which
produced minor toxic effects (biologically significant adverse effects) are assigned an
index of 2. An index of 3 is assigned to exposure levels which produce moderate (more
severe than index of 2) adverse effects which may be considered biologically significant.
When severe adverse effects, such as mortality or irreversible tissue damage, were
observed, the exposure level was given an index of 4. For example, in a bioassay of
rodents exposed to hydrogen chloride, no effect or a non-adverse effect (such as
reversible decrease in respiratory rate) would be assigned a severity index of 1, while an
effect such as slight damage (e.g., ulcerations) to the mucosa of the upper respiratory
tract with no other tissues affected would receive a severity index of 2. More extensive
damage to the mucosa (damage to the squamous epithelium) including damage to the
submucosa would receive a severity index of 3; more serious effects, such as further
damage to underlying support structures and/or death, would receive a severity index
of 4.
Each animal within a dose group was assigned a severity index based, when
possible, on the degree of injury reported by the investigator, as illustrated in Table 5.
For each dose group within a study, the close, duration of exposure, the highest severity
17
-------
Table 3
Preliminary Ranking System for Evaluation
of the Acute Inhalation Toxicity in Animals Exposed to Hydrogen Chloride
Severity
categories
Toxic
effects
Analogous EPA effect levels
1
2
3
4
None No observed adverse effect
Minor Mild adverse effect
Moderate Moderate (to severe) adverse effect
Severe Frank effect
18
-------
Table 4
Response Levels (Ranked in Order of
Increasing Severity of Toxic Effect) Considered in Deriving
Inhalation Reference Concentrations (RfCs) for Systemic Toxicants
NOEL: No-Observed-Effect Level. That exposure level at which there are no
statistically or biologically significant increases in frequency or severity of
effects between the exposed population and its appropriate control.
NOAEL: No-Observed-Adverse-Effect Level. That exposure level at which there are
no statistically or biologically significant increases in frequency or severity of
adverse effects3 between the exposed population and its appropriate control.
Effects are produced at this level, but they are not considered to be adverse.
LOAEL: Lowest-Observed-Adverse-Effect Level. The lowest exposure level in a study
or group of studies that produces biologically significant increases in
frequency or severity of minor adverse effects between the exposed
population and its appropriate control.
AEL: Adverse-Effect Level. The exposure level in a study or group of studies that
produces biologically significant increases in frequency or severity of moderate
adverse effects between the exposed population and its appropriate control.
FEL: Frank-Effect Level5. That exposure level which produces frankly apparent
and unmistakable adverse effects, such as irreversible functional impairment
or mortality, at a statistically or biologically significant increase in frequency
or severity between an exposed population and its appropriate control.
a Adverse effects are defined as any effects resulting in functional impairment and/or
pathological lesions that may affect the performance of the whole organism, or that
reduce an organism's ability to respond to an additional challenge.
b Frank effects are defined as overt or gross adverse effects (severe convulsions,
lethality, etc.).
SOURCE: EPA (1989).
19
-------
Table 5
Reading Data Notation
Number of Animals
Dose Group Severity Index in Each Dose Group Indicator3
1 b,c
2b,d
3e
4b,f
4
4
4
4
10
10
10
10
0
6
10
-1
a Indicator is defined as the number of animals in that dose group with an effect at that
severity level, with the exception of an indicator of -1 (see footnote e).
b In each case when some of the animals are assumed to be of a severity of less than
severity i, which in this case is 4, the model estimates the probability that these animals
are at a severity of less than i, in this case the probability of a severity of 3 or less.
c Indicator is zero. Zero animals had an index of 4 and all ten animals were assumed to
have an index of less than 4.
d Indicator is positive but less than number of animals in the dose group. Six animals
had an index of 4 and the other four animals were assumed to have an index less
than 4.
e Number of animals in the dose group and indicator number are equal. All ten animals
have an index of 4.
f Indicator is -1. The study indicates that at least one animal had an index of 4 but
provides no information on other animals. The other animals were assumed to have a
severity of less than 4.
20
-------
index assigned to that dose group, and the number of animals with an effect of that
severity are entered into the model.
For studies in which dose-response information was adequately reported,
assignment of a severity index was straightforward. In other cases, assignment of a
severity index was complicated by the limitations of reported data. For example (see
Table 5), in studies for which the incidence of lethality was the only result reported
(Darmer et al. 1975, Hartzell et al. 1985, Kaplan et al. 1984, Machle et al. 1942,
Wohlslagel et al. 1976), all dose groups with one or more deaths were assigned a severity
index of 4 with the indicator reporting the actual number of animals that died (see Table
5). However, in a lethality study, when no deaths were reported for one of the dose
groups and no other effects were reported, a severity index of 4 was assigned, but the
number of animals at that severity was designated as 0. Operationally, the model actually
estimates the probability that animals within the dose group were at a severity of less
than 4 (severity index 3 or less).
Only two lethality studies (Burleigh-Flayer et al. 1985, Kaplan 1987) included
control groups. However, no effects were observed in these groups and they were not
included in the analysis. In all of the other studies, due to the nature of the effects
reported, it was assumed that any adverse effects reported were due to hydrogen chloride
exposure. It should be noted that the model used in this analysis can account for a
background rate of an effect when one can be identified.
Toxicity studies were conducted in mice, rats, guinea pigs, and baboons (discussed
further in Appendix A). The effects of hydrogen chloride on non-human primates were
examined in a series of recent studies by Kaplan and his colleagues (Anzueto et al. 1987;
21
-------
Kaplan 1987; Kaplan et al 1984, 1988). In one experiment, seven baboons were exposed
to hydrogen chloride gas for a single 5 minute period to concentrations ranging from 190
to 17,290 ppm (285 to 25,935 mg/m3) (one baboon/exposure concentration); no
concurrent control was included. "Severe irritant effects" were observed at all
concentrations except at the lowest concentration, 190 ppm (285 mg/m3), and all of the
animals survived exposure concentrations up to 11,400 ppm (17,100 mg/m3). Animals
exposed to concentrations of 16,570 and 17,290 ppm (24,855 and 25,935 mg/m3) died
from bacterial pneumonia 18 and 76 days post-exposure, respectively. From this
information, severity categories were assigned as follows: based on the fact that at
190 ppm (285 mg/m3) no irritant effects were observed during exposure or during the
post-exposure period, an index of 1 was ascribed; exposure to 810 to 2780 ppm (1215 to
4170 mg/m3) reportedly caused irritation (coughing and salivation), but no post-exposure
symptoms were observed; therefore, a severity index of 2 was assigned; exposure to
11,400 ppm (17,100 mg/m3) caused immediate agitation and salivation with coughing,
blinking of the eyes, and "raspy breathing" post-exposure when the animal was excited
(assigned a severity index of 3); and effects from the two highest exposures were
classified as indices of 4 because of lethality.
In a separate experiment with anesthetized baboons (Kaplan 1987, Kaplan et al
1988), exposure to hydrogen chloride at nominal concentrations of 500, 5000, and 10,000
ppm (750, 7500 and 15,000 mg/m3) for 15 minutes was evaluated for effect on respiratory
response and pulmonary function. Nine baboons (three per dose) were exposed to
hydrogen chloride, while three controls were exposed to air only. There was a
concentration-related increase in respiratory rate and minute volume at the two higher
concentrations during exposure. Observation of these animals for up to one year post-
22
-------
exposure (Anzueto et al. 1987) revealed no long-term effects in animals exposed to 500
and 5000 ppm (750 and 7500 mg/m3); however, chronic nasal obstruction and mouth
breathing, persistent hypoxemia, increased lung functional capacity, and focal lung fibrosis
in animals exposed to the high concentration were observed. At 500 ppm (750 mg/m3),
no statistically significant alterations in respiratory response or pulmonary function were
observed in the animals, although respiratory rate and minute volume increased slightly
and it was thought that these respiratory effects were biologically significant. For this
effect an index of 2 was assigned. Significant acute responses, including increased
respiratory rate and minute volume, were reported following expiosure to 5000 ppm
(7,500 mg/m3) and were assigned a severity index of 2 because all parameters had
returned to normal by 3 days after exposure. An index of 3 was assigned to the
10,000 ppm (15,000 mg/m3) exposed group for the moderately severe effects observed
(including focal lung fibrosis) in one of the three animals one year after exposure
(Anzueto et al 1987).
The effect of hydrogen chloride inhalation exposure in mice has been reported in
three studies performed by the same research group using similar concentration ranges
and the same exposure intervals, protocols, and laboratory animal species. In the study
by Lucia et al. (1977), mice were exposed to levels of hydrogen chloride ranging from 17
to 7,279 ppm (25 to 10,918 mg/m ) for 10 minutes (nine dose groups of four animals
each). Twenty-four hours after exposure the mice were sacrificed and the heads were
prepared for histopathological examination by sectioning each head into four transverse
sections. A semi-quantitative scale (0 to 4+) of upper respiratory tract damage in mice
was devised by the authors to evaluate the extent of histological insult to the respiratory
tract mucosa. Exposure to hydrogen chloride at 17 ppm (25 mgfai3) for 10 minutes
23
-------
resulted in superficial ulceration of upper respiratory tract mucosa (observed in only one
of four sections microscopically examined) (Lucia et al. 1977), which was considered in
this analysis to be non-adverse (severity index 1). The degree of mucosa and underlying
submucosa/structural damage increased directly with increasing dose. This is one of the
few studies in which all four levels of severity were observed (see Table B-l). The
effects and accompanying severity indices were:
No damage seen at 24 hours - severity index 1
1 to 75% of the mucosa destroyed and some submucosal damage from 131 to
723 ppm (197 to 1,085 mg/m3) - severity index 2
25 to 75% of the mucosa destroyed and some submucosal damage and
damage to unde
severity index 3
damage to underlying support structures at 1,088 ppm (1,632 mg/m3) -
50 to 100% of the mucosa destroyed and extensive destruction of all
underlying tissues at 1,473 to 7,279 ppm (2,960 to 10,919 mg/m3) - severity
index 4. [Note: High-dose animals 7,279 ppm (10,919 mg/m3) also suffered
total destruction of the eyes]
In one study by Barrow et al. (1979), mice exposed to 127 ppm hydrogen chloride
(190 mg/m3) for 10 minutes sustained ulceration of nasal epithelium located in the
inferior of the nasal mucosa. Since the extent of ulceration was not clearly reported, it
was assumed that the animals experienced similar insult as reported by Lucia et al. in
mice exposed to a concentration of 131 ppm (197 mg/m3) for 10 minutes (damage of
upper respiratory tract mucosa in three of four sections examined). Therefore, this dose
group was assigned an index of 2. The next two dose groups [478 and 710 ppm (730 and
1,060 mg/m )] had increasing damage to the nasal epithelium and underlying structures
and were assigned a severity index of 3. All other dose groups in Barrow et al. (1979)
exposed to 1,072 to 19,333 ppm (1,600 to 20,000 mg/m3) had extensive damage to the
24
-------
f
mucosa/submucosa and underlying structures or extensive cornea! damage or death and
were given severity indices of 4.
