EPA/600/R-16/147 I June 2016
www.epa.gov/homeland-security-research
United States
Environmental Protection
Agency
oEPA
Review of Bacillus Dose
Response Data for Human Health
Risk Assessment
Office of Research and Development
National Homeland Security Research Center
-------
This page is intentionally left blank.
-------
&EPA
United States
Environmental Protection
Agency
EPA/600/R-16/147 I June 2016
www2.epa.gov/homeland-security-research
Review of Bacillus anthracis Dose-Response Data for
Human Health Risk Assessment
United States Environmental Protection Agency
Cincinnati, Ohio, 45268
Office of Research and Development
National Homeland Security Research Center
-------
This page is intentionally left blank.
-------
Acknowledgements
The United States Environmental Protection Agency (EPA) would like to acknowledge Mr. Marshal
Gray, Dr. Abdel Kadry, Ms. Eletha Brady-Roberts, Dr. Tonya Nichols, and Dr. Emily Snyder for
their thoughtful review of the report.
Disclaimer
The U.S. Environmental Protection Agency through its Office of Research and Development
funded and managed the research described here under Contract No. SP0700-00-D-3180/CB-l 1-
0232 to Battelle and Contract No. EP-C-14-001 to ICF International, under Battelle contract
13KJB00004 Task Order WA-1-24. It has been subjected to the Agency review and has been
approved for publication. Note that approval does not signify that the contents necessarily reflect
the views of the Agency. Mention of trade names, products, or services does not convey official
EPA approval, endorsement, or recommendation.
Technical Point of Contact
Sarah Taft, Ph.D.
U.S. Environmental Protection Agency
26 West Martin Luther King Drive
MS NG-16
Cincinnati, OH 45628
Taft.Sarah@EPA.GOV
513-569-7037
li
-------
Table of Contents
Acknowledgements ii
Disclaimer ii
Table of Contents iii
List of Tables v
List of Figures vii
List of Appendices viii
Acronyms ix
Executive Summary xii
1 Introduction 1
2 Purpose and Scope 3
3 Framework for Microbial Human Health Risk Assessment 5
4 Problem Formulation 8
4.1 Conceptual Site Model 9
5 Effects Assessment 14
5.1 Hazard Identification 14
5.2 Disease Pathogenesis in the Context of Key Events 20
5.3 Overview of Microbial Dose-Response Analysis 31
5.4 Evaluate the Microbial Dose-Response Data 33
5.4.1 Animal Model Selection Using Concordance of Pathology 33
5.4.2 Identification of Microbial Dose-Response Data 59
5.5 Model the Dose-Response Relationship 84
iii
-------
5.5.1 Determination of Dose Metric 85
5.5.2 Empirical and Mechanistic Modeling Approaches 89
5.5.3 Mathematically Modeling the Microbial Dose-Response Relationship 95
5.6 Conduct Interspecies Extrapolation 97
5.6.1 Review of Interspecies Extrapolation Approaches for Chemical Agents 99
5.6.2 Published Approaches for Interspecies Extrapolation of B. anthracis 103
5.6.3 Proposed Framework for Interspecies Extrapolation for B. anthracis 103
5.6.4 Available Kinetic Data 107
5.6.5 Available Dynamic Data Ill
5.6.6 Summary of Extrapolation Framework for B. anthracis 112
6 Conclusion 114
7 References 127
iv
-------
List of Tables
Table 5-1. Development of Microbial Dose-Response Relationships 32
Table 5-2. Evaluation of Microbial Dose-Response Data 33
Table 5-3. Reported Human Autopsy or Pathology Data by Outbreak or Event 37
Table 5-4. Summary of Human Pathology Relative to Twenhafel (2010) Key Findings 39
Table 5-5. Studies Reporting Inhalation Anthrax Pathology by Rabbit Breed 41
Table 5-6. Summary of Rabbit Pathology Relative to Twenhafel (2010) Key Findings 42
Table 5-7. Studies Reporting Inhalation Anthrax Pathology by Nonhuman Primate Species and
Strain 46
Table 5-8. Summary of Nonhuman Primate Pathology Relative to Twenhafel (2010) Key
Findings 49
Table 5-9. Key Human Histopathological Findings Relative to Time-Dependent Pathology in the
Rabbit and Nonhuman Primate after Single-Dose Exposure 59
Table 5-10. Additional Data for the Human 66
Table 5-11. Single- and Multiple-Dose Key Studies for the Rabbit 73
Table 5-12. Single-Dose Key Study for the Nonhuman Primate 77
Table 5-13. Multiple-Dose Additional Data for the Nonhuman Primate 79
Table 5-14. Oral Dose-Response Data 79
v
-------
Table 5-15. Summary of Number of Key Studies, Supporting Studies, and Additional Data
Sources for the Human, Rabbit, and Nonhuman Primate 81
Table 5-16. Identification of Twenhafel (2010) Key Human Histopathological Findings in Rabbit
and Nonhuman Primate Key Studies 83
Table 5-17. Development of Microbial Dose-Response Relationships 84
Table 5-18. Examples of Mathematical Dose-Response Models for Inhalation Anthrax in the
Rabbit, Nonhuman Primate, or Human by Type of Model 93
Table 5-19. Conduct Interspecies Extrapolation 98
Table 5-20. Summary Table of B. anthracis Deposition Data for the Rabbit 108
Table 5-21. Deposition Efficiencies for Different Annotated Regions in the Rabbit and the
Human Ill
Table 6-1. Summary Table for Data Gaps and Science Policy Gaps 126
vi
-------
List of Figures
Figure 3-1. Elements of the U.S. Environmental Protection Agency (2014a) human health risk
assessment framework and associated report content 6
Figure 4-1. Generic conceptual site model 10
Figure 5-1. Key events determination for inhalation anthrax modified from Hines and Comer
(2012) 23
Figure 5-2. Comparison of mechanistic models relative to biological representation, empirical
curve-fitting, and complexity 92
Figure 5-3. Calculation of an RDDR-based dosimetric adjustment factor 110
Figure 6-1. Science questions and associated elements of the U.S. Environmental Protection
Agency (2014a) human health risk assessment framework 116
vii
-------
List of Appendices
Appendix A - Transmission and Pathogenesis Considerations for Biological Threat Agents
Appendix B - Historical Approaches to Microbial Dose-Response Relationship Development for
Bacillus anthracis
Appendix C - Data Summary Table for End-stage Inhalation Anthrax Pathology of the Human,
Nonhuman Primate, and Rabbit
Appendix D - Bacillus anthracis Dose-Response Data for the Rabbit Characterized as
Supportive Data or Additional Data
Appendix E - Bacillus anthracis Dose-Response Data for the Nonhuman Primate Characterized
as Supportive Data or Additional Data
Appendix F - Conducting Benchmark Dose Analysis for Microbial Pathogens
viii
-------
Acronyms
ADME adsorption, distribution, metabolism, and excretion
AIC Akaike Information Criterion
AMWG Anthrax Modeling Working Group
BBDR biologically based dose-response
BMD benchmark dose
BMDX benchmark dose for response in x% of individuals
BMDLx the 95% lower statistical confidence limit of the BMD when the 95% lower
confidence limit is applied to the estimated slope parameter value
BMR benchmark response
BslA Bacillus anthracis S-layer protein A
CBRN chemical, biological, radiological, and nuclear
CDC U.S. Centers for Disease Control and Prevention
CFD computational fluid dynamics
CSM conceptual site model
CFU colony forming unit(s)
CI confidence interval
DAF dosimetric adjustment factor
DDEF data-derived extrapolation factor(s)
DHHS U.S. Department of Health and Human Services
EISD Exposure - Infection - Symptomatic illness-Death
EPA U.S. Environmental Protection Agency
ET edema toxin
Fr regional fraction deposition
ix
-------
GSD
geometric standard deviation
HED
human equivalent dose
HHRA
human health risk assessment
Ho
null hypothesis
ID
infectious dose
IDX
infectious dose for x% of individuals
LD
lethal dose
LD50
median lethal dose
LOAEL
lowest observable adverse effect level
LT
lethal toxin
MAPKK
mitogen-activated protein kinase kinase
MMAD
mass median aerodynamic diameter
NHP
nonhuman primate
NOAEL
no observable adverse effect level
PBBK
physiologically based biokinetic
PBPK
physiologically based pharmacokinetic
PCR
polymerase chain reaction
POD
point of departure(s)
RDDa
Regional Deposited Dose for the Animal
RDDh
Regional Deposited Dose for the Human
RDDR
Regional Deposited Dose Ratio
SAr
regional surface area
SD
standard deviation
-------
USAMRIID U. S. Army Medical Research Institute of Infectious Diseases
UF uncertainty factor
jam micrometer
-------
Executive Summary
As one of the lead federal agencies supporting decontamination activities after a biological
incident, the U.S. Environmental Protection Agency (EPA) has been systematically evaluating
microbial dose-response data and their application for decision-making to support emergency
management and decontamination activities. Risk-based approaches are desirable because they
provide a formalized process to evaluate the hazard posed by these agents. The hazard posed by
a release of Bacillus anthracis spores has made this agent a focus of considerable research by the
EPA and others to identify and evaluate available data for microbial risk assessment. Given
advances in the body of knowledge, a systematic review of B. anthracis data that can be used to
support the development of a dose-response relationship or the use of B. anthracis dose-response
data in a human health risk assessment (HHRA) is now warranted.
Given the breadth of available microbial dose-response data, science questions were generated to
focus review on data necessary to perform a HHRA for B. anthracis. The following science
questions are considered in the evaluation:
• What natural history data are available to inform development of a site-specific conceptual
site model (CSM) for the generic exposure scenario?
• What data are available to support the development of the hazard identification, including
disease pathogenesis data?
• What data support the use of the rabbit and nonhuman primate animal models for
development of dose-response relationships?
• What dose-response data are available for inhalation and oral exposure in the rabbit,
nonhuman primate, and human that may be appropriate for development of a microbial dose-
response relationship?
xii
-------
• What are available approaches to model a microbial dose-response relationship?
• How might an animal-to-human extrapolation be conducted with B. anthracis dose-response
data and what data are available?
Results were presented using the EPA Framework for Human Health Risk Assessment to Inform
Decision Making (U.S. Environmental Protection Agency, 2014a) (hereinafter: the framework)
as an organizing structure to report results from evaluation of the science questions.
A considerable body of knowledge is now available for the development of a site-specific
HHRA for B. anthracis. There are sufficient data to develop the CSM and generate the hazard
identification, as well as data and methods to generate a dose-response relationship for B.
anthracis and conduct a partial interspecies extrapolation. While there are sufficient data to
generate a quantitative HHRA, data quality and the presence of data gaps may contribute to
potentially high levels of uncertainty in the risk assessment outputs. Depending on the intended
use of the risk assessment outputs, these data may not be acceptable for all types of risk-based
decision-making. Microbial risk assessors who are assisting in the initial planning and scoping
element of the HHRA should take care to communicate these potential data limitations to
decision-makers early in the process.
The most significant data gap relates to the lack of high quality dose-response data, defined as
possessing sufficient quality to be categorized as Key Data. This clearly affects the rigor of the
risk assessment. An additional data gap is the lack of basic mechanistic data for the initiation of
infection and dynamics of the early infection process. These mechanistic data would greatly
assist in the confirmation of appropriate dose metrics and inform the interspecies extrapolation
process. However, alternative dose metrics can be assessed to see if substantive differences in
-------
outputs result from different choices and the interspecies extrapolation process can be conducted
in part to address kinetic elements.
This effort also revealed science policy gaps related to generation of a site-specific HHRA for B.
anthracis inhalation exposure. Science policy gaps also affect current readiness to generate a site-
specific HHRA for B. anthracis inhalation exposure. The selection of appropriate benchmark
response (BMR) targets for reporting and risk-based decision-making for microbial pathogens is
a current policy gap. While technical knowledge may inform BMR selection relative to known
data set characteristics for benchmark dose (BMD) modeling, selection of values for reporting
and risk-based decision-making may incorporate numerous policy considerations. An additional
science policy gap is the management of uncertainty in the interspecies extrapolation given the
current inability to address dynamic differences between the animal model and the human. In
addition to a statement of this uncertainty in the risk characterization, a default adjustment factor
could be considered for use until further data or methodologies are available.
xiv
-------
1 Introduction
As one of the lead federal agencies supporting decontamination activities after a biological
incident, the U.S. Environmental Protection Agency (EPA) has been systematically evaluating
microbial dose-response data and their application for decision-making to support emergency
management and decontamination activities. Risk-based approaches are desirable because they
provide a formalized process to evaluate the hazard posed by these agents. The potential hazard
resulting from exposure to residual biological contamination after buildings or other areas are
cleared for re-entry is a significant concern for decision-makers. The hazard posed by residual
contamination is greatest for biological agents that are highly persistent, resistant to
decontamination, and with potential to cause serious or lethal illness at relatively low doses.
Interest in low-dose dose-response relationships for Bacillus anthracis exposure can be traced to
data gaps made apparent during the civilian response to the 2001 anthrax letter event. The
importance of the assessment of low-dose B. anthracis exposures, such as those potentially
resulting from bioterrorism, was identified in publications shortly after the 2001 anthrax letter
event (Dull et al., 2002; Haas, 2002; Peters and Hartley, 2002; Gutting et al., 2008). Ongoing
preparedness activities have continued to identify the need for the assessment of low-dose
exposures (Coleman et al., 2008; Taft and Hines, 2012; Gutting et al., 2013).
Potential health effects from a release of B. anthracis spores have made this agent a focus of
considerable research by the EPA to identify and evaluate available data for microbial risk
assessment. Although B. anthracis is the most highly studied of the currently known biothreat
agents, significant data gaps have been identified for the microbial dose-response analysis of
human exposure to low-dose exposures (Wilkening, 2006).
-------
There is no technical or regulatory consensus for a B. anthracis dose-response relationship
suitable for risk-based decisions (Taft and Hines, 2012). The lack of a dose-response relationship
for B. anthracis is one significant impediment to the use of risk-based management approaches.
However, there are multiple steps in the risk assessment process that incorporate microbial dose-
response data. There has been considerable research performed since 2001 to better understand
inhalation anthrax and its potential transmission after a biological incident. However, a
systematic review is needed to evaluate currently available open source B. anthracis data to
assess its suitability for use in a human health risk assessment (HHRA) microbial dose-response
analysis. This report conducts a systematic review of B. anthracis dose-response data that can be
used to inform development of a dose-response relationship or to support the use of B. anthracis
dose-response data in a HHRA.
2
-------
2 Purpose and Scope
The primary purpose of this report is to provide open source data and analysis approaches that
can be used to develop a site-specific HHRA for B. anthracis. The report presents the results of
an agent-specific planning activity for B. anthracis that evaluated published dose-response data,
identified data and process gaps for microbial dose-response analysis of the agent, and identified
science policy gaps that may be filled to conduct a site-specific HHRA for this agent. The data
are organized following guidelines in the EPA Framework for Human Health Risk Assessment to
Inform Decision Making (U.S. Environmental Protection Agency, 2014a).
Given the breadth of available microbial dose-response data, science questions were generated to
focus review on data necessary to perform a HHRA for B. anthracis. The following science
questions are considered:
• What natural history data for B. anthracis are available to inform development of a site-
specific conceptual site model (CSM) for the identified exposure scenario?
• What data are available to support the development of the hazard identification, including
disease pathogenesis data?