Barrow et aL (1977) reported a 12% decrease in respiratory rate in mice exposed
to 40 ppm (364 mg/m3) hydrogen chloride for 10 minutes; however, the animals showed
complete recovery in 6 minutes following exposure, indicating a severity level of 1.
Exposure to 100 to 440 ppm (149 to 660 mg/m3) hydrogen chloride resulted in up to a
52% decrease in respiratory rate and was assigned an index of 2. The highest dose group
with reported mortality was given an index of 4.
In the study by Burleigh-Flayer et aL (1985) guinea pigs were exposed to hydrogen
chloride for 30 minutes. Pulmonary irritation and significant acute respiratory parameter
changes were noted, most of which were reversible within 15 days post-exposure. The
lower concentration of 320 ppm (480 mg/m3) was assigned an index of 2 based on the
significant reduction in respiratory frequency achieved during a CO2 challenge 30 minutes
following exposure. One of four animals exposed to 680 ppm (1,020 mg/m3) exhibited
corneal damage; this effect was assigned a severity index of 3. Mortality was reported
among animals exposed to 1,040 and 1,380 ppm (1,560 and 2,070 mg/m3) hydrogen
chloride. As a result, an index of 4 was given to both of these dose groups.
Data extracted from animal bioassays along with severity index assignments and
human equivalent exposure levels may be found in Appendix B.
3.2.2 Human Equivalent Concentration Estimation
Animal exposure concentrations reported in the studies described in the previous
section were converted to a human equivalent dose according to the methods used to
develop inhalation RfCs (EPA 1989). In this study, human breathing rates (BRH) and
25
-------
surface area of the respiratory region affected (SA) were used to convert experimental
concentrations to an equivalent human exposure in air. The human equivalent
concentration (mg/m3)H was calculated by:
(mg/m3)A x [BRArm3/davHSAArcm2>l] = (mg/m3)H
[BRH(m3/day)-HSAH(cm2)]
If:
BRA = animal breathing rate
SAA = animal surface area of the respiratory region affected
BRH = human breathing rate of 20 m /day,
SAH = human surface area of the respiratory region affected
For example, a concentration of 4,800 mg/m3 in a mouse (Darmer et al. 1974) would be
equivalent to a human concentration (based on total lung surface area) of 33,060 mg/m3:
4,800 mg/m3 x 0.064 m3/dav H- 297.7 cm2 = 33,060 mg/m3
20 m3/day + 640758 cm2
Study-specific parameters for breathing rates and surface areas were used whenever
possible. Otherwise, standard values from RfC methodology were used. For the
purposes of this analysis, the human equivalent concentration (HEC) was estimated
based on the assumption that the total lung was the target (HEC-Total). For comparison
purposes, the model was also fit to the data using the unadjusted animal concentration
and a human equivalent concentration based on the assumption that the extrathoracic
region (for two of the animal data sets) as well as the total lung (for all the remaining
data sets) were targets. The experimental data upon which the dose conversions were
based are discussed in section 3.2.1 and in Appendix A.
26
-------
an
Since the longest duration to be considered was one day, no adjustment of time
was made to account for the difference in lifespan between animals and humans. That
is, for short exposures, it was assumed there is no interspecies time difference. Also, in
this investigation, no adjustment for duration of exposure was incorporated into the
human equivalent concentration at this time since the model takes time into
consideration.
3.2-3 Description of the Dose-Duration-Response Model
Two dose-duration-response models were considered for this investigation. Both
models estimate the probability function F(s>i|d,T), which is the probability (P) of
effect of severity (s) greater than the index i at a fixed dose (d) and duration (T).
The two models considered were the Weibull,
P(s > i) = 1 - exp[-qi(d-d0)wTk],
where,
s = severity
q{ = parameter estimate associated with response category i
i = severity index (i=l,2,3...,L-l, with L the number of severity
categories)
d = dose
d0 = threshold dose
T = time
w and k constrained to be > zero,
and the log-logistic,
P(s > i) = 1 /[I + exp[- q{ - W loglo(d-d0)-K log10T]],
27
-------
where,
severity
q{ = parameter estimate associated with response category i
i = severity index (i=l,2,3...,L-l, with L the number of severity
categories)
d = dose
d0 = threshold dose
T = time
W and K are unconstrained
The threshold dose (d0) below which no effect can be attributed to exposure to
the chemical is assumed to be a known constant. For this analysis, threshold dose (d0)
below which no effect can be attributed to hydrogen chloride was assumed to be zero.
This assumption should certainly be conservative for exposures up to one day.
While data should be found and analyzed that provide information on individual
animal effects instead of dose group effects, the data presently available could be used
on an individual animal basis if the proper terms were added to the likelihood as
discussed in Appendk C. For example, at a certain dose, if four out of ten animals died,
then the dose group method would put all ten animals at the highest severity. The
individual animal method discussed would put four at the highest severity and six with
severity less than the highest severity.
Since many of the bioassays available did not report the results for each individual
animal, likelihoods, which convert the information available to a per animal per dose
group basis, have to be considered in this model as follows. For N animals at dose d and
time T with index i, the contribution to the log-likelihood is N In [P^d/T)], where
28
-------
P^d/T) = l-P(s > 1 |d,T)
Pj(d,T) = P(s > i-1 | d,T) - P(s > i | d,T) i = 2,3
P4(d,T) = P(s > 3 | d,T).
Suppose for N animals with fixed d and T, X (number) had an effect of severity
index i, and all that is known about the other animals is that their index is less than i.
For example, at fixed dose and time, three (X=3) out of ten animals (N= 10) died. The
contribution to the log-likelihood is then,
X In [Pj(d,T)] + (N-X) In [l-P(s > i-11 d,T)].
That is, three animals would be assigned a severity index of 4, and six animals would be
assigned, by the likelihood method, severity indices of 3 or less. A special case of this is
when X is zero. For example, the only information is that no animals died. In this case
all animals would be assigned severity indices of 3 or less. This may not be a
conservative assumption in that some animals may have had effects of severity 4;
however, when these data were not available the stated assumption was made. Further
examples of input data and their meaning are found in Table 5.
If for fixed d and T only the maximal effect (index) is given, then at least one
animal has index i and the remainder have index i or less. The contribution to the log-
likelihood is then
In {[1 - P(s > i|d,T)]N - [1 - P(s >(i-l)|d,T)]N}.
There are other possible situations, but these are the only ones necessary to utilize all of
the hydrogen chloride bioassay data on an individual animal basis.
29
-------
Bands may be established where a particular effect would be expected for a given
time and dose. Let a (say 0.01) be given once the model is fitted to the data; curves
generated from the Weibull and log-logistic model would yield lines for which the
probability of a severity index greater than i would be a, Le., [P(s>i | d,T)=a].
For fixed time T and a (e.g., a=0.01), a lower bound on the dose, such that
P(s>i, | d,T)=a, may be found using the likelihood ratio method reported in Crump and
Howe (1985). For fixed time, this is the benchmark approach suggested by Crump
(1984). If the benchmark lower bound is found for various times and connected to form
a line, a benchmark lower bound band can be drawn. (See Appendix C for the
mathematical description of the approach described above.)
3.2.4 Results
While both models have about the same predicted values, the log-logistic model
values are reported in this investigation since this model offers faster convergence of the
optimization routine.
Using the data described above and the log-logistic dose-duration model, the MLE
and the 95% lower bound (confidence limit) on the concentration of hydrogen chloride
are predicted for probability less than 0.05 or 0.01 of observing an effect of a severity
greater than that represented by index i, for various exposure times, T. Expressed
mathematically, estimates of the MLE and 95% lower bound on the MLE for P(s>i | d,T)
= 0.05 or 0.01 were determined. Graphs (Figures 2 to 13) were generated indicating the
MLE and the 95% lower bound (confidence limit) on the MLE. Figures 2 to 5 were
derived using human equivalent concentrations of hydrogen chloride assuming that the
total lung surface area was the target, while Figures 6 to 9 represent the same data
30
-------
100000 -J
10000 - =
1000 i
100
10
Figure 2
Animal to Human Extrapolation Based on
Total Lung Surface Area
01
001
i ii i ~ ~r r ~r r~i
0 1
1
Time (hours)
i i - - i i —r i i
I
10
Maximum Likelihood Estimate of P(S>i | d,T)=0.05
24
-------
Figure 3
Animal to Human Extrapolation Based on
Total Lung Surface Area
100000 _
N>
10000 :
n 1000 .
£ 100
10
01 ,
001
o i
i IT —r~
(=1
1
Time (hours)
1=2
i i
1=3
r T i i r
10
24
95.0% Lower Bound on Maximum Likelihood Estimate of P(S>i | d,T)=0.05
-------
Figure 4
Animal to Human Extrapolation Based on
Total Lung Surface Area
100000
10000
1000
I 100 -=
10
i -
01
001
01
1 1 1 1 1
i-1
-l-T-j
1
Tim« (hours)
i-2
i i i
1-3
10
Maximum Likelihood Estimate of P(S>i | d,T)=0 01
24
-------
100000 .
10000 :
Jo 1000 •_
0,
100 -
10
Figure 5
Animal to Human Extrapolation Based on
Total Lung Surface Area
0 1
001
0 1
r i i
1=1
i r i —r—T~ t
1
Tim.i jh>.i | d,T)=0.01
24
-------
Figure b
10000
100
oot
Animal to Human Extrapolation Based on
Tolal Lung and Extrathoracic Surface Area
0 i I I i I I
0 1
1 1 \-\~T ~T T~
1
Time (hours)
10
24
i-2
Maximum Likelihood Estimate of P(S>i | d,T)=0.05
-------
Figure 7
Animal to Human Extrapolation Based on
Total Lung and Extrathoracic Surface Area
10000
too
i ,
001
i i i "~i r~~r~!~~r
0 I
1
i
Time (hours)
(•=2
i i
-r-i , i |
10
24
95.0% Lower Bound on Maximum Likelihood Estimate of P(S>i | d,T)=0.05
-------
10000
100
I ,
001
Figure 8
Animal to Human Extrapolation Based on
Total Lung and Extrathoracic Surface Area
01
1—I—i—r
l-t
1
Time (hours)
1-2
10
1=3
Maximum Likelihood Estimate of P(S>i | d,T)=0.01
" ~\
24
-------
Figure 3
u>
00
10000
100
co
I
i
001
Animal to Human Extrapolation Based on
Total Lung and Extrathoracic Surface Area
i iii —r -~r ~i~\
0 1
T
i
Time (hours)
i-2
1-3
10
24
95.0% Lower Bound on Maximum Likelihood Estimate of P(S>i | d,T)=0.01
-------
Figure 10
10000
1000 -
P)
E 100
O)
10
i
5 «
0 1
001
Animal to Human Extrapolation Based on
Administered Animal Doses
01
_., T
Time (hcuis)
1=2
I I " I T— 1 1 I
1=3
I
10
24
Maximum Likelihood Estimate of P(S>i | d,T)=0.05
-------
Figure 11
10000 -_
1000
f 100
Q>
10
Animal to Human Extrapolation Based on
Administered Animal Doses
001 i i i i i
I I III —I—I
0 1
1
Time (hours)
1=1
1=2
i i
1-3
10
24
95.0% Lower Bound on Maximum Likelihood Estimate of P(S>i | d,T)=0.05
-------
Figure 12
10000
1000
P)
E
100
to
0 1 ,
001
Animal to Human Extrapolation Based on
Administered Animal Doses
I T I -| I
01
1
Time (hours)
r T r
10
24
Maximum Likelihood Estimate of P(S>i | d,T)=0.01
-------
Figure 13
i
10000
1000
too
10
Animal to Human Extrapolation Based on
Administered Animal Doses
0 1
001 i i I i i ]
0 1
i i 11 —i—r"~r~r
1
Time (hours)
1=2
i i i
1=3
24
95.0% Lower Bound on Maximum Likelihood Estimate of P(S>i | d,T)=0.01
-------
assuming that the human equivalent concentration was based on the total lung surface
area except for the studies by Lucia et al (1977) and Barrow et al (1979) for which the
target was assumed to be extrathoracic. Human equivalent concentrations were derived
based on the methodology used in the development of inhalation RfCs (EPA 1989). For
comparison, additional modeling was done using as input the actual concentrations used
in the animal experiments (Figures 10 to 13). In addition, the MLE and 95% lower
bound hydrogen chloride concentrations for P(s>i|d,T) = 0.05 or 0.01, that is, the
probability of an effect of severity corresponding to an index of 1 for selected time
periods, are given in Tables 6 and 7. The input data upon which the modeling was based
are given in Appendix B and the model output is given in Appendix D.
The results discussed in this section were derived by using data directly, or
interpolating, from tables in Appendix D, and by comparison to the appropriate figures.
Model estimates derived from human equivalent concentrations based on the total lung
as the target tissue (Table D-l in Appendix D) are given in Figures 2 to 5. Based on the
model output, estimation or prediction of the probability (at the 0.05 or the 0.01 level) of
an adverse human health effect of various severity levels for any combination of time and
dose can be made. For example, Figures 2 and 3 indicate the MLE and lower bounds
for P(s>i | d,T)=0.05. Using these figures, the MLE and 95% lower bound for the
probability of an effect of severity represented by index 1 was estimated to be
approximately 1 ppm (2 mg/m3) and 0.5 ppm (0.9 mg/m3), respectively, following an
8-hour exposure [P(s>i|d,8h)=0.05] (Table 6). For a 10-minute exposure
[P(s>i|d,10min)=0.05] the MLE was 55 ppm (85 mg/m3), while the lower bound was
21 ppm (31 mg/m3) (Table 6). Figures 4 and 5 indicate that for [P(s>i| d,T)=0.01], the
MLE was 0.6 ppm (0.9 mg/m3) for an 8-hour exposure and 20 ppm (30 mg/m3) for a
43
-------
Table 6
Maximum Likelihood Estimates and 95% Lower Bounds
for P(s>l | d,T) = 0.05 with Exposure to Hydrogen
Chloride for Various Durations of Exposure
Human Equivalent Concentrations ppm (mg/m3)
Total Lung3
Time
10 mind
4hrs
8hrs
24hrs
MLEe
55
3
1
0.6
(82)
(5)
(2)
(0.9)
LBe
21
1
0.5
0.2
(3D
(2)
(0.9)
(0-3)
Extrathoracicb
MLE
0.3
0.01
(2)
(0.1)
0.007 (6)
3
(2)
LB
13 (0.4)
0.6 (0.02)
0.5 (0.01)
0.2 (4)
Animal0 pp
(mg/m3)
MLE
13
(19)
0.6 (1)
0.5
0.2
(0.8)
(0.3)
m
LB
6
0.4
0.2
0.07
(9)
(0.6)
(0.3)
(0.1)
a Estimates of human equivalent concentrations were based on the assumption that the total
lung was the target.
b Estimates of human equivalent concentrations were based on the assumption thai the total
lung was the target in all studies except that for Lucia et al. (1977) and Barrows
et al. (1979) the extrathoracic area was assumed to be the target.
c No conversion made; concentrations are those used in animal experiments.
d Times are approximate, interpolated from values in Tables D-l to D-3.
e Maximum Likelihood Estimate (MLE) and 95% Lower Bound (LB).
44
-------
10-minute exposure (Table 7). The lower bound on the MLE was 0.1 ppm (0.2 mg/m3)
for an 8-hour exposure and 6 ppm (9 mg/m3) for a 10-minute exposure (Table 7).
Exposure estimates developed from the proposed model were compared to
criteria values and reported data. These estimates are presented in Tables 8 and 9.
While data in the tables are given for both a probability of effect of 0.01 (Table 8) and
0.05 (Table 9), the remainder of this discussion will focus on a probability of 0.01, i.e.,
[P(s>i | d,T)=0.01]. Concentration estimates obtained in this analysis approximate
severity effects within the range of acute criteria values and bracket human data
reported. As an example of reported data, according to Henderson and Haggard (1943),
10 to 50 ppm (15 to 75 mg/m3) was estimated to be the maximum concentration
tolerable for a few hours. An analogous comparison would be to compare model
predictions at 3 hours for a severity index 2. As listed in Table 8 under Total Lung
(derived by using Figures 4 and 5 and interpolating between times as given in Table
D-l), for P(s>2|d,3h)=0.01, the MLE was approximately 56 pprn (84 mg/m3), while the
95% lower bound on the MLE was 16 ppm (24 mg/m3) for 3 hours exposure. The
maximum concentration tolerable for a 1-hour exposure to hydrogen chloride was
reported to range from 50 to 100 ppm (75 to 150 mg/m3) (Henderson and Haggard
1943). The estimated MLE and 95% lov/er bound on P(s>2| d,lh)=0.01 (Figures 4 and
5) were 153 and 43 ppm (229 and 64 mg/m3), respectively. Criteria for acute non-
lethal effects have been established for hydrogen chloride. These values are set for
various exposure time periods and are intended for differing regulatory criteria.
Exposure estimates developed from the proposed model based on total lung were
determined and compared to the criteria values reported. This comparison considered
both the intended time frame and the intended degree of protection for a specified
45
-------
Table 7
Maximum Likelihood Estimates and 95% Lower Bounds
for P(s>l | d,T) = 0.01 with Exposure to Hydrogen
Chloride for Various Durations of Exposure
Time
Human Equivalent Concentrations ppm (mg/m )
Total Lung3
Extrathoracic5
MLEe
LBe
MLE
LB
Animalc ppm (mg/m3)
MLE
LB
10 mind 20 (30) 6 (9)
4 hrs 1 (2) 0.3 (0.5)
8 hrs 0.6 (0.9) 0.1 (0.2)
24 hrs 0.2 (0.3) 0.05 (0.8)
0.1 (0.2) 0.05 (0.07)
0.007 (0.01) 0.001 (0.002)
0.004 (0.006) 0.0005 (0.0008)
0.001 (0.002) 0.0002 (0.0003)
5 (7) 2 (3)
0.3 (0.5) 0.1 (0.2)
0.2 (0.3) 0.07 (0.1)
0.07 (0.1) 0.03 (0.04)
a Estimates of human equivalent concentrations were based on the assumption that the
total lung was the target.
b Estimates of human equivalent concentrations were based on the assumption that the
total lung was the target in all studies, except that for Lucia et al. (1977) and Barrows
et al. (1979) the extrathoracic area was assumed to be the target.
c No conversion made; concentrations are those used in animal experiments.
d Times are approximate, interpolated from values in Tables D-l to D-3.
e Maximum Likelihood Estimate (MLE) and 95% Lower Bound (LB).
46
-------
Tables
Relative Comparison of Several Regulatory
Criteria and Human Data with Maximum Likelihood
and 95% Lower Bounds on P(s>i |d,T)=O.Or
Human Equivalent Concentrations ppm
IDLH
PEL
RfC
Human 1"
Human-2r
Human-3s
Criteria
Value ppm (mg/m3)
100(150)
5(7)
0.005 (0.007)
!Q 50 (15-75)
50-100 (75-150)
1000-2000 (1500-3000)
Reference
Time
(Hours)
0.5
0.25
24
3
1
0.17
Reference
Severity
Index
2
1
1
2
2
3
Total
MLE
281 (421)
13 (20)
0.2 (0.3)
56(84)
153 (229)
2893 (4339)
Luncb
LB
77(115)
2(3)
0.05 (0.08)
16 (24)
43(64)
1104(1656)
(mg/m3)
Extrathoracic'
MLE LB
5(8)
0.07 (0.1)
0.001 (0.002)
1.3 (2)
3(5)
107 (161)
1.3 (2)
0.01 (0.02)
0.0002 (0.0003)
0.3 (0.4)
0.7 (1)
35 (52)
Animal*1 ppm fmu/m3)
MLE LU
63 (94)
3(5)
0.07 (0.1)
15 (22)
36 (54)
437 (655)
35 (53)
1(2)
0.03 (0.04)
8(12)
20 (30)
279 (418)
* All values are based on the probability of a severity greater than 1 at the 0.01 level and are extrapolated from data given in Tables D-1, D-2, and D-3.
b Estimates of human equivalent concentrations were based on the assumption that the total lung was the target.
c Estimates of human equivalent concentrations were based on the assumption that the total lung was the target in all studies, except that for Lucia et al. (1977) and Barrows
et aj. (1979) the extrathoracic area was assumed to be the target.
d No conversion made; concentrations are those used in animal experiments.
c Maximum concentration tolerable for a few (3) hours (Henderson and Haggard 1943).