• What data support the use of the rabbit and nonhuman primate animal models for
development of dose-response modeling of B. anthracisl
• What dose-response data are available for inhalation and oral exposure in the rabbit,
nonhuman primate, and human that may be appropriate for development of a microbial dose-
response relationship for B. anthracis?
• What are available approaches to model a microbial dose-response relationship for
B. anthracis?
3
-------
• How might an animal-to-human extrapolation be conducted with B. anthracis dose-response
data and what data are available?
The intended audience is the human health risk assessor who is familiar with EPA HHRA
guidance and has experience conducting microbial risk assessment. However, individuals with a
research interest in microbial dose-response analysis of B. anthracis may find utility in the report
for planning research to address data gaps or developing methodology for assessment purposes.
4
-------
3 Framework for Microbial Human Health Risk Assessment
The U.S. Environmental Protection Agency (2014a) framework (hereinafter: the framework) for
HHRA is designed for use with physical, chemical, or biological stressors. Stressors in this
context are agents with the potential to cause harm. According to the framework, risk assessment
is the iterative evaluation of the following elements: (1) planning, scoping, and problem
formulation elements prior to the actual risk assessment; and (2) exposure assessment, effects
assessment, and risk characterization steps of the risk assessment (Figure 3-1).
Figure 3-1 also identifies the risk assessment elements in the framework that incorporate
microbial dose-response data and the report sections where the available data for B. anthracis are
summarized and evaluated. Report content addresses two elements of the framework: problem
formulation and effects assessment. In the problem formulation element (Section 4), there is a
systematic identification of the factors (e.g., stressor(s), receptors, regulatory considerations) that
will be evaluated in the risk assessment process (U.S. Environmental Protection Agency, 2014a).
The CSM (Section 4.1) is a primary output of the problem formulation step. This CSM defines
the hazard to be assessed relative to the relationships between the type and source of stressors,
exposure pathways and completeness of these pathways, receptors, and types of endpoints or
effects (U.S. Environmental Protection Agency, 2014a). It is presented as text, with a graphic
showing the movement of the agent from the source to potential points of
5
-------
Planning & Scoping
I
Problem Formulation
• Conceptual Model
• Analysis Plan
i
Risk Assessment
Conceptual Site Model
Section 4.1
Exposure Assessment
Risk Characterization
Effects Assessment
• Hazard Identification
• Dose-Response
Hazard Identification
Section 5.1
Disease Pathogenesis
Section 5.2
Animal Model Selection Using Concordance of Pathology
Section 5.4.1
Identification of Microbial Dose-Response Data
Section 5.4.2
Determination of the Dose Metric
Section 5.5.1
Empirical and Mechanistic Approaches
Section 5.5.2
Model the Dose-Response Relationship
Section 5.5.3
Conduct Interspecies Extrapolation
Section 5.6
Figure 3-1. Elements of the U.S. Environmental Protection Agency (2014a) human health risk assessment framework and
associated report content.
6
-------
human exposure (U.S. Environmental Protection Agency, 2014a). The model may also include
other considerations depending on the site, hazard, or other assessment-specific factors.
In the effects assessment element (Section 5), the hazard identification and dose-response
assessment characterize the potential effects of exposure from the hazard being assessed (U.S.
Environmental Protection Agency, 2014a). Overall, the effects assessment process considers data
on the types of health effects, exposure pathways and routes of exposure associated with health
effects, and associated dose-response relationship data for those effects. Specifically, the hazard
identification (Section 5.1) identifies the type of hazard posed in the context of an identified
exposure scenario (U.S. Environmental Protection Agency, 2014a). As part of the hazard
identification for microbial hazards, data are presented on the likelihood of disease transmission
and disease severity associated with exposure pathways, potentially sensitive subpopulations,
and possible long-term sequelae. An evaluation of the microbial dose-response data (Section 5.4)
considers both available data for animal model selection and the assessment of dose-response
data. The mathematical modeling of dose-response relationship (Section 5.5) incorporates
decisions regarding the dose metric used for analysis, empirical and mechanistic modeling
approaches, and empirical curve-fitting within a benchmark dose analysis framework. As part of
microbial dose-response analysis, approaches to conduct an interspecies extrapolation (Section
5.6) are also considered.
7
-------
4 Problem Formulation
The risk assessment problem is the determination of the human
health hazard posed by contact with low-levels of residual B.
cmthracis spore contamination in the air and on surfaces. An
example of an exposure scenario consistent with this problem
formulation is exposure to low levels of B. cmthracis spores, such
as might be present following application of remedial technologies
after an intentional or unintentional release of spores in an indoor
environment. Exposure to B. cmthracis spores from other scenarios
that are substantively similar in the route(s) and associated
magnitude(s) of exposure may also be assessed using these data.
The problem formulation for this data evaluation is representative
of a simplified, generic site. However, this does not preclude the
potential presence of other exposure pathways when site-specific
conditions are evaluated in an actual HHRA. The data evaluation is
not inclusive of all fate and transport processes leading to
potentially complete exposure pathways following an outdoor release or natural disease
outbreak. For example, fate and transport pathways related to potential contamination of
agricultural products and/or the food supply are not explicitly evaluated. Natural disease
transmission from infected animals or associated fomites (i.e., objects or surfaces) is also not
considered.
Summary of Findings for
Problem Formulation
• Published reports support
the potential for released
B. cmthracis spores to
result in inhalation,
ingestion, and dermal
exposure with disease
transmission.
• A quantitative HHRA
could be developed with
existing data.
• There is the potential for
high levels of uncertainty
associated with the
quantitative HHRA
outputs from limitations
in dose-response data.
• The ingestion and dermal
pathways are also likely
to be complete but there
are insufficient data to
conduct a quantitative
HHRA.
• The available natural
history data are sufficient
to generate a site-specific
conceptual site model.
8
-------
For this assessment, low dose was defined as the Rickmeier et al. (2001) value of less than 105
colony-forming-unit(s) (CFU) inhaled dose. The original source for the low dose value in
Rickmeier et al. (2001) was not identified, though it is presumed to be a consensus expert
opinion identified by project participants. The primary reason for selection of the value of less
than 105 CFU inhaled dose is that it is less than the commonly cited median lethality value of
1.05 x 105 of Zaucha et al. (1998) for the rabbit. Few microbial dose-response and associated
health studies are conducted with doses below the Zaucha et al. (1998) median lethality value.
While it would be desirable that the defined low-dose level was reflective of a lower response
level, it would not have been practical.
The majority of microbial dose-response and associated hazard data evaluated in this report are
derived from spores manufactured for laboratory use, with the noted exception of the data from
exposure to B. anthracis-coni&minated mill aerosols. It is hypothesized that intentionally
released manufactured spores might include some material modification (e.g., dispersants,
detergents) to increase the hazard posed. However, this assessment will assume that no special
processing techniques are used beyond typical laboratory practices to manufacture the spores
with a consistent, highly respirable size for animal challenge studies.
4.1 Conceptual Site Model
A CSM can be a graphical or text description that concisely conveys the source of exposure,
potential fate and transport mechanisms, completed or potentially completed exposure pathways
to receptors, and associated routes of exposure. A generic CSM was generated using the problem
statement description of the human health hazard posed by contact with low-levels of B.
anthracis spore contamination (Figure 4-1). However, the presentation of this generic model
9
-------
does not preclude the presence of other exposure pathways when site-specific conditions are
evaluated. A site-specific evaluation must be conducted prior to the direct use of the generalized
CSM in a site-specific risk assessment for B. cmthracis.
Primary
Aerosolization
Deposition
Re-aerosolization
Ingestion
Dermal
Release of
Manufactured
Spores
Inhalation
Air
Surfaces and
Other Fomites
Explanation
7^ Pathway is or might be complete
^ and could be significant, but data
are lacking to support quantitative
evaluation.
^ Pathway is or might be complete
and could be significant, quantitative
evaluation should be performed.
Figure 4-1. Generic conceptual site model.
The generic exposure scenario assessed is the release of manufactured spores. Spores of B.
cinthrcicis are hardy and persist for extended periods when released in indoor or outdoor
environments (Inglesby et al., 2002). The exact mechanism of release is not defined (e.g.,
envelope, spray), but the spores are aerosolized as they are released to the air. Primary
aerosolization at the point of release is the fate and transport mechanism that transports spores
through the air medium to allow inhalation by the receptor. Spores may be deposited on surfaces
(e.g., tables, computer screens, carpets) where they may be re-aerosolized into the air medium
and remain aerosolized for extended periods. In addition to surfaces, spores may deposit on other
10
-------
fomites (e.g., clothing), which may allow direct contact with the receptor, or the spores may re-
aerosolize or be transported away from the initial release location with the fomite.
Re-aerosolization of B. anthracis spores from indoor surfaces was described after simulated
office activities approximately one month after the primary aerosolization from the 2001 anthrax
letter event in the Hart Senate building (Weis et al., 2002). Measurements of airborne CFU
concentrations varied based on activity levels in the office area (Weis et al., 2002). Re-
aerosolization of spore-containing particles in outdoor environments was also described in
experimental studies using surrogates of B. anthracis spores (Layshock et al., 2012).
Physical transport within and between indoor and outdoor locations may lead to potentially
complete exposure pathways for receptors in areas away from the initial release point. Transfer
between indoor and outdoor environments (and vice versa) through building air intake and
removal structures (e.g., heating, ventilation and cooling equipment), tracking from individuals,
and movement via fomites during sample collection were described in studies using surrogate B.
anthracis spores (i.e., Bacillus thuringiensis var. kurstaki) (Van Cuyk et al., 2011; Van Cuyk et
al., 2012). Secondary contamination of B. anthracis spores in an individual's home and vehicle
were reported after drums were made from contaminated African hides in a location separate
from their home (U.S. Centers for Disease Control and Prevention, 2006). During an
investigation of an anthrax outbreak at a textile mill in 1978, one of four sampled vacuum bags
from workers' homes tested positive for B. anthracis, providing evidence for distant transport of
spores via fomites (Bales et al., 2002). Transmission of cutaneous anthrax to children in the
households of mill workers (presumably through contaminated fomites) was involved in 4% of
11
-------
cases assessed in the Gold (1955) review of 117 anthrax cases in the United States between 1933
and 1955.
Oral exposure of spores is most likely to result from the transfer of spores from fomites (i.e.,
contaminated surfaces, clothing) to the receptor's hand and ultimately their mouth (i.e., hand-to-
mouth exposure pathway). Oral exposure may also occur after inhalation of spores and
subsequent mucociliary clearance from the respiratory tract to the esophagus (U.S. Centers for
Disease Control and Prevention, 2010). Dermal exposure may occur through contact with
deposited spores. However, this exposure pathway will not be assessed further due to the lack of
available dose-response data that more closely match the exposure scenario of interest and do not
involve subcutaneous inoculation.
Exposure duration of receptors may be acute, short-term, or possibly subchronic given the
potential persistence of spores. For example, the exposure duration may be acute from a one-time
visit (e.g., 24-hour or less exposure duration) or may be in the form of recurring daily exposure
as could be anticipated after remediation in a residential or occupational land use. However,
there may be peak exposures resulting in relatively high doses acutely or intermittently over time
depending on receptor activities, environmental conditions, and spore particle characteristics.
Exposure via inhalation or ingestion of spores can result in lethal systemic anthrax illness, with
inhalation anthrax having a significantly higher degree of lethality, even with aggressive medical
treatment. Lethal inhalation anthrax has been associated with both low- and high-dose inhalation
exposures, though this exposure scenario is focused on the assessment of low-dose exposure.
Adult and child receptors are susceptible to inhalation anthrax after inhalation exposure to spores
or to gastrointestinal or oro-pharyngeal anthrax (also termed oral-pharyngeal anthrax) after oral
12
-------
exposure. There is also anecdotal evidence and limited in vitro evidence for the presence of
sensitive subpopulations (e.g., elderly, immune-compromised, toxin-sensitive) that may be more
susceptible to anthrax illness than the general population (Inglesby et al., 2002; Canter, 2005;
Martchenko et al., 2012). Complicated by the low number of published reports on anthrax illness
in pregnant, postpartum, or lactating women, Meaney-Delman et al. (2012) noted preliminary,
though not statistically significant, evidence that cutaneous anthrax may pose the potential for
greater lethality than might be expected in the general non-pregnant population. However,
potential confounding factors were also identified that might explain the observed higher death
rates including lack of timely treatment, type of medical treatment, and location of the cutaneous
lesion (Meaney-Delman et al., 2012).
Direct person-to-person transmission of B. anthracis illness was not identified during a review of
49 anthrax investigations conducted by the U.S. Centers for Disease Control and Prevention
(CDC) between January 1950 and August 2001 (Bales et al., 2002). Anthrax retransmission was
also not described during the 2001 anthrax letter event (Inglesby et al., 2002). Though extremely
rare, transmission of cutaneous anthrax infection has resulted from direct contact with infectious
lesions, contaminated dressings, and contact with a bath item contaminated by an infected
individual (Weber and Rutala, 2001). Published evidence for maternal-to-fetal transmission was
described in case reports of neonatal anthrax illness and was accompanied by anthrax bacilli
identified in organs from fetal and neonatal autopsies (Meaney-Delman et al., 2012).
13
-------
5 Effects Assessment
In the effects assessment element of the risk assessment process, the potential health effects and
endpoints of microbial exposure are identified in conjunction with known relationships between
the exposures as described by the exposure assessment and the likelihood of health effects for
those exposures. Section 5.1 identifies and evaluates available data to conduct a hazard
identification to appropriately inform a site-specific effects
analysis for B. cmthracis. Section 5.2 then builds upon the hazard
identification to provide further detail on the inhalation anthrax
disease pathogenesis. Section 5.3 describes and evaluates
available processes to perform a microbial risk assessment.
5.1 Hazard Identification
The hazard identification identifies the type of health hazard
posed by the potentially complete exposure pathways identified in
the CSM. As further detail to accompany the hazard
identification, a key event identification and description of the
disease pathogenesis of inhalation anthrax is provided in Section
5.2.
The microbial pathogen B. cmthracis exists in two forms:
vegetative bacterium and spore. For B. anthracis, inhalation
exposure of the spore form and associated pathogenic illness is
the human health hazard of greatest concern. Historically, the
spore form has been of greatest human health concern due to its
Summary of Findings for
Hazard Identification
•
The hazard posed by
exposure to B. cmthracis
spores is well
documented.
•
Inhalation anthrax poses
the greatest threat of
lethality because it is
difficult to diagnose
during early stages of
illness and becomes
rapidly lethal.
•
There is considerable
uncertainty in the
mechanistic details of the
disease process.
•
There is not a clear link
between mechanistic
pathway(s) or tissue
dose(s) associated with
the lethality endpoint.
•
There is uncertainty
regarding the mechanistic
process for the initiation
of the infection.
14
-------
persistence in indoor or outdoor environments, demonstrated lethality if infection results from
human inhalation exposure, and prior use in biological terrorism. Vegetative bacteria released to
the environment are generally less of a threat due to their limited persistence and low likelihood
of infection unless directly introduced to the bloodstream (Fisher et al., 2011). There are very
limited published data on the infectious dose (ID) associated with inhalation anthrax illness and
the majority of collected data are for the lethality endpoint.