' Maximum concentration tolerable for one hour (Henderson and Haggard 1943).
* Dangerous for even short exposures.
-------
Table 9
Relative Comparison of Several Regulatory
Criteria and Human Data with Maximum Likelihood
and 95% Lower Bounds on P(s>i |d,T)=0.05*
Human Equivalent Concentrations ppm (mg/m3)
Criteria
Total Lung"
Extrathoracicc
Animal** ppm (mg/m3)
IDLII
PEL
RfC
Human-lc
Human-2f
IIuman-31
Value ppm (mg/m3)
100(150)
5(7)
0.005 (0.007)
10-50(15-75)
50-100 (75-150)
1000-2000 (1500-3000)
Reference
Time
(Hours)
0.5
0.25
24
3
1
0.17
Reference
Severity
Index
2
1
1
2
2
3
MLE
785 (1177)
37 (55)
0.6 (0.9)
157 (236)
427 (640)
8076(12114)
LB
287 (431)
14(21)
0.2 (0.3)
66(99)
166 (249)
4277 (6416)
MLE
48 (72)
0.7(1)
0.01 (0.02)
10(15)
26 (39)
943 (1414)
LB
18 (27)
0.2 (0.3)
0.003 (0.004)
3(5)
10 (15)
452 (678)
MLE
185 (278)
9(13)
0.2 (0.3)
43(64)
106 (159)
1285 (1927)
LB
117(176)
4(6)
0.07 (0.1)
25 (38)
66(99)
954(1431)
* All values are based on the probability of a severity greater than 1 at the 0.05 level and are extrapolated from data given in Tables D-l, D-2, and D-3.
b Estimates of human equivalent concentrations were based on the assumption that the total lung was the target.
c Estimates of human equivalent concentrations were based on the assumption that the total lung was the target in all studies, except that for Lucia et al. (1977) and Barrows
et al. (1979) the extrathoracic area was assumed to be the target.
d No conversion made; concentrations are those used in animal experimenis.
" Maximum concentration tolerable for a few (3) hours (Henderson and Haggard 1943).
' Maximum concentration tolerable for one hour (Henderson and Haggard 1943).
8 Dangerous for even short exposures.
-------
criteria value and compared those to analogous time and severity levels estimated from
the proposed model. For example, the IDLH is an exposure concentration regarded as
immediately dangerous to life and health but is the concentration; that in the event of
respirator failure could be tolerated for 30 minutes. The IDLH for hydrogen chloride
has been established at 100 ppm (150 mg/m3) (HSDB 1991). A similar comparison, for
example, would be to compare the model prediction at 30 minutes for a severity index
of 2. Again referring to Table 8, the MLE and 95% lower bound on P(s>2| d,0.5hr) =
0.01 are 281 and 77 ppm (421 and 115 mg/m3), respectively. In reality, the IDLH may
be analogous to some severity represented by an index between level 2 and 3, indicating
that estimates at level 2 may be lower than would be expected if the data could have
been divided into more severity categories. Therefore, it should be noted that the
severity indices used in this analysis did not equate exactly with effects associated with
IDLH.
The OSHA PEL of 5 ppm (~7 mg/m3) is the concentration that should not be
exceeded at any time during the work day. Concentrations below this value may be
considered to be analogous to a severity category 1; therefore, the PEL was compared to
concentration estimates at a severity index of 1 for 15 minutes (Table 8). Again focusing
on Total Lung, the MLE and 95% lower bound at P(s>l | d,0.25hr)=0.01 were 13 ppm
(20 mg/m3) and 2 ppm (3 mg/m3), respectively, which bracket the 5 ppm (—7 mg/m3)
standard. Conversely, if the dose is held constant and time is estimated from the model
for P(s>l 17 mg/m3,T)=0.01, and if the MLE estimate of concentration is 5 ppm
(~7 mg/m3), then the time corresponding to that dose was approximately 45 minutes
(Figure 4). If the 95% lower bound on that concentration was assumed to be 5 ppm
(~7 mg/m3), the corresponding time was approximately 13 minutes (Figure 5). That is,
49
-------
exposure durations of 13 minutes at 5 ppm (~7 mg/m3) would not be expected to result
in an adverse effect. Stated another way, the probability of exceeding a severity level
corresponding to an index of 1 is less than 0.01 when concentration is 5 ppm (~7 mg/m3)
and time is 13 minutes [P(s>l | 7mg/m3, 13min)=0.01].
The concentration values, as depicted in Figures 4 and 5, agreed with values
estimated using the RD50 method (Barrow et al 1977). The 1/100 RD50 and 1/10 RD50
concentrations reported by Barrow et al (1977) were 3.1 and 31 ppm (4.6 and 46
mg/m3), respectively. If it is assumed that at the lower concentration of 3.1 ppm
(4.6 mg/m3), no adverse effects would be expected to occur (Le., corresponding to a
severity index of 1), then according to this model for P(s>l 14.6 mg/m3,T)=0.01, the
MLE concentration of 3.1 ppm (4.6 mg/m3) corresponds to an exposure time of
approximately 1 and one-half hours. Similarly, the 95% lower bound concentration of 3.1
ppm (4.6 mg/m3) corresponds to approximately 20 minutes. If it is assumed that a
concentration of 31 ppm (46 mg/m3) may be accompanied by some sensory irritation,
then this concentration corresponds to an index of 2, Le., [P(s>2| 46 mg/m3,T)].
As another comparison, the RfC represents the concentration that is not expected
to result in adverse effects should chronic exposure occur at that concentration. The RfC
for hydrogen chloride, 7xlO"3 mg/m3 (5xlO~3 ppm), was based on the adjusted LOAEL in
a lifetime animal study (Sellakumar et al 1985), which was 2.5 mg/m3 (1.7 ppm). When
converted to a human equivalent concentration (6.5 mg/m3 or 5 ppm) and adjusted by
uncertainty factors totaling 1000, the RfC of 7xlO"3 mg/m3 (5xlO~3 ppm) was obtained.
An analogous time and severity index for comparative purposes would be 24 hours at an
index of 1. The MLE and 95% lower bound for P(s>l | d,24hr)=0.01 were 0.2 and 0.05
50
-------
ppm (0.31 and 0.08 mg/m3), respectively. These values are less than the human
equivalent concentration of 5 ppm (—7 mg/m3), but greater than the RfC. In fact, the
model estimates are only 19 to 76 times Jess than the human equivalent concentration
derived from the Sellakumar et al. data. When model estimates were derived from
human equivalent concentrations extrapolated based on total lung and extrathoracic
surface area (from Figures 8 and 9 and Table D-2), the MLE and 95% lower bound on
P(s>l |d,24h)=0.01 were l.lxlO'3 to IQxKT4 ppm (2xlO'3 to 3x10^ mg/m3) (Table 8).
When comparing model estimates based on Total Lung to those based on Total Lung
and Extrathoracic, the latter more closely approximate the RfC (the model estimates for
the MLE closely approximate the RfC, the 95% lower bound is an order of magnitude
smaller than the RfC). However, when comparing across all other criteria values, the
model estimates based on Total Lung provide a closer match, especially the 95% lower
bound values.
Although there are three curves (since these are log-log plots, the curves are seen
as straight lines) on each graph, the curve indicating no adverse effect is the lowest curve,
i = 1. A severity curve for severity index 4, i = 4, is not depicted, since the model solves
for maximum likelihood or lower bound on dose to exceed severity of i, and severity
exceeding index 4 (life-threatening effects or death) is undefined. What can be said is
that, should the hydrogen chloride air concentration exceed approximately 2.935 to
1,121 ppm (4,403 to 1,682 mg/m3) for as little as 10 minutes or 31 or 11 ppm (46 or
16 mg/m3) for 24 hours, irreversible, severe damage is likely to occur [for the MLE and
95% lower bound on P(s>31 d,T)=0.01] (Figures 4 and 5). This agreed with those data
reported by Henderson and Haggard (1943) who estimated that exposure to hydrogen
51
-------
chloride at concentrations of 1,000 to 2,000 ppm (1,500 to 3,000 mg/m3) for even short
periods of time was dangerous to human health.
33 Conclusions and Relevance to Human Health
There is limited information in the literature on the inhalation toricity of hydrogen
chloride in humans, and that which is available presents some conflicting information.
According to the EPA (1988), reported odor threshold levels range from 0.07 ppm
(0.1 mg/m3) to as high as 204 ppm (306 mg/m3). Hydrogen chloride is expected to be a
strong irritant, especially to mucous membranes lining the nasal passages and respiratory
tract. However, the ambient air concentration at which irritation may be first
experienced and the severity of that irritation over various concentrations and durations
is not well characterized.
There is an important difference between this method and methods used to derive
short-term standards. In existing methods a NOEL or a NOAEL from either animal or
human studies is divided by uncertainty factors that reflect the type of data and the
confidence in the data. Although numerous studies may have been reviewed, in the
NOAEL/uncertainty factor approach only one data point (one dose, from one study) is
actually used. In this proposed method, all data from all studies are subjected to
analysis, and the MLE of the dose for fixed time and probability, which may or may not
be any one of the actual experimental doses associated with an effect of specific severity,
is derived. Experimental doses are first converted to human equivalent doses. Then the
MLE, as estimated by the model, is the most likely dose, given all the evidence, that at a
specified time is not expected to be above a specified level of severity at a given level of
52
-------
statistical probability. The lower bound or confidence limit on the MLE is also estimated
by the model. (Lower bounds that may be considered include 90%, 95%, 97.5% and
99%.)
Applying the model to the hydrogen chloride data illustrates that the model is
robust enough to use a minimal data base; minimal in terms of number of studies
available and the lack of extensive data reported in some of the studies.. These
limitations in the data base, as well as any inconsistencies in results among studies, are
reflected in both the fit of the model to the data and the estimation of the lower
confidence limits on these data. The distance between the MLE and the 95% lower
confidence limit is reflective of the data limitations. For example, as illustrated in figures
for the 95% lower confidence limit, the curve representing a severity equal to or greater
than one is considerably lower than the curve for the corresponding MLE in each case.