Complete human exposure pathways with B. anthracis spores associated with anthrax illness
include agricultural contact with livestock, recreational contact with wildlife, associated fomites
from livestock or wildlife (e.g., soil, meat, leather, wool or hair, bone meal) (Shadomy and
Smith, 2008), and occupational contact with contaminated animal products (e.g., woolen textile
mill) (Brachman et al., 1960). Prior to the 2001 anthrax letter event, approximately 80% of
anthrax illness in the United States was associated with industrial contact with contaminated
materials and 20% was associated with agricultural exposure (Brachman, 1984). For those
exposed occupationally, the primary risk factor for anthrax illness was contact with contaminated
goat hair from Iran, Iraq, India, or Pakistan (Coleman et al., 2008). Incidental contact with
contaminated animal products (e.g., shaving brush bristles, yarn, animal hide drums, bone meal)
is associated with anthrax illness but tends to be extremely rare (Vaswami, 1955; Suffin et al.,
1978; U.S. Centers for Disease Control and Prevention, 2010; Marston et al., 2011). Two
releases of manufactured B. anthracis spores have resulted in human anthrax disease outbreaks:
the accidental release of spores manufactured by a former Soviet Union bioweapons facility in
Sverdlovsk in 1979, and the anthrax letter event in the United States in 2001.
15
-------
The four types of anthrax illness are differentiated based on the route of exposure associated with
the initiation of infection: inhalation exposure (i.e., inhalation anthrax), oral exposure (i.e.,
gastrointestinal anthrax or intestinal anthrax, oro-pharyngeal anthrax), dermal exposure (i.e.,
cutaneous anthrax), and injection exposure (i.e., injection anthrax) from subcutaneous,
intramuscular, or intravenous injection of drugs contaminated with B. anthracis spores (Inglesby
et al., 2002; Grunow et al., 2013). With the exception of the deliberate release of manufactured
spores, anthrax illness is relatively rare in developed countries and most often results from
contact with infected animals or contaminated animal products (Passalacqua and Bergman,
2006).
Anthrax illness has been described as having three phases: asymptomatic or incubation,
prodromal or latent with nonspecific flulike symptoms, and fulminant with "severe symptomatic
disease" (Bravata et al., 2006). Fulminant anthrax infection is characterized by the development
of overt clinical symptoms resulting from bacteremia and subsequent systemic dissemination of
bacteria and associated toxins. These symptoms can include respiratory distress (i.e., dyspnea,
stridor, cyanosis leading to mechanical ventilation after respiratory failure) and shock (Holty et
al., 2006). Meningoencephalitis is present in up to 50% of human fulminant inhalation anthrax
cases reviewed in Holty et al. (2006). Though each type of anthrax illness can progress to a
fulminant infection, inhalation anthrax poses the greatest threat of lethality because it is difficult
to diagnose during early stages of illness and becomes rapidly lethal after development of severe
symptoms (Inglesby et al., 2002). Even with modern medical treatment and early diagnosis, the
case fatality rate of those with inhalation anthrax during the 2001 anthrax letter event was 45%
(Inglesby et al., 2002). However, the fatality rate is generally estimated to be almost twice as
high without antibiotics or intensive medical treatment (Inglesby et al., 2002; Hilmas et al.,
16
-------
2009). In the United States, 32 cases of inhalation anthrax were reported from 1900 through
2005 (Holty et al., 2006). Slightly more than half of the cases resulted from sources of
manufactured spores or contaminated animal products. Eleven cases were associated with the
2001 anthrax letter event, five occupational cases were associated with the Manchester goat hair
processing plant outbreak in 1957, and one case in 1966 from a man working across the street
from the Manchester plant almost a decade after the 1957 outbreak (Holty et al., 2006). From
2006 through 2013, two additional cases of inhalation anthrax were reported in the United States
(U.S. Centers for Disease Control and Prevention, 2006; Griffith et al., 2014).
There are two forms of anthrax illness associated with oral exposure: gastrointestinal and oro-
pharyngeal. The fatality rate for identified cases of gastrointestinal anthrax ranges from 25% to
60% (U.S. Centers for Disease Control and Prevention, 2000), though it is unknown to what
extent the estimate may be biased high from overrepresentation of more clinically apparent
and/or more severe cases. In a similar fashion to inhalation anthrax, early diagnosis of
gastrointestinal anthrax can be difficult due to non-specific disease symptoms (Cote et al., 2011).
Oro-pharyngeal anthrax generally presents in a milder form and is associated with lower fatality
levels than gastrointestinal anthrax (Hilmas et al., 2009). Case fatality rate estimates for
gastrointestinal and oro-pharyngeal anthrax have high uncertainty as these forms of illness are
likely to be both underreported and present as a "spectrum" of severity levels ranging from
subclinical to lethal illness (Sirisanthana and Brown, 2002). Anthrax infection following oral
exposure is most typically associated with less developed countries (Weiner and Glomski, 2012);
this may be related to increased exposure opportunities due to differing cultural norms and
routine food safety practices in less developed countries. Historically, a large-scale
gastrointestinal anthrax epidemic of approximately 15,000 people in Saint-Domingue (Haiti)
17
-------
during the 1700s was hypothesized to result from ingestion of uncooked beef, highlighting its
potential for significant foodborne outbreaks (Morens, 2002).
In the United States, gastrointestinal anthrax in an occupational setting has been reported co-
incident with cutaneous anthrax, with hand-to-mouth contact of spore-contaminated materials
identified as a potential route of exposure (MacDonald, 1942). Gastrointestinal anthrax was
suspected after ingestion of contaminated meat, though anthrax was not clinically confirmed in
the Minnesota family event in 2000 (U.S. Centers for Disease Control and Prevention, 2000).
Gastrointestinal anthrax in one individual in the United States was also reported after use of a
contaminated animal hide drum (U.S. Centers for Disease Control and Prevention, 2010).
Hypothesized pathways of exposure of the drum user included inhalation and subsequent
ingestion of airborne spores, ingestion of food that had been contaminated by individuals that
previously contacted spores, ingestion of food contaminated by direct deposition of aerosol, and
incidental hand-to-mouth contact after spore contact (U.S. Centers for Disease Control and
Prevention, 2010). However, the absence of gastrointestinal anthrax in laboratory animals after
oral challenge with very large doses of B. anthracis spores has led to the hypothesis that
infection from the oral route may require exposure to significant amounts of vegetative bacteria
(e.g., ingestion of undercooked contaminated meat) (Inglesby et al., 2002; Xie et al., 2013). Host
conditions may predispose individuals to infection even at lower doses where others may be
unaffected (U.S. Centers for Disease Control and Prevention, 2010).
Cutaneous anthrax currently accounts for approximately 95% to 99% of all reported human cases
of anthrax illness worldwide (Shadomy and Smith, 2008), with reported lethality rates of
approximately 1% with antibiotic treatment and 10 to 20% without treatment (Beatty et al.,
18
-------
2003). Eleven of the 22 cases of anthrax illness during the 2001 anthrax letter event were
suspected or confirmed to be cutaneous (Inglesby et al., 2002). A 7-month old infant developed
cutaneous anthrax after contact with B. anthracis contamination during the 2001 anthrax letter
event that later resulted in severe systemic illness with hemolytic anemia, renal involvement, and
persistent hyponatremia (Freedman et al., 2002). This constellation of symptoms appears to be
unique relative to other descriptions of cutaneous anthrax in children, as well as the development
of severe systemic symptoms after timely treatment with antibiotics and corticosteroids
(Freedman et al., 2002). Children who develop cutaneous anthrax typically respond very well
with appropriate treatment, but the severity of presentation in this case is atypical (Freedman et
al., 2002). However, it is unknown how much of the literature describing cutaneous anthrax
includes consideration of cases in children less than one year of age.
First described in 2000, injection anthrax is a relatively new phenomenon for human exposure
and subsequent anthrax infection (Grunow et al., 2013). This form has only been identified in
European countries to date and it has been hypothesized that all cases over the past decade may
have resulted from a common contamination source in heroin (Grunow et al., 2013). The case
fatality rate for injection anthrax is estimated to be 30% (Grunow et al., 2013).
Long-term health impacts, also termed sequelae, have been associated with infectious disease for
a number of pathogens. For example, the toxins produced by some bacteria can cause serious
organ damage in those infected (e.g., kidney damage from Escherichia coli infection) (Food and
Agriculture Organization and World Health Organization (FAO and WHO), 2003).
Alternatively, post-infection response to infectious disease can include the development of auto-
19
-------
immune diseases such as reactive arthritis and Guillain-Barre syndrome (Food and Agriculture
Organization and World Health Organization (FAO and WHO), 2003).
The potential for long-term sequelae from inhalation or gastrointestinal anthrax infection is
unknown. Opportunities to conduct studies on the potential long-term health effects associated
with surviving inhalation anthrax have been extremely limited due to the rarity of cases and
survival after the illness. Reissman et al. (2004) assessed the presence of long-term health effects
from bioterrorism-related B. anthracis infection in an adult study population that survived either
inhalation anthrax or cutaneous anthrax. The study took place one year after illness from the
2001 anthrax letter event in the United States. Survivors reported somatic symptoms associated
with multiple body systems, psychological distress, poor life adjustment, and reduced
functioning (Reissman et al., 2004). However, the confounding of bioterrorism-related exposure
with anthrax illness limits the ability to draw conclusions solely attributable to anthrax illness.
Reissman et al. (2004) noted that their results were supportive of other studies with the United
States population that identified both physical and mental health problems associated with
surviving a terrorism event.
5.2 Disease Pathogenesis in the Context of Key Events
A key events analysis provides the analytical framework and structure to evaluate host-pathogen
interactions from exposure through response (Buchanan et al., 2009). The base assumption of the
key events approach is that a series of "causally linked biochemical or biological key events" can
describe the process from initial exposure through the endpoint of interest (Meek et al., 2014).
Though originally developed for chemical dose-response analysis, a key events framework for
the food-borne pathogen Listeria monocytogenes was generated to assist in the development of a
20
-------
dose-response relationship (Buchanan et al., 2009). Using the general approach described by
Julien et al. (2009) and Buchanan et al. (2009), a preliminary key events process for inhaled B.
anthracis spores was generated for discussion purposes by Hines and Comer (2012). While the
motivation for the B. anthracis key events description was to facilitate identification of data gaps
in the disease process, it provides a useful framework to organize the presentation of anthrax
pathogenesis data (Figure 5-1). Appendix A provides background on transmission and
pathogenesis elements for biological threat agents with relevance to microbial dose-response
analysis.
Key Event 1: Inhalation and Deposition of Respirable B. anthracis Spore Particles
The first key event in B. anthracis pathogenesis is inhalation and deposition of respirable B.
anthracis spore particles. For the development of inhalation anthrax, spores must be inhalable,
deposit in the respiratory tract, and remain viable to initiate infection. It is traditionally accepted
that the transmission of inhalation anthrax infection is optimized when inhalation exposure
occurs to respirable spore particles that are less than 10 |im, which have a higher deposition
potential in the deeper regions of the lung than larger particles. However, Thomas (2013) notes
that deposition should be evaluated as a "continuum" through the entire respiratory tract, with
the potential for infection recognized along the different tissue types present.
Consistent with other inhaled microbial pathogens, larger particle size doses are generally
associated with presumed infection in the upper versus lower portions of the respiratory tract
(Thomas, 2013). Higher doses for lethality are hypothesized to result from higher levels of
clearance in the upper respiratory tract and tissue-specific colonization features (Thomas, 2013).
Particle clearance capabilities in the upper respiratory tract also favor movement of particles to
21
-------
the gastrointestinal tract (Thomas, 2013). In the murine model, gastrointestinal involvement was
only identified in mice challenged with 12 |im particles, but not with 1 |im particles (Thomas et
al., 2010). In this same murine model, the inhalation challenge with 12 |im particles was also
22
-------
Key Event 1
A
Key Event 2
1
Key Event 3
Key Event 4
I
Inhalation &
Deposition of
Respirable
Bacillus anthracis
Spores
\m
Spore
Germination
and Survival
of Vegetative
Trojan Horse
Model
Phagocytic Celts
Jailbreak Model
Lymphoid Tissue
Jailbreak Model
Epithelial Tissue
*
Vegetative Growth &
Toxin Production in
Mediastinal Lymph
Node, Bacteria Enter
Circulatory System
fly
i
Trojan Horse Model
Jailbreak Model
Phagocytic Cells
Lymphoid Tissue
Epithelial Tissue
• Spore Intake by
• Spore Deposition in
• Spore Intake by Lung
Alveolar Macrophage or
Lymphoid Tissue
Epithelial Cell
Dendrite
¦ Extracellular
• Germination & Bacterial
• Germination & Bacterial
Germination & Bacterial
Survival, Toxin
Survival, Toxin
Survival, Toxin
Production, Transport
Production, Refease
Production, Disruption
to Interstitium and
During
of Endothelial Barrier,
Lymph or Vasculature
Transport/Arrival at
Transportto Lymph
Lymph Node
Node
I
Figure 5-1. Key events determination for inhalation anthrax modified from Hines and Comer (2012).
-------
associated challenge with 12 |im particles was also associated with longer average time-to-death
measures than 1 |im particles (i.e., 161 ± 16.1 h versus 101.6 ± 10.4 h, respectively) (Thomas et
al., 2010).
Consistent with other inhaled microbial pathogens, larger particle size doses are generally
associated with presumed infection in the upper versus lower portions of the respiratory tract
(Thomas, 2013). Higher doses for lethality are hypothesized to result from higher levels of
clearance in the upper respiratory tract and tissue-specific colonization features (Thomas, 2013).
Particle clearance capabilities in the upper respiratory tract also favor movement of particles to
the gastrointestinal tract (Thomas, 2013).
Anthrax infection from inhalation exposure to larger particle sizes is associated with larger
reported median lethal dose (LD50) values and differences in disease presentation for nonhuman
primate and guinea pig challenge studies (Druett et al., 1953; Goodlow and Leonard, 1961). The
exposure of nonhuman primates to particle sizes greater than 10 |im has been associated with
disease initiation in the upper respiratory tract, as evidenced by edema of the face and head for
days prior to death from anthrax illness (Druett et al., 1953). Similar presentations of human
anthrax infection were reported where infection was identified in the upper, but not lower,
respiratory tract (Thomas, 2013). In the human, a limited number of case reports have been made
of anthrax infection with clear indications of upper respiratory tract infection, but without any
typical manifestations in the lower respiratory tract (Thomas, 2013).
Key Event 2: Spore Germination, Proliferation, and Movement to Bloodstream
The second key event for pathogenesis is spore germination and vegetative proliferation,
ultimately leading to the release of vegetative bacteria to the bloodstream. Spore germination
24
-------
leading to vegetative proliferation is indicative of infection. Two models have been developed to
conceptualize the initiation of B. anthracis infection from inhalation exposure: the Trojan horse
model (Guidi-Rontani, 2002) and the jailbreak model (Weiner and Glomski, 2012). The Trojan
horse model is the first and most frequently cited model for initiation of inhalation anthrax since
its publication in 2002 (Weiner and Glomski, 2012). Most of the early in vitro mechanistic work
cited in the initial description of the Trojan horse model used the murine animal model or
murine-derived cell lines (Hanna et al., 1993; Guidi-Rontani et al., 1999; Dixon et al., 2000),
though Shafa et al. (1966) evaluated macrophages from the rabbit. The Trojan horse model
hypothesizes the establishment of inhalation anthrax infection as an intracellular competition
between the B. anthracis spore, host macrophage, and toxins expressed by vegetative
B. anthracis (Guidi-Rontani, 2002). The Trojan horse model implicates both lethal toxin (LT)
and edema toxin (ET) in the initiation of infection.
In the Trojan horse model, infection is initiated through engulfment of the spore by an alveolar
macrophage and subsequent spore germination either during transport to, or upon arrival in, the
lymph node (Guidi-Rontani, 2002). Consistent with the Trojan horse model, lung-associated
lymph nodes were identified as the primary location of germination in rabbits after
bronchoscopic administration1 of spores (Lovchik et al., 2012). However, the murine (Glomski
et al., 2007; Sanz et al., 2008; Dumetz et al., 2011) and guinea pig models (Twenhafel, 2010)
provide preliminary evidence that transport to or through a regional lymph node may not be
necessary for spore germination and anthrax illness after inhalation exposure.
1 Bronchoscopic administration likely precludes initiation of infection in the upper respiratory tract and nasal-
associated lymphoid tissue (NALT).
25
-------
After the Trojan horse model was published, additional phagocytic cell types capable of
transporting B. anthracis spores to lymph nodes were identified through in-vitro studies of
human dendritic cells2 (Brittingham et al., 2005) and murine B cells (Rayamajhi et al., 2012).
Spore germination outside of phagocytic cells in a murine animal model after inhalation and oral
exposure was reported in the lymphoid tissue of the respiratory tract and Peyer's patch tissues of
the intestine, respectively (Glomski et al., 2007; Lowe et al., 2013). Spore translocation into lung
epithelial cells was also reported from an in vivo murine study, providing a hypothetical direct
intracellular route for spores to the lymphatic system (Russell et al., 2008).
To accommodate these new data, the jailbreak model expanded the Trojan horse model for
initiation of infection in three important ways: (1) increased emphasis on the host-pathogen
interactions in lymphoid and epithelial tissues, (2) broadened the role of alveolar macrophages to
include important elements of host defense, and (3) expanded the number of potential cellular
carriers to initiate infection (Weiner and Glomski, 2012). The jailbreak model is unique because
it provides a conceptually consistent approach to model the early stages of infection across the
three natural routes of exposure: inhalation, gastrointestinal, and cutaneous anthrax (Weiner and
Glomski, 2012). Multiple pathways by which inhalation anthrax may be initiated from the same
route of exposure were identified (Weiner and Glomski, 2012). The use of multiple distinct
pathways for infection would not be unique to B. anthracis as multiple pathways have been
identified for other microbial pathogens (e.g., salmonellae, shigellae, Listeria monocytogenes)
(Weiner and Glomski, 2012). Lowe et al. (2013) has also clarified that the identification of
multiple pathways does not imply that mediastinal lymph node-initiated infections are not
2 Dendritic cells were identified in the original description of the Trojan horse model as possibly providing an
additional vehicle for transport to the lymphatic system and subsequent germination location (Guidi-Rontani, 2002).
26
-------
occurring, but that alternative or additional pathways may not be recognized without study
approaches designed to capture the data.
New concepts introduced in the jailbreak model include the potential for extracellular
germination of spores that does not require an intracellular phagocytic location for germination,
while still allowing for subsequent transport to the lymph system (Weiner and Glomski, 2012).
The differing role for toxins in early infection is also notable. In the jailbreak model, spores
germinate in an extracellular environment and toxins are necessary to damage the integrity of
cellular barriers to facilitate access to the lymph system (Weiner and Glomski, 2012). In contrast,
toxins in the Trojan horse model facilitate successful intracellular germination through
modulation of host defenses in the phagocytic cell (i.e., the oxidative burst process) (Weiner and
Glomski, 2012).
Key Event 3: Vegetative Proliferation Leads to Measurable Bacteremia and Toxemia
The establishment of anthrax infection requires the successful germination of spores in a host
environment that is conducive for proliferation and dissemination of vegetative bacteria to the
bloodstream (Guidi-Rontani, 2002). Systemic infection then allows for continued bacterial
proliferation in blood and tissue, toxin production, and other virulence factors that are necessary
for potential development of fulminant anthrax.
Lowe et al. (2013) hypothesized that the host environment for germination and growth may have
downstream effects on the dissemination pattern of systemic infection. Similarities in the
terminal bacterial burden in organs, but varying numbers of bacteria and differing kinetics of
release based on the initial site of spore germination (e.g., lymphoid tissue versus phagocytes in
27
-------
draining lymph node) were identified in murine studies (Lowe et al., 2013). Equivalent studies
have yet to be conducted in the rabbit and nonhuman primate.
Dissemination allows for the vegetative bacteria to be presented to new host environments
relative to the initial environment(s) associated with germination and initial proliferation. An in
vitro evaluation of the germination of B. anthracis Sterne spores and proliferation of vegetative
bacteria in rabbit, nonhuman primate, and human sera found the rabbit sera to be the most
hospitable proliferation medium relative to the nonhuman primate and human (Bensman et al.,
2012). Interestingly, the same in vitro study reported differences in the species sera most
hospitable to germination, with spore germination highest in nonhuman primate sera, moderate
in human sera, and only limited in rabbit sera (Bensman et al., 2012).
Few inhalation anthrax datasets for the rabbit report survival after measurable bacteremia.
Survival without medical treatment after development of anthrax bacteremia was reported in two
unvaccinated animals in the multiple-dose, low-dose rabbit study (U.S. Environmental Protection
Agency, 2012b). Fellows et al. (2001) also reported survival after anthrax bacteremia when
vaccinated rabbits were challenged with isolated strains from diverse geographic locations.
Incidence of bacteremia for two isolates were reported as 70% and 80%, with accompanying
survival rates of 90% and 100%, respectively. However, bacteremia levels were relatively low
(i.e., <100 CFU/mL) (Fellows et al., 2001).
In contrast, survival after measurable low-level bacteremia was reported more often for
unvaccinated nonhuman primates (Albrink and Goodlow, 1959; Saile et al., 2011; Henning et al.,
2012) and vaccinated nonhuman primates (Ivins et al., 1996; Ivins et al., 1998; Fellows et al.,
2001). Consistent with reports for the vaccinated rabbit, the levels of bacteremia were low (i.e.,
28
-------
<100 to 200 CFU/mL) in the vaccinated nonhuman primate. From a key events perspective, the
presence of measurable bacteremia appears to be strongly correlated with development of lethal
anthrax infection, but in itself is not 100% predictive.
Bacteremia provides for a significant toxin loading to develop due to the upregulation of toxin
production by vegetative bacteria (Cote et al., 2011). The LT and ET anthrax toxins may be
released through extracellular vesicles containing toxin or in association with the capsule (Ezzell
et al., 2009). Host cell proteins are receptors for the toxins, with differential expression of these
proteins in cell lines associated with varying levels of cellular lethality when exposed to anthrax
toxin (Martchenko et al., 2012). Each toxin affects cell signaling pathways that are present
throughout the body in almost every cell type (Moayeri and Leppla, 2009). As a result, the
response to the toxin is varied and dependent on the exposure and dose of exposed cells and
tissues.
The LT is a zinc metalloproteinase that affects the mitogen-activated protein kinase kinases
(MAPKKs) that are critical to many diverse cellular functions (Moayeri and Leppla, 2009). The
ET is a calmodulin-dependent adenylate cyclase that produces cyclic 3',5'-adenosine
monophosphate (cAMP), a compound also capable of affecting cellular signaling pathways
(Moayeri and Leppla, 2009). The level of cooperative action of the toxins is a current area of
uncertainty. The anthrax toxins have been described to work in an "additive or synergistic"
fashion when both toxins are present (Lovchik et al., 2012), with the potential for "cooperative"
action of the two toxins also reported for in vitro cellular studies using murine dendritic cells
(Tournier et al., 2005). Recent reviews should be consulted for more detailed information on
29
-------
toxins and toxin action (Tournier et al., 2007; Moayeri and Leppla, 2009; Guichard et al., 2012;
Lowe and Glomski, 2012).
Key Event 4: Development of Fulminant Infection
Fulminant anthrax is associated with a presentation of "severe symptomatic disease" that can
rapidly progress to severe respiratory distress, shock, and death (Bravata et al., 2006). Terminal
bacteremia (i.e., vegetative bacteria in bloodstream) can be extremely high relative to other
microbial pathogens, with levels of 109 CFU/mL reported in the nonhuman primate (Friedlander
et al., 1993). More typical reported values for the nonhuman primate range from 106 to 108
CFU/mL, with published examples including Ivins et al. (1996) and Ivins et al. (1998). In the
rabbit animal model, terminal bacteremia concentrations were identified in the range of 105 to
107 CFU/mL in the single-dose study and 101 to 105 CFU/mL in the multiple-dose study (U.S.
Environmental Protection Agency, 201 la, 2012b). However, there were also animals in each
study that died with anthrax-illness related symptoms but no measureable bacteremia
concentrations (U.S. Environmental Protection Agency, 2011a, 2012b). In contrast, toxemia can
be more variable in its presentation from the appearance of symptoms to death, with non-
detection even in animals that die with symptomatic disease.
While the action of toxins in the early stages of anthrax illness is thought to affect the
functioning of phagocytic cells, the systemic accessibility of toxins in the later illness stages
provides for the expression of widespread and tissue-specific toxicity. However, there is
considerable uncertainty in known connections between the cell type and the pathway(s)
associated with the toxicity (Moayeri and Leppla, 2009). To date, there are no mechanistic
pathway(s) or tissue dose(s) that can be definitively associated with the lethality endpoint.
30
-------
There is also in-vitro evidence for non-toxin mediated virulence factors that may be associated
with lethality. Lethality in the rabbit resulted from intravenous challenge with B. anthracis
Vollum strain vegetative bacteria mutants that lost production of toxins (Levy et al., 2014).
However, additional non-toxin virulence factors that are hypothesized to contribute to anthrax
lethality include sepsis from high bacteremia levels, proteases, B. anthracis S-layer protein A
(BslA), and other factors yet to be identified (Friedlander, 2001; Guichard et al., 2012; Weiner
and Glomski, 2012; Coggeshall et al., 2013; Remy et al., 2013). The sepsis hypothesis has
received the most attention to date. The hypothesis acknowledges the role of toxins in reducing
immune system effectiveness, but associates lethality with the extremely high bacteremia levels
of fulminant illness (Stearns-Kurosawa et al., 2006; Coggeshall et al., 2013). Alternately, Cote et
al. (2011) recognized the high terminal bacteremia concentration and hypothesized that host
death resulted from a combination of toxemia and additional virulence factors.
5.3 Overview of Microbial Dose-Response Analysis
Dose-response analysis evaluates the relationship between exposure and the likelihood of
identified health effects or outcomes (U.S. Environmental Protection Agency, 2014a). The
resulting dose-response relationship is then compared to the results of the exposure assessment to
determine the likelihood of adverse effects. There are three main steps in the development of a
microbial dose-response relationship: (1) evaluation of microbial dose-response data, (2)
modeling the dose-response relationship, and (3) conducting interspecies extrapolation to a
human equivalent dose (HED) (Table 5-1). Table 5-1 identifies the key questions associated with
each main step and the report section where data to evaluate the key questions are presented. The
evaluation of data to answer the key questions is guided by current microbial dose-response
analysis practice and data describing B. anthracis pathogenesis. As additional information to
31
-------
supplement Section 5.1, Appendix B provides a review of historical themes in modeling B.
cmthrcicis dose-response relationships.
Table 5-1. Development of Microbial Dose-Response Relationships
Steps in Microbial Dose-
Response Analysis
Key Questions
Report Section
Evaluate the microbial
dose-response data
(Section 5.4)
What animal models are appropriate to
generate dose-response data for B.
anthracis'?
Section 5.4.1 Animal Model Selection
Using Concordance of Pathology
What dose-response data are available and
of sufficient quality to generate a dose-
response relationship for B. anthracis?
What endpoints can be evaluated with
available dose-response data?
Section 5.4.2 Identification of
Microbial Dose-Response Data
Model the dose-response
relationship
(Section 5.5)
What dose metrics can be supported based
on available disease pathogenesis and other
dose-response data?
What assumptions are associated with a
given dose metric?
Section 5.5.1 Determination of Dose
Metric
What types of empirical and mechanistic
models may be suitable for B. anthracis?
Can mechanistic models be supported by
available dose-response data for B.
anthracis1
Section 5.5.2 Empirical and
Mechanistic Modeling Approaches
What approaches can be used to
mathematically model the dose-response
relationship and estimate the POD?
Section 5.5.3 Mathematically
Modeling the Microbial Dose-
Response Relationship
Conduct interspecies
extrapolation to a HED
(Section 5.6)
What is a general framework that can be
used for interspecies extrapolation of B.
anthracis1
Section 5.6.3. Proposed Framework
for Interspecies Extrapolation for B.
anthracis
What data for the rabbit, nonhuman primate,
and human are available to evaluate the
kinetics and dynamics of B. anthracis
pathogenesis?
Section 5.6.4 Available Kinetic Data
Section 5.6.5 Available Dynamic Data
How can available data be incorporated in
the extrapolation process?
Section 5.6.6 Summary of
Extrapolation Framework for B.
anthracis
POD — point of departure
HED — human equivalent dose
32
-------
5.4 Evaluate the Microbial Dose-Response Data
The evaluation of microbial dose-response data in this section will consider determination of
appropriate animal models to generate a dose-response relationship relevant for the human and
evaluation of available dose-response data for the appropriate animal models and the human
(Table 5-2).
Table 5-2. Evaluation of Microbial Dose-Response Data
Step in Microbial Dose-
Response Analysis
Key Questions
Report Section
Evaluate the microbial
dose-response data
(Section 5.4)
What animal models are appropriate to
generate dose-response data for B.
anthracis'?
Section 5.4.1 Animal Model Selection
Using Concordance of Pathology
What dose-response data are available and
of sufficient quality to generate a dose-
response relationship for B. anthracis?
What endpoints can be evaluated with
available dose-response data?
Section 5.4.2 Identification of
Microbial Dose-Response Data
5.4.1 Animal Model Selection Using Concordance of Pathology
This section will evaluate suitability of the rabbit and nonhuman primate animal models for the
development of human dose-response relationships for inhalation anthrax. Based on general
similarity in the pathology of the human and the animal models, the rabbit and nonhuman
primate are identified as suitable for inhalation anthrax studies of pathogenesis (Zaucha et al.,
1998; Leffel and Pitt, 2006; U.S. Food and Drug Administration, 2007; Goossens, 2009;
Twenhafel, 2010). While rodent species (e.g., mouse) have been used for studying various
elements of anthrax pathogenesis, potential variation in response to fully virulent strains and
differences in immune system activity may limit the utility of these animal models for broader
applications (U.S. Food and Drug Administration, 2007). Given the relative scarcity of oral
33
-------
dosing studies reporting pathology, an animal model assessment was not conducted for this route
of exposure. Animal model selection should be based on the utility of an animal model to answer
the specific research question(s) being considered (Goossens,
2009). However, a process to assess animal model suitability for
extrapolation to a human B. cmthrcicis dose-response relationship
has not been proposed (Pitt and LeClaire, 2005; Leffel and Pitt,
2006; Coleman et al., 2008).
For this evaluation, the suitability of animal models for
extrapolation to human B. cmthrcicis dose-response relationships
was determined by an assessment of general concordance in
published anthrax pathology between the human and the animal
models. The key human histologic findings for the assessment of
animal models identified by Twenhafel (2010) were used to
assess published anthrax pathology of the rabbit and nonhuman
primate relative to that of the human. The ultimate use (e.g., basic pathogenesis research,
medical countermeasures) of the selected animal models was not specified in Twenhafel (2010).