This reflects both the few data points with a severity index of 1 and some of the overall
variability in the data.
Guth et al. (1991) showed that visually improved model fit was attained by
application of categorical regression to a specific toxicological endpoint for methyl
isocyanate and tetrachloroethylene. Modeling of data for specific endpoints or data only
from studies in humans for propylene oxide and dichloromethane resulted in poor model
fit due to data limitations imposed by selecting specific subsets of the data. Although
analysis of data disaggregated to the level of endpoint or species may clarify
interpretation of the results by increasing the similarity of the various data points used,
data requirements limit the extent of data reduction that is possible. Criteria for model
fit and minimum data base have not been established for this method.
53
-------
Since hydrogen chloride did not have data for all time periods and all severity
indices, i.e., data gaps, the model fills in these data gaps by the likelihood method. While
this is a useful procedure, it introduces a degree of uncertainty in the estimation of the
dose for fixed severity, probability, and time when there are data gaps. Other
uncertainties in this method are those inherent in the use of any model for dose-
response, such as the use of the appropriate model, the use of the correct explanatory
variables (i.e., sex, age, or species, etc. in addition to time and dose).
The use of other models that may be superior in their predictive abilities to the
log-logistic and Weibull models should be explored. Methods should also be developed
that will validate the assumptions implicit in the model whenever the regression is
performed. Some goodness-of-fit criteria should be developed which measure the
usefulness of the regression procedure relative to a particular data set. Due to the
partial likelihoods used, even graphs of the data and concordance statistics as presently
developed could not be used to get some idea of the fit. Since one would expect the
biological mechanism to change as the severity of the effect increased, a more general
model that allows the time and dose parameters to change over severity levels should be
considered.
Another limitation in the wide-spread application of this model is that it requires
critical evaluation of all the data and construction of a severity scale that is reflective of
the entire data base. While this process can be somewhat codified, since severity
schemes have been proposed and used in other applications, it still requires some
subjective judgement and, therefore, should be approached in a way that gains some
consensus among reviewers.
54
-------
Another uncertainty is in the interspecies extrapolations. For other methods,
extrapolation is accomplished by either expressing dose in units i:hat are assumed to be
equivalent across species, such as the surface area equivalence assumed in cancer
extrapolation, or by the application of uncertainty factors. The RfC method incorporates
both by expressing units as a human equivalent concentration and by the application of
uncertainty factors. With this dose-duration-response method, doses are expressed as
human equivalent concentrations; however, further species extrapolation is not explicitly
done. What is done is to estimate the 95% lower confidence limit at the specified level
of probability, in this case at the 0.05 and 0.01 levels, for each dose/duration combination
that is not expected to cause adverse effects in exposed populations (i.e., the probability
of exceeding a severity index greater than 1). Other confidence limits, e.g., the 99%
confidence limit, or other probability levels could have been selected. For hydrogen
chloride, the 95% lower limit at the 0.01 level adequately characterized the human data
(Table 8).
It appears that much of the data reported on air concentrations for human
exposure to hydrogen chloride are derived from the same source (Elkins 1959.
Henderson and Haggard 1943), the interpretation of which may be open to question.
Use of all data on adverse effects following short-term hydrogen chloride exposure, as in
this proposed method, may provide a more scientifically defensible rationale for setting
guidelines for protection of human health and for guidelines on response to emergency
situations.
The proposed dose-duration response model estimated values for hydrogen
chloride that closely agreed with reports of human exposure and established regulatory
criteria. The proposed dose-duration-response model, when verified, may provide a
55
-------
useful tool for the risk assessor/risk manager. At a glance, the risk assessor may be able
to evaluate if an adverse effect is likely to occur from exposure to a chemical at a
specified dose and time by referring to a graph or table that incorporates the output of a
dose-duration-response model.
56
-------
4. References
Alarie Y. 1973. Sensory irritation by airborne chemicals. CRC Crit Rev Toxicol
2:299-363.
American Conference of Governmental Industrial Hygienists. 1980. Documentation of
Ttireshold Limit Values in Workroom Air. ACGIH. Cincinnati, Ohio. p. 222.
American Conference of Governmental Industrial Hygienists. 1987. Threshold Limit
Values and Biological Exposure Indices for 1987-1988. ACGIH. Cincinnati, Ohio.
p. 23.
Anzueto A, Switzer W, Kaplan H, et al. 1987. Long-term effects of hydrogen chloride
on pulmonary function and morphology in nonhuman primates. Toxicologist 7:192 (As
cited in EPA 1988).
Archuleta D, Stavert D, Behr M, et al. 1990. Pathologic responses to HF, HBr, and HC1
inhaled by pseudo-mouth breathing and nose breathing rats. Toxicologist 10.
Barrow C, Lucia H, Alarie Y. 1979. A comparison of the acute inhalation toxicity of
hydrogen chloride veiius the thermal decomposition products of polyvinylchloride.
/ Combust Toxicol 6:3-12.
Barrow C, Alarie Y, Warrick J, et al. 1977. Comparison of the sensory irritation
response in mice to chlorine and hydrogen chloride. Arch Environ Health 32:68-76.
Burleigh-Flayer H, Wong K, Alarie Y. 1985. Evaluation of the pulmonary effects of
HC1 using CO2 challenges in guinea pigs. Fund Appl Toxicol 5:978-985.
Commonwealth of Massachusetts. 1990. Summary of Massachusetts' Methodology for
Developing Allowable Ambient Limits. Department of Environmental Quality
Engineering. Office of Research and Standards.
Crump K. 1984. A new method for determining allowable daily intakes. Fund Appl
Toxicol 4:854-871.
Crump K, Howe R. 1985. A review of methods for calculating confidence limits in low
dose extrapolation. In: Toxicological Risk Assessment. Krewski D, ed. CRC Press,
Inc. Canada.
Darmer K Jr, Kinkead E, DiPasquale L. 1974. Acute toxicity in rats and mice exposed
to hydrogen chloride gas and aerosols. Am Ind HygAssocJ 35:623-631.
Dourson M. 1986. New approaches in the derivation of acceptable daily intake (ADI).
Comments Toxicol 1:35-48.
57
-------
Dourson ML, Hertzberg RC, Hartung R, Blackburn K. 1985. Novel methods for the
estimation of acceptable daily intake. Toxicol Ind Health 1:23-41.
Elkins H. 1959. Tlie Chemistry of Industrial Toxicology. 2nd Edition. John Wiley and
Sons, New York. pp. 79-80.
Environmental Protection Agency (EPA). 1980. Guidelines and methodology used in
the preparation of health effects assessment chapters of the consent decree water
quality criteria. Fed Reg 45:79347-79357.
Environmental Protection Agency (EPA). 1988. Health Assessment Document for
Chlorine and Hydrogen Chloride. Office of Health and Environmental Assessment.
September 1988.
Environmental Protection Agency (EPA). 1990. Interim Methods for Development of
Inhalation Reference Concentrations. EPA/600/8-88/006A. Environmental Criteria
and Assessment Office, Office of Health and Environmental Assessment. August
1990.
^Environmental Protection Agency (EPA). 1991. Integrated Risk Information System
(IRIS). Hydrogen chloride. January 14, 1991.
GEOMET Technologies, Inc. 1981. Hydrogen Chloride: Report 4, Occupational
Hazard Assessment. U.S. Department of Health and Human Services. NIOSH,
Cincinnati, OH. p. 56.
Guth DJ, Jarabek AM, Wymer L, Hertzberg RC. 1991. Evaluation of Risk Assessment
Methods for Short-Term Inhalation Exposure. Presented at 84th Annual Meeting Air
& Waste Management Association, Vancouver, British Columbia, June 16-21, 1991.
Hartung R. 1986. Ranking the severity of toxic effects. In: Hemphill D, ed. Trace
Substances in Environmental Health. XX. University of Missouri-Columbia,
Columbia, Missouri, pp. 204-211.
Hartzell G, Packham S, Grand A, et al. 1.985. Modeling of lexicological effects of fire
gases: III. Quantification of post-exposure lethality of rats from exposure to HC1
atmospheres. J Fire Sci 3:195-207.
Hazardous Substances Data Base (HSDB). 1991.
Henderson Y, Haggard H. 1943. Noxious Gases. Reinhold Publishing Corporation, New
York. pp. 124-141.
58
-------
Hertzberg R. 1989. Fitting a model to categorical response data with application to
species extrapolation of toxicity. Proceedings of 26th Hanford Life Sciences
Symposium. Modeling for Scaling to Man: Biology, Dosimetry and Response. Health
Phys (Suppl 1):405-409.
Hertzberg R, Knauf L. 1989. Statistical Methods for Estimating Risk for Exposure
Above the Reference Dose. Environmental Criteria and Assessment Office, Office of
Health and Environmental Assessment, Office of Research and Development, U.S.
Environmental Protection Agency, Cincinnati, OH.
Hertzberg R, Miller M. 1985. A statistical model for species extrapolation using
categorical response data. Toxicol Ind Health 1:43-57.
Kaplan H. 1987. Effects of irritant gases on avoidance/escape performance and
respiratory response of the baboon. Toxicology 47:165-179.
Kaplan H, Anzueto A, Switzer W, et al 1988. Effects of hydrogen chloride on
respiratory response and pulmonary function of the baboon. / Toxicol Environ Health
23:473-493.
Kaplan H, Anzueto A, Switzer W, et al. 1986. Respiratory effects of hydrogen chloride
in the baboon. Toxicologist 6:52 (As cited in EPA 1988).
Kaplan H, Grand A, Rogers W, et al 1984. A Research Study of the Assessment of
Escape Impairment by Irritant Combustion Gases in Postcrash Aircraft Fires.
AD-A146-484. Prepared by South West Research Institute, San Antonio. Texas.
Prepared for Federal Aviation Administration, Atlantic City Airport, NJ.
Kusewitt D, Stavert D, Ripple G, et al. 1989. Relative acute toxicity in the respiratory
tract of inhaled hydrogen fluoride, hydrogen bromide and hydrogen chloride.
Toxicologist 9
Lucia H, Barrow C, Stock M, et al. 1977. A semiquantitative method for assessing
anatomic damage sustained by the upper respiratory tract of the laboratory mouse,
Mus musculis. J Combust Toxicol 4:472-486.
Machle W, Kitzmill K, Scott E, Treon J. 1942. The effect of inhalation of hydrogen
chloride. J Ind Hyg Toxicol 24:222-225.