Twenhafel (2010) evaluated human pathology data from Sverdlovsk (Abramova et al., 1993;
Grinberg et al., 2001) and the 2001 anthrax letter event (Jernigan et al., 2001) to generate the
following list of key human pathological findings: pneumonia; splenic lymphoid depletion;
meningitis; hepatic, gastrointestinal, and urogenital hemorrhage and/or inflammation; anthrax
bacteremia; and anthrax toxemia.
Summary of Findings for
Animal Model Selection
• The rabbit and nonhuman
primate exhibit many
commonalities in the type
of lesions and tissues
identified for inhalation
anthrax in the human.
• Differences were not
identified between the
rabbit and the nonhuman
primate for anthrax
pathology that do not
have a time-dependency
for incidence or severity
in presentation.
• The rabbit and nonhuman
primate are suitable
animal models for
development of dose-
response relationships for
the human.
34
-------
5.4.1.1 End-stage Pathology
The vast majority of published pathology data for the evaluated animal models are representative
of end-stage illness. However, exceptions include a nonhuman primate serial sacrifice study that
evaluated a subset of tissues associated with early infection events (Berdjis et al., 1962) and a
serial pathology study in the rabbit at 30, 60, and 72 hours post-challenge for selected tissues
(Peterson et al., 2007). The pathology reported from scheduled sacrifice studies may also include
animals that have inhalation anthrax in earlier stages of the disease (i.e., not end-stage) and may
therefore introduce early or intermediate disease stages in the described pathology. However,
these occurrences are not specifically identified in reports and therefore cannot be systematically
evaluated.
Comparisons of anthrax pathology provided in published reports can be challenging for many
reasons. Vasconcelos et al. (2003) noted the inherent difficulty in comparisons of pathology
reported in nonhuman primate studies due to fundamental differences in study design and quality
controls (e.g., animal age, B. anthracis strain, dose, particle size, pre-existing lung lesions from
mites). Differences in pathology descriptions and disease definitions also complicate
comparisons of the presence, absence, or severity of identified pathological conditions (Fritz et
al., 1995; Vasconcelos et al., 2003). Additionally, distinguishing between gross versus
histopathologic observations can be challenging based on the limited data reported for some
studies (Fritz et al., 1995). Likewise, the existence of anthrax pathology can be missed for
animals lacking gross lesions typically associated with inhalation anthrax (i.e., atypical disease
presentations) if microscopic examination of tissues is not conducted (Vasconcelos et al., 2003).
The lack of these data could bias the reported data set of inhalation anthrax pathology toward
only the histopathology associated with gross pathological features.
35
-------
Characteristics of inhalation anthrax pathology also affect the comparison of reported study
results. One key consideration is the potential role of time-dependency in lesion development,
whereby lesion progression and anatomical location for specified tissue locations are associated
with survival time post-challenge. For example, defined pathological outcomes (e.g., meningeal
hemorrhage, adrenal inflammation and necrosis, hepatic necrosis) were reported more commonly
in nonhuman primates that survived four or more days post-challenge relative to those that
survived shorter time periods (Vasconcelos et al., 2003). Similarly, the rapidity of death from
inhalation anthrax in the rabbit has been attributed to a decreased incidence and severity of
mediastinal lesions relative to the human, who typically exhibits a longer survival time (Zaucha
et al., 1998). The extension of human survival afforded by medical treatments (e.g., antibiotics,
aggressive medical care) may confound comparisons with animal pathology unless similar
medical treatments are employed to extend the illness duration in the animal model. The
prolongation of survival through the use of antibiotics in later stages of illness without
prevention of death was reported during early studies of nonhuman primates by Gleiser (1967).
As an additional complicating factor, dose-dependency has also been hypothesized to affect
formation of specific lesions (Gleiser et al., 1963). As these factors have relevance for the
comparison of animal model data, they will be considered further in Section 5.4.1.5.
A detailed summary table of end-stage pathology for the rabbit, nonhuman primate, and human
is provided in Appendix C, Data Summary Table for End-stage Inhalation Anthrax Pathology of
the Human, Nonhuman Primate, and Rabbit.
36
-------
5.4.1.2 Human
Available human anthrax pathology data originate from the 1957 anthrax occupational outbreak,
the 1979 Sverdlovsk outbreak, the 2001 anthrax letter event, and anecdotal published case
reports (Table 5-2). There are varying levels of comprehensiveness and detail in the reported
pathology, ranging from complete pathological descriptions to highlights or generalized findings.
Table 5-2 identifies primary sources describing human inhalation anthrax cases. However, there
may be potential for some overlap in descriptions of an individual case. Interpretation of the
published data on the pathology of human inhalation anthrax is complicated by: (1) application
of varying types of medical treatment and (2) cases resulting from exposure to different strains or
spore products (e.g., Ames strain manufactured spore products versus mill aerosol strains).
Table 5-3. Reported Human Autopsy or Pathology Data by Outbreak or Event
Outbreak/Event (Strain)
Reported Data
1957 Occupational Outbreak
(Unknown mill aerosol strain[s])
Albrink et al. (1960)
Plotkin et al. (2002)
1979 Sverdlovsk Outbreak
(Unknown multiple strains)*
Abramova et al. (1993)
Grinberg et al. (2001)
2001 Anthrax Letter Event
(Ames strain)
Barakat et al. (2002)
Borio et al. (2001)
Bushet al. (2001)
Gill and Melinek (2002)
Guarner et al. (2003)
Jernigan et al. (2001)
Mina et al. (2002)
Anecdotal Events
(Unknown strains)
Albrink (1961) - Electrician who worked in microbiology laboratory in an
unspecified year (Unknown strain)
Brachman et al. (1961) - Male with sarcoidosis in 1958 and woman in
1948 (Unknown mill aerosol strain[s])
Gold (1955) - Handyman in carding room of mill in 1942 (Unknown mill
aerosol strain[s])
LaForce et al. (1969) - Worker across alleyway from goat hair processing
plant in 1966 (Unknown mill aerosol strain[s])
Suffin et al. (1978) - Weaver exposed to yarn in 1976 (Unknown multiple
strains associated with animal-origin yarn)
U.S. Communicable Disease Center (1961) - Secretary in goat hair and
wool plant outside Philadelphia in 1961 (Unknown mill aerosol strain[s])
* See Jackson et al. (1998) for more information on Sverdlovsk strains
t Hie 1951 case described as the "housewife" in Brachman et al. (1961) did not include autopsy or pathology data
37
-------
Data on human inhalation anthrax pathology without any medical treatment are not available, as
most individuals receive medical treatment when the severity of illness associated with fulminant
anthrax is exhibited. For example, inhalation anthrax cases in the 1957 occupational outbreak
were given antibiotics at some point prior to final diagnosis or death. This makes it difficult to
fully determine the human pathology without medical treatment. To obtain comparable animal
model pathology data, animal studies would need to incorporate the same types of medical
treatments. The effectiveness of medical treatment for inhalation anthrax has increased
substantially between the earlier outbreaks (e.g., 1957 occupational outbreak, Sverdlovsk) and
the 2001 anthrax letter event outbreak; this has led to higher survival rates and possibly longer
times to death for those that do not survive. However, strain-specific differences in inhalation
anthrax pathology have also contributed to the identified differences. Many of the pre-
Sverdlovsk cases resulted from exposure to unknown strains of animal mill aerosol or finished
product (e.g., yarn) of animal origin, whereas the Sverdlovsk outbreak, the 2001 anthrax letter
event outbreak, and the case in the electrician described by Albrink (1961) resulted from
exposure to Ames or an unidentified manufactured spore product strain(s) (Table 5-3).
Table 5-4 provides a summary of reported human pathology relative to the Twenhafel (2010) list
of key histopathological findings. The two most pronounced gross autopsy findings of human
inhalation anthrax victims were pleural effusions and mediastinal lymph nodes with edema and
hemorrhage (Guarner and del Rio, 2011). Serosanguinous pleural effusions were identified in
five of the eight patients who died during the 2001 anthrax letter event, with the confirmed
presence of B. anthracis antigens in the pleurae thought to explain the reported severity of these
lesions (Guarner et al., 2003). "Massive hemorrhagic mediastinitis" was identified in two of
38
-------
three of the fatal inhalation anthrax cases in the 1957 occupational outbreak reviewed by Plotkin
et al. (2002), with mediastinal lymph nodes described as enlarged and edema-filled.
Notable differences in pathology were described between the victims of Sverdlovsk and the 2001
anthrax letter event, with greater progression of disease reported in Sverdlovsk victims (Guarner
and del Rio, 2011). The first point of difference between the Sverdlovsk and the 2001 anthrax
letter event victims was the relative presence of high- and low-pressure hemorrhages. In the 2001
anthrax letter event, higher pressure hemorrhages were less prominent than in the Sverdlovsk
cases. The second main difference was that the Sverdlovsk victims exhibited extensive
Table 5-4. Summary of Human Pathology Relative to Twenhafel (2010) Key Findings
Pathology
Human
Pneumonia
Pleural effusions (at autopsy or drained prior to death) (LaForce et al., 1969; Jernigan et
al., 2001; Barakat et al., 2002; Mina et al., 2002; Guarner et al., 2003)
Pulmonary edema (Abramova et al., 1993; Mina et al., 2002), including intra-alveolar
and interstitial edema with focal hemorrhage and fibrin deposition (Barakat et al., 2002)
Necrotizing, hemorrhagic pneumonia with primary foci present (Abramova et al., 1993)
Perihilar interstitial pneumonia (Grinberg et al., 2001) and acute bronchial pneumonia
(Grinberg et al., 2001)
Splenic lymphoid
depletion
Splenomegaly with hemorrhage (Albrink et al., 1960), congestion (Suffin et al., 1978),
and necrosis (Barakat et al., 2002; Guarner et al., 2003)
Moderate to marked lymphocytolysis, minimal atrophy of follicles, and thickening of
Bilroth cords (Grinberg et al., 2001)
Meningitis
Meningitis (Inglesby et al., 2002), including hemorrhagic meningitis (Plotkin et al.,
2002)
Cardinal's Cap from hemorrhage of leptomeninges (Inglesby et al., 2002); more
frequently identified from Sverdlovsk than 2001 anthrax letter event victims (Guarner
and del Rio, 2011)
Hepatic hemorrhage
or inflammation
Intrasinusoidal inflammation present (Grinberg et al., 2001)
Kupffer cells mildly to moderately hypertrophic and hyperplastic, minimal to mild
centrilobular, and coagulation necrosis noted infrequently (Grinberg et al., 2001)
Gastrointestinal
hemorrhage or
inflammation
Gastrointestinal submucosal lesions (Abramova et al., 1993; Inglesby et al., 2002)
Necrosis, hemorrhage, and edema of the ileum (Albrink et al., 1960)
Urogenital
hemorrhage or
inflammation
None reported for human in identified sources
39
-------
hemorrhage in the meninges (i.e., Cardinal's Cap) and higher burdens of B. anthracis in the brain
and intestines (Guarner and del Rio, 2011). In contrast to the identification of meningeal spread
in approximately 80% of the Sverdlovsk cases as reported by Grinberg et al. (2001), a
considerably lower case rate of meningitis or post-mortem evidence of meningeal spread was
identified in the 2001 anthrax letter event cases (Guarner et al., 2003). Hypothesized reasons for
these differences included differing B. anthracis strains, earlier case recognition, and more
effective treatment protocols in the 2001 anthrax letter event (Guarner et al., 2003; Guarner and
del Rio, 2011).
Splenomegaly was reported during the 1957 occupational outbreak (Albrink et al., 1960),
anecdotal case reports (Suffin et al., 1978), and the 2001 anthrax letter event (Barakat et al.,
2002; Guarner et al., 2003). Splenic congestion, a condition that can contribute to presentation of
splenomegaly, was also identified in 86% of the 41 cases for which microscopic data were
evaluated in the Sverdlovsk outbreak (Grinberg et al., 2001).
5.4.1.3 Rabbit
Inhalation anthrax pathology for B. anthracis Ames strain exposure has been described for two
rabbit breeds (Table 5-5). Table 5-5 identifies studies reporting pathology of end-stage inhalation
anthrax, with the exception of Peterson et al. (2007). The New Zealand white rabbit is the most
commonly used breed of domesticated rabbit (Oryctolagus cuniculus) for anthrax pathology
studies. Peterson et al. (2007) reported that the pathology in the Dutch-belted dwarf rabbit
resulting from intranasal B. anthracis administration was generally consistent with that identified
for New Zealand white rabbits by Zaucha et al. (1998) and Yee et al. (2010) after aerosol
challenge. An absence of sex-related differences in the development of antigenemia or
40
-------
bacteremia after aerosol challenge was described in the New Zealand white rabbit (Yee et al.,
2010). In a comparison of the pathology resulting from bronchoscopic versus aerosol challenge,
Lovchik et al. (2012) reported that the "typical" histopathology lesions identified were consistent
with those described by Zaucha et al. (1998) and Yee et al. (2010).
Table 5-5. Studies Reporting Inhalation Anthrax Pathology by Rabbit Breed
Rabbit Breed
Study Citation (Strain)
New Zealand White
Rabbit
Lovchik et al. (2012) (Ames)
Peterson et al. (2007)* (Ames)
U.S. Enviromnental Protection Agency (2011a) (Ames)
U.S. Enviromnental Protection Agency (2012b) (Ames)
Yee et al. (2010) (Ames)
Zaucha et al. (1998) (Ames)
Dutch-belted Rabbit
Peterson et al. (2007) (Ames)
* Reports serial sacrifice pathology for up to 72 hours post-challenge, no end-stage pathology
A detailed summary table of end-stage pathology for the rabbit, nonhuman primate, and human
is provided in Appendix C, Data Summary Table for End-stage Inhalation Anthrax Pathology of
the Human, Nonhuman Primate, and Rabbit. Table 5-6 summarizes the published rabbit
pathology relative to the key findings of Twenhafel (2010).
A review describing gross lesions identified in New Zealand white rabbits after aerosol challenge
found blood from the nose, splenomegaly, adrenal gland hemorrhage, hemorrhage in the
mandibular lymph node, and lung edema (Twenhafel, 2010). In the same review, reported
histopathology included interstitial pneumonia, splenitis, and lymphadenitis with destruction of
lymphoid tissues noted in the spleen; mediastinal, mandibular, and mesenteric lymph nodes; and
Peyer patches in the small intestine and sacculus rotundus (Twenhafel, 2010).
41
-------
Table 5-6. Summary of Rabbit Pathology Relative to Twenhafel (2010) Key Findings
Pathology
Key Findings
Pneumonia
No reported pneumonia, but suppurative inflammation in lung (U.S.
Enviromnental Protection Agency, 201 la, 2012b)
Splenic lymphoid depletion
Splenomegaly, with acute fibrinous splenitis (Zaucha et al., 1998; Yee et
al., 2010; Lovchik et al., 2012); necrosis (Zaucha et al., 1998; Yee et al.,
2010; Lovchik et al., 2012); hemorrhage (Zaucha et al., 1998; Lovchik et
al., 2012); lesions (Lovchik et al., 2012)
Lymphocyte depletion (Lovchik et al., 2012)
Meningitis
Meningitis with suppurative inflammation (U.S. Enviromnental Protection
Agency, 201 la); bacilli in meninges (Peterson et al., 2007)
Brain and/or meningeal lesions with no leukocytic infiltrate (Zaucha et al.,
1998)
Hepatic hemorrhage or
inflammation
Pathology not reported after inhalation exposure, one identification after
intravenous dosing in Nordberg et al. (1961)
Gastrointestinal hemorrhage or
inflammation
Hemorrhage, necrosis, and lymphoid depletion in appendix (U.S.
Enviromnental Protection Agency, 2012b)
Edema, hemorrhage, and necrosis in cecum (U.S. Enviromnental Protection
Agency, 2012b)
Urogenital hemorrhage or
inflammation
Ovarian hemorrhage (Zaucha et al., 1998)
Two high-dose studies reported the pathology for New Zealand white rabbits challenged with
single inhaled doses of approximately 107 inhaled CFU of B. cmthrcicis Ames spores (Zaucha et
al., 1998; Yee et al., 2010). Zaucha et al. (1998) is the classic anthrax pathology rabbit study.
The most "prominent" pathology findings reported for the 22 New Zealand white rabbits were
hemorrhage and edema in the spleen, lymph nodes, lungs, gastrointestinal tract, and adrenal
glands (Zaucha et al., 1998). Lesions were typically hemorrhagic, necrotic, and exhibited
minimal localized leukocytic response (Zaucha et al., 1998). Necrosis was reported in the
mediastinal lymph nodes of 100% of the challenged rabbits, in the mandibular lymph nodes of
89% of the challenged rabbits, and in the mesenteric lymph nodes of 59% of the challenged
rabbits (Zaucha et al., 1998). Zaucha et al. (1998) hypothesized that the increased incidence and
severity of lesions in the submandibular node may have been associated with direct
42
-------
oropharyngeal deposition or mucociliary clearance of previously deposited spores lower in the
respiratory tract. Acute mediastinitis was infrequently identified, with lesions noted to be less
severe than in the human (Zaucha et al., 1998). The spleen exhibited necrosis, inflammation,
hemorrhage, and significant lesions (Zaucha et al., 1998). Pathology was also reported for a
single high-dose control group that was identified as generally consistent with that identified by
Zaucha et al. (1998) (U.S. Environmental Protection Agency, 2011a).
Data were also obtained from studies where limited pathology results were reported as part of a
larger study design. Yee et al. (2010) conducted a high-dose exposure study and noted general
pathological concordance with the Zaucha et al. (1998) results. Peterson et al. (2007) described
the pathology identified during serial sacrifices of aerosol-challenged animals at 36 hours (n=3),
60 hours (n=3), and 72 hours (n=l). Histologic lesions, by order of prominence, were present in
the mediastinal lymph node, lungs, spleen, and thymus (Peterson et al., 2007). Lesions exhibited
edema/fibrin, necrosis/depletion, hemorrhage, and differing levels of leukocytic infiltration
(Peterson et al., 2007). Lovchik et al. (2012) reported consistency in the pathological lesions in
rabbits bronchoscopically challenged with lethal doses of B. anthracis with that previously
described in the rabbit by Zaucha et al. (1998) and Yee et al. (2010) for aerosol challenges.
Pathology from low-dose B. anthracis aerosol challenge studies was also reported for the New
Zealand white rabbit (U.S. Environmental Protection Agency, 201 la, 2012b). An acute single
low-dose study with a challenge dose of approximately 102 to 105 inhaled CFU was conducted
(U.S. Environmental Protection Agency, 201 la). A follow-on study using a similar design that
incorporated multiple doses of approximately 102 to 104 inhaled CFU per day for 15 days was
43
-------
then performed (U.S. Environmental Protection Agency, 2012b). The challenges took place
Monday through Friday; there were no weekend challenges.
Gross and microscopic pathology reported for both studies was concordant with Zaucha et al.
(1998), with gross lesions correlated with histological findings of hemorrhage, necrosis, edema,
and suppurative inflammation (U.S. Environmental Protection Agency, 2012b).
One pathological finding of interest was the identification of granulomas/pyrogranulomas in one
individual (Rabbit 38) of the U.S. Environmental Protection Agency (2012b) multiple-dose
study. In the single-dose study, multinucleated giant cells were reported as tending toward
formation of granulomas, though no actual granulomas were identified. One interpretation for the
presence of the granuloma or pre-granulomas was that the removal of organic debris (e.g., food
particles or hair and debris from vascular access ports) (Taketoh et al., 2009) was impaired by
systemic macrophage dysfunction that can be associated with high levels of bacteremia and
associated sepsis (U.S. Environmental Protection Agency, 2012b). However, the
pathophysiological data for the rabbit did not include signs indicative of fulminant anthrax
necessary to induce sepsis (i.e., showed elevation in telemetry parameters with abnormality only
in the respiratory rate, single low positive bacteremia sample). There is one other pyrogranuloma
reported in the literature relating to inhalation anthrax and it was described in a vaccinated
animal that survived inhalation anthrax (U.S. Food and Drug Administration, 2002).
Interestingly, the pulmonary lesions reported by Gleiser et al. (1968) were consistent with the
characteristics of an early granuloma and were identified in animals thought to be innately
resistant to inhalation anthrax infection. In this context, the granuloma may simply be a non-
specific indicator of a vigorous host response to a bacterial challenge.
44
-------
However, the use of a venous access port in the U.S. Environmental Protection Agency (2012b)
and U.S. Environmental Protection Agency (201 la) studies may provide an additional
confounding factor to the interpretation of the granuloma in the multiple-dose study and the early
stage granulomas described in the single-dose study. Granulomas were reportedly associated
with the use of venous access ports in studies of rats (Taketoh et al., 2009); however, the study
did not have a control group for statistical comparison. Accordingly, further study using a fully
virulent low-dose B. anthracis spore strain without the inclusion of confounding factors (e.g.,
venous access port, vaccination status) will be necessary before the granuloma can be attributed
to its proper cause.
5.4.1.4 Nonhuman Primate
Published pathology data from inhalation exposure to B. anthracis were identified by nonhuman
primate species, B. anthracis strains, and sources (Table 5-7). With the exception of Berdjis et al.
(1962), who used a serial sacrifice study design, the identified reports describe end-stage
pathology from inhalation of B. anthracis aerosols. Reported pathology outcomes from studies or
treatment groups that included medical treatments (e.g., anti-toxins, antibiotics) or other
treatment protocols were not included in the summary pathology table in Appendix C. An
example of data from a treatment protocol would include the pathology reported from penicillin-
treated monkeys in Gochenour et al. (1962).
45
-------
Table 5-7. Studies Reporting Inhalation Anthrax Pathology by Nonhuman Primate Species
and Strain
Nonhuman Primate (Species)
Study Citation (Strain)
Chimpanzee (Pan troglodytes)
Albrink and Goodlow (1959) (Vollum rB)
Rhesus Monkey
(Macaca mulatto)
Berdjis et al. (1962)^ (Vollum-189)
Gochenour et al. (1962) t (Vollum-189)
Friedlander et al. (1993) (Vollum IB strain)
Fritz et al. (1995) (Vollum IB strain. Ames strain)
Gleiser et al. (1963) t (Vollum-189)
Cynomolgus Macaque
(Macaca fascicularis)
Brachman et al. (1966) (Goat Hair Mill Aerosol, Unknown Strain[s])
Dalldorf et al. (1971) (Goat Hair Mill Aerosol, Unknown Strain[s])
Hcnning et al. (2012) (Ames)
Vasconcelos et al. (2003) (Ames)
African Green Monkey
(Chlorocebus aethiops)
Twenhafel et al. (2007) (Ames)
Common Marmoset
(Callithrix jacchus)
Lever et al. (2008) (Ames)
* Serial sacrifice pathology reported Days 1 through 6, no end-stage pathology reported
t Originating technical report for papers is Gochenour (1961)
i Papers report pathology from same study of nonhuman primate exposure to goat hair mill aerosol in South Carolina, originating
technical report for papers is Dalldorf and Kaufman (1966)
In the 1950s through the 1960s, anthrax studies by the U.S. Army laboratories (predecessors of
the current U.S. Army Medical Research Institute of Infectious Diseases [USAMRIID]
laboratories) typically used the cynomolgus monkey (Macacafascicular!s) (U.S. Food and Drug
Administration, 2002). There was one published study reporting pathology after exposure to goat
hair mill aerosols of unknown strain(s) that used the cynomolgus monkey (Brachman et al.,
1966; Dalldorf et al., 1971). The rhesus monkey (Macaca mulatto) was also used in the 1960s in
controlled exposure laboratory studies with the Vollum-189 strain (Berdjis et al., 1962;
Gochenour et al., 1962; Gleiser et al., 1963). During the resurgence period of anthrax research
from 1990 through 2000, the rhesus monkey was the most commonly used species until the
rhesus monkey became increasingly expensive and difficult to access (Twenhafel et al., 2007).
Since that time, additional nonhuman primate species were evaluated including the African green
monkey (Chlorocebus aethiops) and common marmoset (Callithrix jacchus) (Lever et al., 2008;
46
-------
Twenhafel, 2010), while the cynomolgus monkey also experienced a resurgence in use (e.g.,
Vasconcelos et al. (2003); Henning et al. (2012)). All studies conducted in 2003 or later with
these nonhuman primate species or the cynomolgus monkey used the Ames strain of B.
anthracis.
The assessment of nonhuman primate pathology of the lung is complicated by lung mite
(Pneumonyssus simicola) parasitism in most rhesus monkeys used for testing during the 1960s.
Studies that reported lung mites in challenged monkeys include Berdjis et al. (1962) and Gleiser
et al. (1963). At the time of challenge, the mites contributed to lung lesions, which became sites
of superinfection with B. anthracis (Fritz et al., 1995). Therefore, comparisons of lung pathology
between rhesus monkeys and other nonhuman primates may be difficult based on the availability
of one study (Fritz et al., 1995) that reported pathology of rhesus monkeys without mite
infection. Noting the similarity in pathology between the rhesus monkey and the fulminant
necrotic and hemorrhagic pneumonia described by Abramova et al. (1993) during the Sverdlovsk
outbreak, Fritz et al. (1995) hypothesized that the described nonhuman primate pathology
resulting from infection under conditions of pre-existing lung lesions may mimic that of the
human with pulmonary compromise and have utility in that context. Hemorrhagic pneumonia
has been reported in the nonhuman primate (Albrink and Goodlow, 1959; Fritz et al., 1995;
Lever et al., 2008), as well as hemorrhage of varying severity, absent pneumonia, in the lung
(Gleiser et al., 1963; Vasconcelos et al., 2003; Twenhafel et al., 2007).
This assessment examined nonhuman primate species as one group for the evaluation of the
pathology data. However, species-specific data that indicate a lack of concordance with expected
human pathology or that of other nonhuman primate species were also highlighted.
47
-------
The nonhuman primate species exhibited generally consistent clinical and pathological outcomes
after exposure to lethal inhalation doses of B. anthracis (Twenhafel, 2010). Though few low-
dose studies have been conducted, one study reported similar pathology across a range of low-
dose (200 to 2 x 104 CFU) and high-dose (2 x 104 CFU to 1 x 107 CFU) challenges for the
African green monkey (Twenhafel et al., 2007). Similarities in response were identified for age
(e.g., adult versus juvenile) and sex (e.g., male versus female) in the dose range of 2 x io4 to
5 x 1010 CFU (Twenhafel, 2010).
A detailed summary table of end-stage pathology for the rabbit, nonhuman primate, and human
is provided in Appendix C, Data Summary Table for End-stage Inhalation Anthrax Pathology of
the Human, Nonhuman Primate, and Rabbit. Table 5-8 summarizes the published nonhuman
primate pathology relative to the key findings of Twenhafel (2010).
48
-------
Table 5-8. Summary of Nonhuman Primate Pathology Relative to Twenhafel (2010) Key
Findings
Pathology
Key Findings
Pneumonia
Pleural effusions (Albrink and Goodlow, 1959; Dalldorf et al., 1971; Vasconcelos et al., 2003;
Twenhafel et al., 2007); though not reported in rhesus macaque (Twenhafel et al., 2007)
Hemorrhagic pneumonia (Albrink and Goodlow, 1959; Lever et al., 2008); low incidence of
pneumonia (2/13) but presence of hemorrhages (Fritz et al., 1995);
Alveoli filled with edema often mixed with fibrin, hemorrhage, macrophages, and neutrophils
(Twenhafel et al., 2007); acute suppurative inflammation (4/14) (Vasconcelos et al., 2003)
Splenic lymphoid
depletion
Splenomegaly (Albrink and Goodlow, 1959; Middleton and Standen, 1961; Gleiser et al., 1963;
Lever et al., 2008); low incidence identified from one study (3/13) (Fritz et al., 1995) or
described as mild (Twenhafel et al., 2007);
Histiocytosis (Fritz et al., 1995); hemorrhage in splenic marginal zone (Fritz et al., 1995);
necrosis of lymph follicles and/or necrosis of red and white pulp with hemorrhage (21/23)
(Dalldorf etal., 1971)
Meningitis
Meningitis (9/21) (Dalldorf et al., 1971); suppurative meningitis (10/13) (Fritz et al., 1995)
Hepatic
hemorrhage or
inflammation
Liver congestion (Albrink and Goodlow, 1959; Lever et al., 2008)
Diffuse hepatic congestion fibrin deposition and expanded germinal center (Lever et al., 2008);
lymphocytic depletion (Fritz et al., 1995)
Acute inflammation/leukocytosis (13/14) and acute necrosis (5/14) in liver (Vasconcelos et al.,
2003); sinusoidal leukocytosis (9/10), necrosis (6/10) and acute inflammation (4/10) (Henning
etal., 2012)
Gastrointestinal
hemorrhage or
inflammation
Hemorrhage of various severity in the small and large intestine serosa and esophagus mucosa
(Fritz et al., 1995); or stomach mucosa and/or submucosal tissues (Fritz et al., 1995;
Vasconcelos et al., 2003);
Acute colitis with necrotizing vasculitis (1/13) (Fritz et al., 1995), necrosis of villus tips in
ileum or jejunum (9/14) (Vasconcelos et al., 2003), or with stomach inflammation (2/14) or
ulceration (1/14) (Vasconcelos et al., 2003)
Edema, congestion, and hemorrhage in the gastrointestinal tract (Twenhafel et al., 2007)
Urogenital
hemorrhage or
inflammation
Periovarian or peritesticular congestion and/or hemorrhages (Twenhafel et al., 2007)
Ovarian hemorrhage and necrosis (1/14) (Vasconcelos et al., 2003)
Gross pathology in the nonhuman primate from inhalation anthrax includes edema, hemorrhage,
and varying levels of necrosis in the lungs, lymph nodes, and spleen (Fritz et al., 1995; Leffel
and Pitt, 2006). Generally mild levels of leukocytic infiltration in the tissues were also reported,
with mild levels typically indicating a highly susceptible host (Fritz et al., 1995). Gross and
histological changes in lymphoid tissues are a key pathological outcome in fulminant inhalation
anthrax. Lymphoid tissues exhibiting consistent pathology across nonhuman primate species are
the mediastinal lymph nodes (specifically, the tracheobronchial lymph node) and the spleen, with
49
-------
hallmark lesions, including the presence of necrohemorrhagic lymphadenitis and generalized
lymphoid depletion (Twenhafel, 2010). However, the marmoset animal model as described by
Lever et al. (2008) exhibits a relatively low incidence of "classic lesions" in the lymph nodes (1
of 6 marmoset with enlarged and hemorrhagic tracheobronchial lymph nodes) and an absence of
meningitis (Twenhafel, 2010).