McCune E. 1990. Personal communication. North Carolina Division of Environmental
Management, Air Quality Section. February 2, 1990.
National Air Toxics Information Clearinghouse (NATICH). 1988. NATICH Data Base
Report on State. Local and EPA Air Toxics Activities. USEPA, Office of Air Quality
Planning and Standards. Research Triangle Park, NC. July 1988.
59
-------
North Dakota State Department of Health. 1990. Policy for the Control of Hazardous
Air Pollutant Emissions in North Dakota. Air Quality Management Branch.
Patty F. 1963. Industrial Hygiene and Toxicology. Vols. 1 and 2. New York:
Interscience Publishers.
Pavlova T. 1976. Disturbance of development of the progeny of rats exposed to
hydrogen chloride. Bull Exp Biol Med 82:1078-1081.
Rhode Island Department of Environmental Management. 1990,, Rhode Island Air
Toxics Guidelines.
Sellakumar A, Snyder C, Solomon J, Albert R. 1985. Carcinogenicity of formaldehyde
and hydrogen chloride in rats. Toxicol Appl Pharmacol 42:417-423.
Wohlslagel J, Dipasquale L, Vernot E. 1976. Toxicity of solid rocket motor exhaust:
effects of HC1, HF, and alumina on rodents. / Combustion Toxicol 3:59-70.
60
-------
APPENDIX A
DISCUSSION OF ACUTE ANIMAL TOXICITY DATA
-------
Hydrogen chloride (HC1) forms a corrosive acid in air and, as such, is acutely
toxic. The primary target for acute inhalation exposure is the respiratory tract. Adverse
effects following acute exposure to hydrogen chloride include irritation, pulmonary
edema, nasal and respiratory tissue injury, and death. These adverse respiratory effects
have been assessed in laboratory animals and have been evaluated by focusing primarily
on three endpoints: lethality, alterations in respiratory function, and histopathological
changes to the nasal or pulmonary tissues.
Review of the available literature indicated that there are some data gaps in our
knowledge concerning the adverse health effects from acute exposure to sublethal
concentrations of hydrogen chloride that impact development of dose-duration-response
relationships. One of these data limitations is that few studies have reported the results
of all three of the endpoints mentioned above by dose group when evaluating the toxicity
of hydrogen chloride. For example, a study may report on alterations in respiratory
function without corresponding histopathological injury to the respiratory tissue or a
study may present mortality data without describing any of the nonlethal effects also
observed. Even though some of these effects have been reported separately in animal
bioassays, they are not independent responses. Rather, these effects represent a
continuum of toxicity, the severity of which is dependent on the concentration of
hydrogen chloride to which the subject was exposed, the duration of that exposure, and
the individual sensitivity. As either the concentration or the time increased, toxic effects
would be expected to increase from slight irritation to severe irritation which overlaps
with slight, reversible tissue injury and progresses to severe tissue destruction, pulmonary
distress and finally death. As such, it is important to develop an overall perspective for a
dose-response assessment of the adverse health effects of hydrogen chloride by analyzing
A-l
-------
data on multiple endpoints across studies. For example, Barrow et al (1977) observed
decreases in the respiratory rate of mice over time which were concentration related.
The significance of these responses relative to the animals' health may be qualitatively
assessed, but when combined with the histopathological effects reported in the same
species of mice exposed by the same protocol to similar concentrations and exposure
duration (Lucia et al 1977), a more quantitative assessment can be made, Le., the
relative degree of respiratory damage experienced by the animals that corresponds to the
changes observed in respiratory rates. By comparing across studies it may be possible to
generalize these results and equate alterations in respiratory function with
histopathological injury and thereby assess the extent and permanence of the respiratory
damage.
Another limitation of the data is that only a limited number of exposure durations
have been tested. Most exposure protocols required examination for effects of high
concentrations for durations of less than 30 minutes. Only two studies reported effects
from single-exposure durations of greater than 30 minutes. In one study, Machle et al.
(1942) exposed guinea pigs and rabbits (three animals per group) to concentrations
ranging from 33 to 13,667 ppm (50 to 20.500 mg/m3) for single-exposure periods up to 6
hours. The other investigation, Hartzell et al. (1985), was a series of acute toxicity
studies in rats and the longest exposure duration was 60 minutes. To overcome this
limitation, evaluation of studies should be conducted across studies rather than evaluation
of each study independently.
It is important to note that the mathematical model described in Chapter 3 of
this report allows an analysis of these data across endpoint, time, and specific study. The
data from the available literature used in the dose-duration-response analyses (presented
A-2
-------
briefly in Chapter 3) are discussed in the following paragraphs. The studies considered
are grouped according to the main endpoints reported in these studies. Some cross-study
or cross-endpoint comparisons have been made; however, comparisons across studies,
especially as they relate to the severity index assigned for analysis, are described in
section 3.2.1 and are not repeated in this section.
Acute lethality studies
The LC50 values derived in acute toxicity studies in rats, mice, rabbits, and guinea
pigs indicate that the hydrogen chloride concentrations that result in mortality lie within
fairly similar ranges for all species tested. However, rats may be somewhat less sensitive
to the acute lethal effects of hydrogen chloride than mice and guinea pigs. The results of
these lethality studies demonstrated that the toxicity of hydrogen chloride is related to the
"exposure dose," which is dependent upon the concentration of hydrogen chloride to
which the animals were exposed and the duration of exposure. In a series of studies by
Darmer et al. (1974) rats and mice were exposed to high concentrations of hydrogen
chloride for either 5 or 30 minutes. Groups of ten male Sprague-Dawley rats were
exposed to concentrations ranging from 30,000 to 57,290 ppm (45,000 to 85,935 mg/m3)
for 5 minutes, or 2,078 to 6,681 ppm (3,117 to 10,022 mg/m3) for 30 minutes. Calculated
LC50 values were 40,989 ppm (61,483 mg/m3) for the 5-minute exposure but were
4,701 ppm (7,051 mg/m3) for the 30-minute exposure. During the 7-day post-exposure
observation period, no deaths occurred in rats exposed to the lowest concentration for
either exposure time. In the study with CF-1 male mice, concentrations ranged from
3,200 to 30,000 ppm (4,800 to 45,000 mg/m3) for 5 minutes, or 410 to 5,363 ppm (615 to
8,045 mg/m3) for 30 minutes. LC50 values of 13,745 and 2,644 ppm (20,617 and
A-3
-------
3,966 mg/m3) were determined for the 5 and 30 minute exposures, respectively. A low
incidence of mortality (a total of four out of 30 mice) occurred in the 5-minute exposure
groups at concentrations up to 6,410 pprn (9,615 mg/m3) hydrogen chloride, the fourth
highest exposure concentration. No deaths were reported in the mice exposed to
410 ppm (615 mg/m3) hydrogen chloride for 30 minutes, but mortality increased rapidly
with increasing hydrogen chloride concentration.
In studies by Hartzell et al. (1985), male Sprague-Dawley rats (6 or 8 rats/group)
were exposed to multiple concentrations of hydrogen chloride over a range of 5 to 60
minutes. The values for the 14-day LC50s appeared to be a function of both time and
concentration, e.g., no deaths occurred ai: a 5-minute exposure to 9,200 ppm (13,800
mg/m3), whereas exposure for 10 minutes to 9,170 ppm (13,755 mg/m3) caused~100%
mortality. At the longest exposure duration, 60 minutes, no mortality occurred at a
concentration of 1,793 ppm (2,690 mg/m3). The LC50s values from the 5-, 10-, 15-, 30-,
and 60-minute exposures were 15,900, 8,370, 6,920, 3,715, and 2,810 ppm (23,850, 12,555,
10,380, 5,572, 4,215 mg/m3), respectively.
In guinea pigs and rabbits of unspecified strains, an exposure concentration of
3,667 ppm (5,500 mg/m3) for 5 minutes, caused no mortality, but at 4,333 ppm
(6,500 mg/m3) for 30 minutes or 667 pprn (1,000 mg/m3) for 2 to 6 hours, 100% mortality
occurred in both species (Machle et al. 1942). Mortality was observed in English smooth-
haired guinea pigs at a lower exposure concentration of 1,040 ppm (1,560 mg/m3) for 30
minutes, when the animals were subjected to hydrogen chloride by head-only exposures
(Burleigh-Flayer et al 1985). In baboons, hydrogen chloride concentrations up to 11,400
ppm (17,100 mg/m3) for 5 minutes were tolerated without mortality (Kaplan et al. 1984).
A-4
-------
The apparent discrepancy in the magnitude of LC50 values in rats from the
5-minute exposure groups may be attributed to several factors. Whole body exposures
were used in the studies by Darmer et al. (1974), while Hartzell et al (1985) subjected
animals to head-only exposures. This difference in protocol may have altered the actual
dose of hydrogen chloride received enough to result in a difference in the observed
toxicity in the shorter-term exposures (e.g., 5 minutes). Another distinction is that post-
exposure observation times varied from 7 days in the Darmer et al. study to 14 days in
the Hartzell et al. study. It is reasonable to assume that extending the observation time
two-fold would provide data points of moribund animals which might have succumbed
within the additional time period and therefore, lower the median lethal dose value. The
differential values reported for toxicity in guinea pigs may also be attributed to exposure
protocols of head-only versus whole body; however, the study by Machle et al. (1942) is
confounded by exposure in a copper chamber, by hydrogen chloride gas generated from
sulfuric acid, and by a non-homogeneous test population.
Histopathological changes in nasal and respiratory tract tissues
When male Swiss-Webster mice (four mice/group) were exposed to a range of
hydrogen chloride concentrations for 10 minutes, superficial ulceration of the epithelium
of the external nares was evident at 17 ppm (26 mg/m3), and as hydrogen chloride
concentrations increased, progressive damage occurred (Lucia et al. 1977). Total
destruction of the upper respiratory tract was observed at 7.279 ppm (10,919 mg/m3).
However, the extent of injury to the lower respiratory tract was not evaluated. In a
similar study, groups of eight male Fischer-344 rats were exposed for 30 minutes to
A-5
-------
concentrations of 100 to 1,000 ppm (150 to 1,500 mg/m3) (Kusewitt et al 1989). The
observed injury scaled with concentration and was confined only to the nasal cavity. In
male English smooth-haired guinea pigs (four guinea pigs/group) following exposure to
1,040 ppm (1,560 mg/m3) hydrogen chloride for 30 minutes, tissue damage in both the
large airways and alveolar regions was noted. Squamous metaplasia and bronchitis in the
large conducting airways along with acute inflammation of the alveoli were observed
(Burleigh-Flayer et al. 1985).