Splenomegaly (i.e., enlargement of the spleen) is a gross pathology outcome commonly
identified in the nonhuman primate (Albrink and Goodlow, 1959; Gleiser et al., 1963; Lever et
al., 2008). Splenomegaly was also identified in the pathology reported when fulminant anthrax
developed from intracutaneous dosing of rhesus monkeys with B. anthracis (Vollum IB) spores
(Middleton and Standen, 1961). Gross splenic changes have been described as enlargement with
rounded edges, dark red color, and an appearance to similar to "blackberry jam" (Twenhafel,
2010).
Though there is some variation in the frequency with which splenomegaly was reported across
nonhuman primate species, there may not be a meaningful splenic pathology difference across
the nonhuman primate species when considering the general consistency in reported
histopathological data. Fritz et al. (1995) reported a lower incidence of splenomegaly (3 of 13) in
the rhesus monkey relative to that reported as "frequently seen" by Gleiser et al. (1963).
However, Fritz et al. (1995) also reported characteristic microscopic lymphoid changes (e.g.,
splenic histiocytosis [12/13], lymphoid depletion [13/13], and hemorrhage in spleen marginal
zone [7/13]) present in the majority of monkeys without regard to the presence of gross
splenomegaly. Gleiser et al. (1963) identified histopathological changes in the spleen as similar
to the lymph node (i.e., necrosis, hemorrhage, "depopulated" state), though a quantitative
50
-------
description was not provided. Vasconcelos et al. (2003) reported mild splenomegaly (enlarged
1.5- to 2-fold) in 13 of 14 cynomolgus monkeys challenged with high doses of B. anthracis.
However, the reported histopathological data provided did not extend beyond a general
description of lymphocytolysis and general congruence in the pathology with the intrathoracic
lymph nodes. Dalldorf et al. (1971) reported splenic changes of necrosis in the red and white
pulp of the spleen with hemorrhage in 14 of 23 cynomolgus monkeys, but did not identify gross
pathology relating to general spleen enlargement. Lever et al. (2008) noted gross pathology
indicative of splenomegaly in 2 of 6 marmoset, yet described microscopic findings in 6 of 6
marmoset of lymphoid depletion, necrosis, fibrin, and hemorrhage, as well as acute
inflammation.
There were also species-specific differences in the reporting of pleural effusions across the
nonhuman primate species. Pleural effusions were identified in the chimpanzee, cynomolgus
macaque, and African green monkey (Albrink and Goodlow, 1959; Dalldorf et al., 1971;
Vasconcelos et al., 2003; Twenhafel et al., 2007). However, pleural effusions were not reported
in the rhesus monkey (Gleiser et al., 1963; Twenhafel et al., 2007) and marmoset (Lever et al.,
2008). The relevance of this difference is not currently known.
Cardiac tissue lesions or the associated myocardium were identified more frequently in the
cynomolgus monkey than in the rhesus monkey when Vasconcelos et al. (2003) compared their
cynomolgus study results with those for the rhesus monkey reported by Fritz et al. (1995) and
Gleiser et al. (1963). These lesions have not been reported in the human (Vasconcelos et al.,
2003), though pericardial effusions were identified in 2001 anthrax letter event cases (Jernigan et
al., 2001). The presence of differing pathology may be an area of true difference in tissue or
51
-------
organ-specific susceptibility among the nonhuman primate species (Vasconcelos et al., 2003)
and the human.
5.4.1.5 Results of Concordance Analysis for Similarity between Rabbit, Nonhuman
Primate, and Human Pathology
The rabbit and nonhuman primate exhibit many commonalities in the type of lesions and tissues
associated with inhalation anthrax pathology in the human. For example, Zaucha et al. (1998)
identified that the end-stage pathology of anthrax in the rabbit as being "remarkably similar" to
the human. Vasconcelos et al. (2003) reported that the "pattern of inhalation anthrax lesions" was
similar among the cynomolgus monkey, rhesus monkey, and the human.
The principal anthrax lesions of edema, hemorrhage, and necrosis are present in a variety of
common tissues in the rabbit, nonhuman primate, and human. However, this constellation of
pathology is generally consistent with descriptions of animal models susceptible to fulminant
inhalation anthrax infection (Gleiser et al., 1963) and is not unique to the rabbit and nonhuman
primate animal models. Lesion differences among susceptible animals are manifested by
differing levels of inflammation and infiltration of leukocytic elements into existing lesions (U.S.
Food and Drug Administration, 2002), whereby less susceptible animals exhibit greater
inflammation and leukocytic infiltration than more susceptible animals, which rapidly succumb
to illness.
The lymphoid tissues are the primary target for anthrax lesion development in susceptible
animals (U.S. Food and Drug Administration, 2002), specifically the lymph nodes draining the
lungs, pharynx, or gastrointestinal tract, the spleen, and lymphoid tissues associated with the
gastrointestinal tract (e.g., Peyer's patches, sacculus rotundus, appendix). The most commonly
affected lymph nodes are the thoracic lymph nodes, including the mediastinal lymph nodes
52
-------
(human, rabbit, and nonhuman primate), the submandibular (rabbit), and the cervical lymph
nodes (nonhuman primate). Anthrax pathology of the affected lymph nodes includes necrosis,
hemorrhage, and depletion and/or destruction of lymphocytes (lymphocytolysis), with these
characteristics identified in the overall pathology of the rabbit, nonhuman primate, and human
(Dalldorf et al., 1971; Abramova et al., 1993; Zaucha et al., 1998; Guarner et al., 2003).
Movement to and through the lymph node or other more direct routes to the bloodstream allow
for systemic accessibility of the pathogen. This allows for infection and associated pathology to
be exhibited in distant nonlymphoid tissues in the rabbit and nonhuman primate, including
adrenal glands, ovarian or testicular tissues, and myocardial tissue (Gleiser et al., 1963; Fritz et
al., 1995; Zaucha et al., 1998; Vasconcelos et al., 2003; Twenhafel et al., 2007).
There are two areas of difference between the anthrax presentation in the human and the
nonhuman primate. The first is the presentation of splenomegaly or splenic histopathology. As
described in Section 5.4.1.4, variation among the nonhuman primates in the presence or absence
of splenomegaly has been reported. There are also conflicting reports regarding the presence or
absence of splenomegaly in the human. Based on reports from Sverdlovsk from Abramova et al.
(1993) and Grinberg et al. (2001), Vasconcelos et al. (2003) determined that humans do not
typically exhibit splenomegaly. Fritz et al. (1995) also identified a limited occurrence of
splenomegaly in the human. However, splenomegaly was described in earlier human inhalation
anthrax case reports by Albrink et al. (1960) and Suffin et al. (1978). Histopathology conducted
on four of the 2001 anthrax letter event cases described splenic histopathology to include
congestion (3 of 4 individuals) and necrosis (1 of 4 individuals) (Guarner et al., 2003). Available
gross pathology is very limited from the Sverdlovsk outbreak and frequency of splenomegaly is
unknown. However, histopathology on stored tissues from Sverdlovsk reported splenic
53
-------
pathology to include lymphocytosis, splenic congestion, and presence of neutrophils (Grinberg
et al., 2001), which are not inconsistent with presentation of splenomegaly. Alternately, B.
anthracis strain-specific effects may be contributing to apparent differences in the presentation
of splenomegaly in the human, as the historic human data were reflective of exposure to mill
aerosol strains, whereas later data reflected exposure to the Ames strain in the 2001 anthrax letter
event or the mixture of strains in Sverdlovsk.
Meningitis has been reported as a second differentiator in anthrax pathology between the rabbit
and the nonhuman primate animal model due to identified absence of meningitis in the rabbit
(Twenhafel, 2010). Zaucha et al. (1998) described a low incidence of hemorrhage associated
with B. anthracis bacilli in the rabbit brain or meninges and noted the lack of leukocytic
infiltration in these lesions. Since that report, one study reported meningitis with suppurative
inflammation in a high-dose (c. 106 CFU) control group rabbit (1 of 25 rabbits) (U.S.
Environmental Protection Agency, 2011a). The absence of "full blown" meningitis is
hypothesized to result from the rapidity of disease progression in the rabbit, which limits the
opportunity for inflammation and leukocytic response (Zaucha et al., 1998; Leffel and Pitt,
2006). Meningitis lesions in the rabbit were typically noninflammatory when compared to the
suppurative, inflammatory lesions described in the nonhuman primate and human (U.S. Food
and Drug Administration, 2002). Interestingly, the rabbit that exhibited meningitis in the U.S.
Environmental Protection Agency (201 la) study had a time-to-death of four days, which was at
the high end of the range for time-to-death values (i.e., 2 to 3 days, mean of 2.4 days) reported by
Zaucha et al. (1998). Alternately, Zaucha et al. (1998) hypothesized that strain differences could
be contributing to variation in the incidence of meningitis in the rabbit versus nonhuman primate
as earlier nonhuman primate studies used Vollum strains as reported in Fritz et al. (1995) and
54
-------
Gleiser et al. (1963). However, studies conducted since that time with the Ames strain in
nonhuman primate species have reported meningitis in the African green monkey with a similar
incidence as prior nonhuman primate studies (Twenhafel et al., 2007), as well as suppurative
meningitis with hemorrhage in the cynomolgus monkey (Vasconcelos et al., 2003).
Time-dependency in anthrax pathology also contributes to differences in lesion tissue location
and presentation among the rabbit, nonhuman primate, and human (U.S. Food and Drug
Administration, 2002; Leffel and Pitt, 2006). However, this poses a challenge for the systematic
evaluation of anthrax pathology of animal models and the human because of recognized
differences in the time-to-death values typically associated with each group. As described earlier,
the rabbit typically exhibits the shortest time-to-death values as evidenced by the commonly
cited value of 2 to 3 days of Zaucha et al. (1998). The nonhuman primate exhibits a wider range
of values for time-to-death, with 3 to 8 days reported in Fritz et al. (1995). The human with a
slightly longer time-to-death values as evidenced by the reported range of value of 5 to 8 days in
Jernigan et al. (2001). As identified earlier, a complicating factor for interpretation of human
pathology data is the unknown contribution that magnitude of dose or initiation of medical
treatment may play in resulting time-to-death and/or pathology.
The relationship between survival time and lesion development was first recognized over 50
years ago in the nonhuman primate (Albrink et al., 1960; Berdjis et al., 1962). When evaluating a
possible connection between the use of antibiotics and the presence of meningitis in study
animals, Albrink et al. (1960) hypothesized that antibiotics may reduce damage in non-central
nervous system tissues and prolong life, such that individual bacteria that travel to the meninges
have sufficient time to multiply and develop into meningitis. Time-dependent development of
55
-------
lesions was also described in the nonhuman primate without medical or antibiotic treatment post-
challenge. Nonhuman primates with an extended survival time post-challenge relative to shorter-
lived animals in the same study were more commonly found to exhibit disease progression in
specific tissues (e.g., adrenal inflammation and necrosis, hepatic necrotic lesions, meningeal
hemorrhage, cerebral vasculitis) (Vasconcelos et al., 2003). As would be expected, the severity
of lesions may also be affected by the length of survival time for disease progression. Lesions
and associated inflammation in the mediastinal area (mediastinitis) were described in the
nonhuman primate and the human, though a lesser severity of mediastinitis was noted for the
rabbit relative to the human (Zaucha et al., 1998). Zaucha et al. (1998) hypothesized that a longer
disease progression would provide necessary time for expansion of the infection from the lymph
nodes to the surrounding mediastinal tissues.
Overall, the human exhibits less susceptibility than the rabbit and nonhuman primate, with the
result being a longer period of disease progression (i.e., longer time-to-death after challenge)
(U.S. Food and Drug Administration, 2002; Leffel and Pitt, 2006). The increased time of length
of disease allows for development of more inflammatory elements of the pathology (U.S. Food
and Drug Administration, 2002). As an example, the rabbit typically exhibits less severe
mediastinal lesions, reduced incidence of pneumonia, and a lack of leukocyte invasion in the
meninges and brain than species less susceptible to anthrax (Leffel and Pitt, 2006) and generally
has the shortest time-to-death after challenge.
The purpose of the concordance review was to evaluate available pathology data for the
nonhuman primate and to select appropriate dose-response data for lethality to extrapolate to the
human. However, this review should not be directly applied to other endpoints (e.g., infection)
56
-------
without additional analysis. As noted previously, the key histopathology findings in the human
identified by Twenhafel (2010) were used as a starting point. These findings included hepatic,
gastrointestinal, and urogenital hemorrhage and inflammation; pneumonia; splenic lymphoid
depletion; and meningitis. In the evaluation of animal models for the testing of medical
countermeasures, a close replication of the human disease state is desired to ensure the treatment
being assessed is protective of a full range of adverse anthrax illness outcomes in addition to
lethality (e.g., meningitis, organ, or tissue damage). In contrast, animal model selection for dose-
response analysis focuses identification on key events associated with disease progression
relative to the identified endpoint of interest (i.e., lethality for this assessment).
Uncertainty in the key events process for development of inhalation anthrax complicates the use
of a disease progression approach from initiation of infection through end-stage illness. To
reduce reliance on a strict disease progression interpretation, the animal model selection
assessment evaluated general concordance in tissue location and pathology associated with
inhalation anthrax in the animal models and the human. As the data were analyzed, time-
dependency was considered to play a potential role in the relative development of pathology
across hosts and was incorporated as an element of the final assessment. While the lack of serial
sacrifice data for the animal models limits the ability to draw conclusions for the precise timing
and relative sequence of events of inhalation anthrax pathology, the identification of differences
in the appearance of pathology between animals dying earlier and later may assist in determining
those elements associated with a longer duration of infection (e.g., meningitis).
Table 5-9 shows general concordance in the anthrax pathology between the rabbit and nonhuman
primate with regard to presence or absence of lesions and inflammation in target tissues
57
-------
associated with anthrax pathology in the human. The pathological lesions identified in the human
for which the rabbit animal model differs with the nonhuman primate have a time-dependent
element in their presentation, with the rabbit differing from the nonhuman primate either in the
severity as defined by level of inflammation or leukocytic infiltration or general incidence (Table
5-9).
There were no identified differences between the rabbit and the nonhuman primate animal
models for elements of anthrax pathology that do not have a time-dependency regarding
incidence or severity in presentation. However, those elements of pathology that showed
differences between the rabbit and nonhuman primate animal model preliminarily indicate that
that time-dependency may be related to their pathological presentation. The results of the
concordance assessment of pathology support the use of the rabbit and nonhuman primate animal
models for development of dose-response data.
58
-------
Table 5-9. Key Human Histopathological Findings Relative to Time-Dependent Pathology
in the Rabbit and Nonhuman Primate after Single-Dose Exposure
Pathology
Rabbit
Nonhuman Primate
Evidence for Time-
Dependency in Severity or
Incidence
Pneumonia
Yes - Zaucha et al.
(1998) with noted lower
incidence and severity
than NHP and human
Yes - Alb rink and
Goodlow (1959), Fritz et
al. (1995)
Yes - Progression to pneumonia
is associated with inflammatory
process, lower incidence, and
lesser severity reported in rabbit
Splenic lymphoid
depletion
Yes - Zaucha et al.
(1998),
Lovchik et al. (2012)
Yes - Fritz et al. (1995)
No - Spleen is an early disease
target in inhalation anthrax
Meningitis
Yes-U.S.