The extent and location of damage to the respiratory tract may be species-specific
and may also be related to breathing patterns. A recent study with Fischer-344 rats (five
to eight rats/group) demonstrated that lesion sites in the respiratory tract are primarily
dependent upon inhalation pathway, Le., mouth versus nose breathing of hydrogen
chloride gas (Archuleta et al. 1990). Pseudo-mouth breathing (PM), which consisted of
animals being fitted with mouthpieces with attached silastic tracheal tubes, and normal
nose-breathing (NB) rats, were exposed to 1,000 ppm (1,500 mg/m3) hydrogen chloride
for 30 minutes. Histopathological analysis of the upper and lower respiratory tract
revealed that tissue injury was confined to the nasal region only for NB exposure,
whereas the PM exposure produced major tissue disruption in the trachea and peripheral
lung damage. However, this difference may be less apparent as the hydrogen chloride
concentration increases because the filtering action of the nasal mucosa may be less
efficient as concentration increases.
A-6
-------
Irritation
Hydrogen chloride is both a sensory and pulmonary irritant (Burleigh-Flayer et al
1985). Sensory irritation was manifested by a decrease in the respiration rate at lower
hydrogen chloride concentrations, but pulmonary irritation characterized by an increase
in respiratory rate followed by a decreased rate occurred at higher doses (Burleigh-Flayer
et al, 1985).
The onset of sensory irritation in mice or guinea pigs was concentration and time
dependent (Barrow et al 1977, Burleigh-Flayer et al 1985). Within 5 to 6 minutes at
exposure to 40 ppm (60 mg/m3), irritation was evident in male Swiss-Webster mice (four
mice/group). As the concentrations increased, sensory irritation occurred within 1 minute
of exposure, and the RD50 (concentration producing 50% decrease in respiratory rate) in
mice was 309 ppm (460 mg/m3) (Barrow et al 1977).
In guinea pigs, pulmonary irritation occurred after 20 minutes exposure to
320 ppm (480 mg/m3) and in less than 4 minutes to 1,380 ppm (2,070 mg/m3) (Burleigh-
Flayer et al. 1985). Persistent impairment of respiratory function was observed in guinea
pigs after 30 minute exposure to 1,040 or 1,380 ppm (1,560 or 2,070 mg/m3) which was
supported by evidence (in the 1,040 ppm exposed animals) of histological damage in both
the airways and alveolar regions.
The no-effect level for irritation in the baboon was 190 ppm (285 mg/m )
hydrogen chloride for 5 minutes (Kaplan et al. 1984). Irritation occurred at a level of
810 ppm (1,215 mg/m ) but animals were asymptomatic post-exposure, whereas exposure
to 2,780 ppm (4,170 mg/m3) caused a reversible post-exposure cough. Baboons tolerated
an exposure of 5,000 ppm (7,500 mg/m3) for 15 minutes without evidence of long-term
A-7
-------
effects. An exposure to 10,000 ppm (15,000 mg/m ) for 15 minutes caused long-term
pulmonary effects, such as chronic nasal obstruction, persistent hyporia, and focal lung
fibrosis (Kaplan et ai 1988, Anzueto et ai 1987). Death from bacterial pneumonia
resulted at levels of 16,570 and 17,290 ppm (24,855 and 25,935 mig/m3) for 5 minutes
(Kaplan 1987).
Extra-pulmonary effects
Extra-pulmonary effects of inhaled hydrogen chloride have been reported in some
experimental studies. Degenerative and necrotic lesions in the heart, liver, and kidneys
have been reported in some studies in which guinea pigs and rabbits were exposed to
hydrogen chloride concentrations ranging from 150 to 3,075 ppm (225 to 4,613 mg/m3)
for periods of 5 minutes to 12 hours (Machle et al. 1942). Exposure of male Swiss-
Webster mice (four mice/group) to concentrations of 130 to 20,000 ppm (190 to 30,000
mg/m3) hydrogen chloride for 10 minutes, produced passive congestion in the intestine,
liver, and kidney (Barrow et al. 1979). However, in both studies it was difficult to
determine the direct contribution of hydrogen chloride since insufficient respiratory
function can cause extra-pulmonary deleterious effects.
Pregnant Wistar rats (8 to 15 rats/group) exposed for 1 hour to 300 ppm
(450 mg/m3), the LC50, exhibited severe lung lesions and disturbed kidney function.
These maternotoxic effects were directly related to disturbances of embryogenesis and
postnatal organ dysfunction in the progeny of the exposed females (Pavlova 1976).
A-8
-------
APPENDIX B
ANIMAL BIOASSAY DATA FOR INPUT INTO THE
DOSE-DURATION-RESPONSE MODEL
-------
Table B-l
Acule Toxicity Data Derived from Animal Bioassays
Hydrogen
exposure
•ng/m3Anima,
45,000
48,383
59,775
67,800
85,935
3,117
4,017
4,607
7,770
9,102
10,022
4,800
7,590
9,218
9,615
11,288
12,098
13,914
20,483
39,728
45,000
chloride
level
mg/m3Human
220,000
236,539
292,233
331,467
420,127
15,239
19,639
22,523
37,987
44,499
48,9%
33,060
52,277
63,489
66,224
77,747
83,326
95,833
141,078
273,629
309,940
Time in
minutes
5
5
5
5
5
30
30
30
30
30
30
5
5
5
5
5
5
5
5
5
5
Severity
index
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
Number of
laboratory
animals
Darmer el aL
10
10
10
10
10
10
10
10
10
10
10
Darmer et aL
10
10
10
10
10
10
10
10
15
15
Indicator
1974 - Rats
0
1
6
7
9
0
1
0
5
8
10
1974 - Mice
1
1
2
0
6
2
5
6
13
13
Comments
Determined post-exposure lethality; other
loxicity symptoms reported were not attributed
to specific exposure concentrations.
Determined post-exposure lethality; other
loxicity symptoms reported were not attributed
to specific exposure concentrations.
-------
Table B-l (continued)
Acute Toxicity Data Derived from Animal Bioassays
Hydrogen
exposure
«.g/m3Animal
615
1,701
4,017
4,082
4,413
4,607
6,068
6,114
8,045
DO
tv>
13,800
16,178
18,876
21,460
23,189
30,450
8,166
11,444
12,171
12,638
13,755
6,540
9,257
11,970
13,440
14,985
chloride
level
-------
Table B-l (continued)
Acute Toxicity Data Derived from Animal Bioassays
CO
Hydrogen chloride
exposure level
roe/"1 Animal mg/m3||Uman
Time in
minutes
Severity
index
Number of
laboratory
animals Indicator
Comments
17,700
21,615
22,875
27,645
33,390
37,950
38,775
41,535
68,250
80,850
115,095
131,490
86,533
105,673
111,833
135,153
163,240
185,533
189,567
203,060
333,667
395,267
562,687
642,840
5
5
5
5
5
5
5
5
5
5
5
5
3
3
4
4
4
4
4
4
4
4
4
4
285
1,215
1,335
1,410
4,170
17,100
24,855
25,935
657
2,803
3,080
3,253
9,619
39,447
57,336
59,828
Kaplan el al 1984 •• Rats
1
1
1
1
Monitored clinical symptoms during and post-
exposure and body weight and survival post-
exposure.
5
5
5
5
5
5
5
5
1 1
2
2
2
2
2
4
4
Kaplan et al 1984 -- Baboons
1
Monitored clinical symptoms during and post-
exposure, and survival post-exposure.
26
197
420
740
1,085
1,632
2,960
4,665
10,919
179 (5)a
1,357 (38)
2,893 (82)
5,097 (145)
7,473 (212)
11,240(319)
20,387 (578)
32,130(911)
75,205 (2132)
10
10
10
10
10
10
10
10
10
1
2
2
2
2
3
4
4
4
Lucia et al. 1977 -- Mice
4 ' 4
4
4
4
4
4
4
4
4 4
Historically evaluated anatomical injury to the
upper respiratory tract. Authors ranked effects
using a semi-quantitative scale.
-------
Table B-l (continued)
Acute Toxicity Data Derived from Animal Bioassays
-------
Table B-l (continued)
Aculc Toxicity Data Derived from Animal Bioassays
Hydrogen
exposure
•"^Animal
2,720
3,872
4,911
5,912
6,683
834
1,478
2,081
2,853
3,714
chloride
level
m&/m3Human
13,298
18,959
24,(X)9
28,903
32,672
5,744
10,180
14,333
19,650
25,580
Time in
minutes
60
60
60
60
60
60
60
60
60
60
Severity
index
4
4
4
4
4
4
4
4
4
4
Number of
laboratory
animals
Wohlslagel el aL
10
10
10
10
10
Wohlslagel el aL
10
10
10
10
10
Indicator
1976 - Rats
0
2
6
8
10
1976 - Mice
2
3
6
8
10
Comments
Determined post-exposure lethality. Observed
toxicity symptoms and gross histopathology were
not discussed per specific exposure
concentration.
Determined post-exposure lethality. Observed
toxicity symptoms and gross histopathology were
not discussed per specific exposure
concentration.
Burleigh-Flayer el aL 1985 - Guinea pigs
480
1,020
1,560
2,070
1(K)
1,000
5,5(X)
6,500
1,163
2,471
3,780
5,016
154
1,540
8,471
10,011
30
30
30
30
1,800
360
5
30
2
3
4
4
3
4
3
4
4
4
8
8
Machle et aL 1942
3
3
3
3
-1
-1
2
3
-- Guinea pigs
0
3
3
3
Measured effects of hydrogen chloride on pulmo
nary function. Examined pulmonary histo-
pathology for only one exposure concentration.
Corneal damage and mortality also reported.
Data on lethality were the only useable
information for dose-response effects.
Machle el aL 1942 - Rabbits
100
1,000
5,500
6,500
128
1,278
7,026
8,304
1,800
360
5
30
3
4
3
4
3
3
3
3
0
3
3
3
Data on lethality were the only useable
information for dose-response effect:;.
-------
03
Table B-l (continued)
Acute Toxicity Data Derived from Animal Bioassays
Hydrogen chloride
exposure level Time in
nig/in Animal nig/iTi i-iijiTian iiiinuies
190
730
1,060
1,600
4,570
10,700
12,000
20,000
29,000
60
149
368
660
1,415
1,309 (37)a
5,028 (143)
7,301 (207)
11,020(312)
31,476 (893)
73,697 (2090)
82,651 (2344)
137,751 (3905)
199,739 (5664)
413
1,026
2,535
4,546
9,746
10
10
10
10
10
10
10
10
10
10
10
10
10
10
Severity
index
2
3
3
4
4
4
4
4
4
1
2
2
2
4
Number of
laboratory
animals
Barrow et aL
4
4
4
4
4
4
4
4
4
Barrow el aL
4
4
4
4
4
Indicator
1979 -- Mice
4
-1
-1
-I
-1
-1
2
2
4
1977 -- Mice
4
-1
-1
-1
-1
Comments
Examined histopathological lesions in the upper
respiratory tract and mortality.
Evaluated sensory irritation as measured by
decreases in respiratory rate. Reported one
incidence of mortality.
All dose conversions based on total lung surface area except for values in parentheses which are based on extrathoracic surface area.
-------
APPENDIX C
DERIVATION OF MAXIMUM LIKELIHOOD ESTIMATE BOUNDS
AND LOWER CONFIDENCE LIMITS
-------
Log linear bands in d and T for
P(s>i|d,T) = a
may be found by taking the natural logarithm of the above expression and solving for a
linear equation in logarithm to the base ten for dose and time. For the Weibull model
the linear equation is
Io8l0
-ln(l-a)
= Wlog10(d-d0) + Klog10T
and for the log-logistic model the linear equation is
In
l-o
- qj . W Iog10(d-d0) + K log10T.
For fixed a and using the MLE for qj, W and K bands or lines in log time and dose
may be graphed.
A benchmark band or a lower bound on the P(s>i | d,T) = a line is not as easy to
find. However, a good approximation is to find individual points on the line and connect
them with straight line segments. Since the time considered was 24 hours, twelve equally
spaced intervals in Iog10 (time) were used. For each fixed time (T:) where:
j = 1 to 12, and
T; = equal divisions of time in terms of Iog10 time
the following problem was solved:
C-l
-------
Minimize d,
subject to: P (s > i j d,T:) = o
and LL(0) - LL(0) = 1/2 X2lj2p
Here 9 represents the vector of MLEs and 6 represents the vector of estimates which
satisfy the constraints with LL(*) representing the loglikelihood. For this paper, p = 0.05
or 95% confidence intervals are presented.
C-2
-------
r *
APPENDIX D
COMPUTER OUTPUT FOR DOSE-DURATION-RESPONSE MODEL
-------
Table D-l
Maximum Likelihood Estimates and 95% Lower Bounds
on Hydrogen Chloride Concentrations for Various Exposure Times
and Severity Levels. Human Equivalent Concentrations
Based on Total Lung Surface Area as the Target.
Human Equivalent Concentration fms/m ^
Index Time CHours^
P(s>i)=0.01 1 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
P(s>i)=0.01 2 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
P(s>i)=0.01 3 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
MLE
95.439
56.691
33.675
20.003
11.883
7.0582
4.1927
2.4905
1.4794
.88883
.52111
.31008
3718.1
2208.5
1311.9
779.29
462.92
274.97
163.34
97.024
57.635
34.627
20.301
12.080
14225.
8449.3
5019.0
2981.4
1771.0
1052.0
624.89
371.19
220.50
132.47
77.667
46.215
95% LB
27.789
16.630
9.9205
5.8999
3.4983
2.0681
1.2192
.71688
.42048
.24895
.14344
8.37019E-02
966.32
577.60
346.33
208.36
125.78
76.139
46.166
27.988
16.918
10.297
6.0848
3.6268
5306.1
3194.2
1914.7
1143.0
679.62
402.57
237.65
139.85
82.061
48.574
27.963
16.305
D-l
-------
Table D-1 (continued)
Human Equivalent Concentration fme/m3>)
Index Time (Hours)
P(s>i)=0.05 1 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
P(s>i)=0.05 2 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
P(s>i)=0.05 3 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
MLE
266.45
158.27
94.016
55.847
33.174
19.706.
11.705
6.9531
4.1304
2.4815
1.4549
.86570
10380.
6165.9
3662.6
2175.6
1292.4
767.68
456.01
270.88
160.91
96.673
56.678
33.725
39713.
23589.
14012.
8323.5
4944.4
2937.0
1744.6
1036.3
615.60
369.85
216.84
129.03
95% LB
98.822
59.223
35.367
21.042
12.482
7.3775
4.3463
2.5529
1.4954
.88388
.50828
.29603
3674.6
2139.2
1262.3
761.34
469.34
293.61
183.62
113.27
68.698
41.543
24.250
14.253
20319.
12307.
7409.5
4432.5
2635.5
1558.2
916.75
536.96
313.23
184.20
105.24
60.894
D-2
-------
Table D-2
Maximum Likelihood Estimates and 95% Lower Bounds
on Hydrogen Chloride Concentrations for Various Exposure Times
and Severity Levels. Human Equivalent Concentrations Based on
Total Lung and Extrathoracic Areas as the Target.
Human Equivalent Concentration
Index Time (Hours')
P(s>i)=0.01 1 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
P(s>i)=0.01 2 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
P(s>i) = 0.01 3 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
MLE
.69755
.41507
.24698
.14697
8.74551E-02
5.20386E-02
3.09654E-02
1.84259E-02
1.09646E-02
6.59866E-03
3.87560E-03
2.31013E-03
72.281
43.010
25.593
15.229
9.0622
5.3923
3.2087
1.9093
1.1362
.68376
.40159
.23938
526.38
313.21
186.38
110.90
65.994
39.269
23.367
13.904
8.2739
4.9794
2.9245
1.7432
Cmg/m3>)
95% LB
.12260
7.29395E-02
4.31714E-02
2.54210E-02
1.48935E-02
8.68202E-03
5.03716E-03
2.90902E-03
1.67275E-03
9.69608E-04
5.45357E-04
3.10538E-04
19.544
11.745
7.0064
4.1492
2.4394
1.4240
.82564
.47572
.27249
.15718
8.78675E-02
4.96971E-02
168.10
101.11
60.314
35.680
20.936
12.183
7.0377
4.0373
2.3018
1.3213
.73477
.41358
D-3
-------
Table D-2 (continued)
Human Equivalent Concentration (me/m3)
Index Time (Hours)
P(s>i)=0.05 1 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
P(s>i)=0.05 2 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
P(s>i)=0.05 3 4.26775E-02
7.58947E-02
.13496
.24000
.42677
.75895
1.3496
2.4000
4.2677
7.4947
13.522
24.000
MLE
6.1183
3.6406
2.1663
1.2891
.76707
.45643
.27160
.16161
9.61709E-02
5.78771E-02
3.39993E-02
2.02623E-02
633.98
377.24
224.47
133.57
79.485
47.296
28,143
16.747
9.9653
5.9973
3.5224
2.0096
4616.9
2747.2
1634.7
972.73
578.84
344.43
204.95
121.96
72.571
43.674
25.651
15.290
95% LB
1.4091
.84117
.49920
.29446
.17268
.10069
5.83847E-02
3.36785E-02
1.933 16E-02
1.11804E-02
6.27099E-03
3.56038E-03
236.80
143.54
86.198
51.258
30.183
17.604
10.177
5.8356
3.3228
1.9035
1.0558
.59257
2138.5
1303.6
783.65
465.62
273.16
158.33
90.791
51.585
29.085
16.510
9.0660
5.0427
D-4
-------
Table D-3
Maximum Likelihood Estimates and 95% Lower Bounds
on Hydrogen Chloride Concentrations for Various Exposure Times
and Severity Levels. Based on Concentrations
Used in Animal Experiments
Index Time (Hours)
P(s>i)=0.01 1 .426775E-01
.758947E-01
.134962
.240000
.426775
.758947
1.34962
2.40000
4.26775
7.49466
13.5219
24.0000
P(s>i)=0.01 2 .426775E-01
.758947E-01
.134962
.240000
.426775
.758947
1.34962
2.40000
4.26775
7.49466
13.5219
24.0000
P(s>i)=0.01 3 .426775E-01
.758947E-01
.134962
.240000
.426775
.758947
1.34962
2.40000
4.26775
7.49466
13.5219
24.0000
MLE
19.448
12.066
7.4862
4.6447
2.8819
1.7880
1.1093
.68827
.42704
.26772
.16412
.10199
695.43
431.46
267.69
166.09
103.05
63.935
39.668
24.611
15.270
9.5733
5.8688
3.6470
1949.7
1209.7
750.52
465.65
288.91
179.25
111.21
69.001
42.812
26.840
16.454
10.225
mg/m
95% LB
8.6214
5.3566
3.3182
2.0493
1.2618
.77466
.47425
.28955
.17634
.10830
6.4825 1E-02
3.92769E-02
388.18
242.98
151.41
93.911
57.979
35.630
21.801
13.286
8.0678
4.9367
2.9417
1.7738
1226.4
768.65
479.15
296.99
183,04
112.18
68.403
41.521
25.104
15.295
9.0727
5.4472
D-5
-------
Table D-3 (continued)
Index Time (Hours')
P(s>i)=0.05 1 .426775E-01
.758947E-01
.134962
.240000
.426775
.758947
1.34962
2.40000
4.26775
7.49466
13.5219
24.0000
P(s>i)=0.05 2 .426775E-01
.758947E-01
.134962
.240000
.426775
.758947
1.34962
2.40000
4.26775
7.49466
13.5219
24.0000
P(s>i)=0.05 3 .426775E-01
.758947E-01
.134962
.240000
.426775
.758947
1.34962
2.40000
4.26775
7.49466
13.5219
24.0000
MLE
57.171
35.470
22.007
13.654
8.4718
5.2561
3.2611
2.0233
1.2554
.78702
.48247
.29982
2044.3
1268.4
786.94
488.25
302.94
187.95
116.61
72.350
44.890
28.143
17.252
10.721
5731.6
3556.0
2206.3
1368.9
849.32
526.94
326.93
202.84
125.85
78.901
48.369
30.058
mg/m
95% LB
27.370
17.028
10.560
6.5276
4.0218
2.4699
1.5122
.92315
.56200
.34499
.20636
.12494
1271.5
799.19
499.64
310.59
191.96
117.95
72.092
43.850
26.559
16.205
9.6259
5.7859
4137.5
2612.7
1637.8
1018.2
627.57
383.63
232.87
140.54
84.443
51.129
30.140
17.993
D-6
------- |