Enviromnental Protection
Agency (2011a) in 1/25
rabbits, lower incidence
than NHP and human
Yes - Fritz et al. (1995),
Gleiser et al. (1963),
Lever et al. (2008),
Twenliafel et al. (2007),
Vasconcelos et al. (2003)
Yes - Hypothesized as time-
dependent in Zaucha et al.
(1998), not identified in any of
NHP serial sacrifices reported in
Berdjis et al. (1962)
Hepatic hemorrhage
or inflammation
No - Not reported after
inhalation exposure
pathology, one report
after intravenous dosing
inNordberg et al. (1961)
Yes - Vasconcelos et al.
(2003), Henning et al.
(2012), Lever et al.
(2008)
Yes - Reported as time-
dependent in NHP by
Vasconcelos et al. (2003)
Gastrointestinal
hemorrhage or
inflammation
Yes - Zaucha et al.
(1998), U.S.
Enviromnental Protection
Agency (2012b)
Yes - Fritz et al. (1995);
Vasconcelos et al. (2003)
No - Hemorrhagic spread to
gastrointestinal tract seems to
occur early in the disease process
Urogenital
hemorrhage or
inflammation
Yes - Zaucha et al.
(1998) but noted as rare
Yes - Twenliafel et al.
(2007), Vasconcelos et
al. (2003)
Unknown - Evidence or reports
for time-dependency are lacking
NHP - nonhuman primate
5.4.2 Identification of Microbial Dose-Response Data
A literature search was conducted for open source rabbit, nonhuman primate, and human dose-
response data, including dose-response data sets, modeled LD50 values, or reported parameter
values (e.g., probit slope values). Given the scarcity of available human data, dose-response data
were more broadly defined for the human to include epidemiological and qualitative dose-
response data. Dose-response studies that reported either acute (i.e., less than 24-hour or single-
dose) or multiple-dose exposures were identified. Dose-response studies that reported infection
and/or lethal endpoints were also collected in the literature search. The search evaluated
published literature from January 1950 through January 2014. However, documents of historical
59
-------
relevance (i.e., pre-1950) that provided background or context for
selected secondary data were also identified as part of the literature
search.
A dose-response relationship describes "the relationship between a
quantified exposure (dose) and the proportion of subjects
demonstrating specific biologically significant changes in incidence
and/or in degree of change (response)" (U.S. Environmental
Protection Agency, 201 lc). To model the dose-response relationship,
response data must be reported for each individual or dose group. For
the inhalation route of exposure, the exposure dose must have been
reported as an inhaled dose or a deposited inhaled dose, or sufficient
data was provided to derive an inhaled dose. For data that did not
report an inhaled or deposited dose metric, an allometric equation
could be used to calculate an exposure dose if environmental
concentration with individual or group animal weight data were
available. Oral dose-response data were collected without regard to
dose metric or animal model due to the recognized scarcity of
published data.
For the human, acceptable dose-response data were more broadly defined to include additional
data types. Published epidemiological data, modeled values, and parameters developed from
animal and/or human inputs or fitted parameter values, and data derived from expert elicitation
processes were all targeted by the literature search. If the animal data were identified as
60
Summary of Findings for
Identification of Microbial
Dose-Response Data
• Few inhalation challenge
studies were identified as
Key Studies for the rabbit
and nonhuman primate;
there were no Key Studies
or Supporting Studies
identified for the human.
• There were very few
single or multiple dose
challenge studies using
low doses.
• Dose-response data are
available for the rabbit
and nonhuman primate
that may be suitable for
development of a human
dose-response
relationship.
• The uncertainty
associated with the use of
these data may be high.
• Depending on the level of
acceptable uncertainty in
the analysis outputs, there
may be limitations on
how these data may be
used in decision-making.
-------
appropriate to apply to the human, they were evaluated for use as human dose-response data.
Acceptable epidemiological data identified known exposure characteristics associated with
human outbreaks of anthrax illness. Qualitative data describing the relative susceptibility of the
human to anthrax infection were also collected as they were identified.
After identification by the literature search, all dose-response data sets and modeled dose-
response values were evaluated using general quality criteria identified in the U.S.
Environmental Protection Agency (2003) data quality guidance. Data sets that met the general
quality criteria were then further evaluated using the project-specific criteria described in the
next section.
5.4.2.1 Categorization of Dose-Response Data
Project-specific criteria in the form of assessment questions and defined rules for data handling
were used to categorize the identified dose-response data as Key Studies, Supporting Studies,
and Additional Data. The process described in U.S. Environmental Protection Agency (2012c)
was the starting point for the development of assessment questions and the evaluation process.
U.S. Environmental Protection Agency (2012c) evaluated publishedB. anthracis dose-response
data relative to its utility for developing dose-response relationships, especially in the low-dose
region. The assessment questions presented in U.S. Environmental Protection Agency (2012c)
addressed: (1) the availability of raw dose-response data (i.e., original data set), (2) the
availability of particle size distribution data, including reported use of single spore particles in
the challenge, (3) the presence of dose groups with less than 50% lethality rate or an overall
lethality rate of less than 50% when individual doses were reported, (4) the use of real-time
61
-------
methods to derive inhalation rates, (5) sufficient animal numbers in individual dose groups (n >
5) or total number tested for individual dose measurements (n > 12).
The purpose of the report is to generate a comprehensive picture of available dose-response data
and models for B. anthracis. As a result, dose-response data were sought even if an individual
data set might be insufficient to derive a dose-response relationship. To incorporate this change,
modifications were made to the process identified in U.S. Environmental Protection Agency
(2012c): (1) dose-response data (e.g., model parameter values and outputs, epidemiological data
for the human) were defined more broadly, (2) the quantitative scoring process was not used, and
(3) different output assessment categories were employed. Given the potential for inhalation
rates derived from allometric data to significantly under- or overestimate the actual dose (U.S.
Environmental Protection Agency, 2012c), one additional modification was made to the process:
the use of real-time methods (e.g., plethysmography) was a Key Study design requirement.
Using knowledge gained from the implementation of the assessment questions in U.S.
Environmental Protection Agency (2012c), default rules were developed to place data in the
Additional Data category. Dose-response data that consisted solely of high-dose challenge of a
control group for a medical countermeasure study were automatically categorized as Additional
Data. The dose levels used in the high-dose challenges are dose typically 100 to 200 times the
Zaucha et al. (1998) LD50 value. If the original dose-response data set was not available, a all
modeled values (e.g., probit slope values, fitted parameter values, LD50) were placed in
Additional Data. If the original data set was identified, modeled values were reported alongside
their originating data set in the summary tables. All identified epidemiological data for the
human were categorized as Additional Data. Dose-response data were not quantitatively scored
62
-------
as they were in U.S. Environmental Protection Agency (2012c), but were categorized based on
the sufficiency of the published data for modeling dose-response relationships for low-dose
exposures or for informing dose-response relationships of higher quality data.
Key Studies were defined as representative of the highest quality dose-response studies that met
criteria for selection during the literature search. Quality was defined by the availability of study
data, study design with real-time inhalation rate and particle size measurement, data elements
including evaluation of low dose and associated response levels (i.e., between 1% and 50%
lethality), and sufficient number of animal and dose group numbers to mathematically model a
dose-response relationship. Supporting Studies had identifiable limitations in assessment quality
indicators relative to Key Studies, yet were found to have potential in bounding potential dose-
response relationship(s) as described by Key Studies. Additional Data were defined by the lack
of data critical to assessing dose-response relationships (e.g., original dose and response data set)
or study design elements that limit utility for development of low-dose dose-response
relationships. As a result, their utility in dose-response analysis may be limited to providing
corroborative support for higher quality data.
Key Studies are presented in summary text and tables in the following sections, with strengths
and weaknesses relative to the use of these data in dose-response analysis also identified.
Modeled dose-response values that are re-analyses of previously published primary data are
associated with the primary data set, if the data set was identified. Highly relevant or often cited
Additional Data were also reported in conjunction with Key Studies to provide additional context
for the presented data. Summary of dose-response data that were categorized as Supporting Data
63
-------
or Additional Data are provided in Appendices D and E for the rabbit and nonhuman primate,
respectively.
5.4.2.2 Results from Literature Search 0/° Bacillus anthracis Dose-Response Data
The development of a human-relevant dose-response relationship for B. anthracis is challenged
by a lack of suitable data sets for dose-response analysis (U.S. Department of Homeland Security
and U.S. Environmental Protection Agency, 2009). One area of particular concern is the limited
number of low-dose exposure studies for single- and multiple-dose challenges. The majority of
animal dose-response data identified through the literature search originated from single-dose
studies at very high doses, sometimes as high as 200 times the identified LD50 value. Single
high-dose studies have limited value for the assessment of repeated low-dose exposure (U.S.
Environmental Protection Agency, 2012c).
Few studies that reported dose-response data were designed to derive data for dose-response
analysis. Reported study purposes for recent data sets included evaluation of the pathology or
pathophysiology of infection, or assessment of the efficacy of medical countermeasures. These
studies were often conducted using a single high-dose challenge to ensure a high likelihood of
systemic anthrax infection in the challenge animals. Historical data were often developed to
report an LD50 value for use in military applications or early anthrax research and little attention
was paid to representation of low doses.
Few studies were identified as Key Studies for the rabbit and nonhuman primate; there were no
Key Studies or Supporting Studies identified for the human. The two Key Studies for the rabbit
were the single-dose U.S. Environmental Protection Agency (201 la) study and the multiple-dose
U.S. Environmental Protection Agency (2012b) study. No studies were categorized as
64
-------
Supporting Studies. For the nonhuman primate, one single-dose Key Study (Lever et al., 2008)
and one single-dose Supporting Study (Druett et al., 1953) were identified.
5.4.2.3 Human Inhalation Data
All identified human dose-response data for the human were categorized as Additional Data.
Human dose-response data included epidemiological data, modeled data from the nonhuman
primate that were identified for human application (with or without the addition of human
relevant values), and specific values or ranges elicited from experts for modeled values of
interest (e.g., LD50). Dose-response data were primarily reported using the lethality endpoint.
However, the ID and LD were identified as equivalent by expert elicitation (Rickmeier et al.,
2001), in the presentation of a range of median infectious dose (ID50) values, (U.S. Army
Medical Research Institute of Infectious Diseases, 2011), or incorporated in modeling (Webb and
Blaser, 2002; Wein et al., 2003; Craft et al., 2005; Toth et al., 2013).
No open source studies reported human dosing with B. anthracis. The lack of available human
dose-response data has been previously reported (Taft and Hines, 2012; Toth et al., 2013).
Environmental exposure or dose data were not reported with human outbreak data (e.g.,
Sverdlovsk, 2001 anthrax letter event). However, there was one study (Dahlgren et al., 1960),
with subsequent reanalysis by Cohen and Whalen (2007), that reported two days of air
measurements to which a mixture of vaccinated and unvaccinated mill workers were exposed
without incidence of anthrax illness.
Primary citations of human dose-response data identified through the literature review are
presented in Table 5-10. Repeated secondary citations of the same human dose-response data
were not included here. For example, there were numerous citations of the Inglesby et al. (2002)
65
-------
human LD50 range. Qualitative assessments regarding relative susceptibility that were identified
through the literature search are also summarized.
When reviewing Table 5-10, it is important to recognize that the LD50 values come from a
variety of data sources with varying levels of data quality and reproducibility (e.g., expert
elicitation, combinations of human epidemiological and animal model challenge data), as well as
variability in fundamental study design elements (e.g., animal model, strain). The literature
search identified a number of incorrect citations of previously published data (i.e., secondary
data). These unique values are included in Table 5-10 and identified as incorrect, but are not
considered further in the report.
Table 5-10. Additional Data for the Human
Published Study
Value or Associated Model (B. anthracis Strain)
Basis for Value or Model Specification
Cohen and Whalen (2007)
(Originating data set: Mill aerosol, unknown strain[s])
600 inhaled respirable spores over an 8 hour day is the
"lower boundary of the maximum noninfectious dose for
inhalation anthrax" in a healthy individual "who is not
egregiously predisposed to anthrax or lung disease, or is
immunocompromised"
Data reported in Dahlgren et al. (1960), Brachman
et al. (1966), and assumptions regarding the human
exposure rate were used to derive the 600 inhaled
respirable spores value
Craft et al. (2005)
Age-dependent linear dose-response model to predict the
probability of infection for a given age
(unknown strain)
P(s,a)-imn(l,(ci_c;j)
s = dose, a = age
c1 = 38,000
c2 = 450
Amax = 80
Age distribution U[0,A]and pdf f(a) = A_1
Craft et al. (2005) is an independent paper by
members of AMWG. Used ID values from Table 3
in Webb and Blaser (2002). Original data source
for nonage-dependent ID values was Rickmeier et
al. (2001)
Curling et al. (2010)
(Originating Druett et al. (1953) data set strain: M36)
Exponential model, fitted parameter X = 1.36 x 10 5
LD50 = approximately 51,000 spores
Druett et al. (1953) nonhuman primate data for
single spore clouds, model fitted parameter and
output reported in the NATO Planning Guide for
the Estimation of Chemical, Biological,
Radiological, and Nuclear (CBRN) Causalities,
Allied Medical Publication - 8(c) (Curling et al.,
2010)
66
-------
Published Study
Value or Associated Model (B. anthracis Strain)
Basis for Value or Model Specification
Dahlgren et al. (1960)
(Originating data set: Mill aerosol, unknown strain[s])
Approximately 1,300 spores (510 spores in particles 5 |im
and less in size) may be inhaled over 8 hours by
nonimmunized individuals in an occupational setting
without infection
Airborne measurements of B. anthracis spores
taken in Pennsylvania textile mill during time
period with no reported incidence of human
inhalation anthrax in a population where only 33%
were vaccinated
Defense Intelligence Agency (1986)
(Unknown strain)
LD50 = 8,000 to 10,000 spores
Unspecified studies
Franz et al. (1997)
(Unknown strain)
ID = 8,000 to 50,000 spores
Unspecified studies, Franz et al. (1997) identified
USAMRIID as general source of information for
values, U.S. Army Medical Research Institute of
Infectious Diseases (2011) identifies the same
range of values for ID
Inglesby et al. (1999), Inglesby et al. (2002)
(Unknown strain)
LD50 = 2,500 to 55,000 inhaled spores [sic]
Incorrect identification of Defense Intelligence
Agency (1986) reported LD50 range
Rickmeier et al. (2001)
(Unknown strain)
IDso = between 8,000 and 10,000 spores (calculated as
8,940 spores)
IDio = 1,000 to 2,000 spores (calculated as 1,135 spores)
ID90 = 50,000 to 100,000 spores
Calculated probit slope = 1.43 probits/logio dose
Subject matter expert opinion elicited for ID values
and used to calculate probit slopes
Toth et al. (2013)
Exponential model with time-dependence
(Originating data set: Mill aerosol, unknown strain[s])
Simplified Equation:
I (d,t) = 1 — exp(—rd( 1 — e~0t))
r = 6.4 x 10-5
(CI = 4.0 x 10"5to 9.5 x 10"5)
EISD model populated with human and nonhuman
primate data sources, Brachman et al. (1966) for
nonhuman primate dose-response data,
Brookmeyer et al. (2005) reported value for rate of
clearance (0) = 0.07 day-1 based on Henderson et
al. (1956) nonhuman primate, and Holty et al.
(2006) for human Sverdlovsk data
r value determined after setting the following parameters:
Rate of clearance (0) = 0.07 day-1
Best fit T distribution shape parameter a = 5.43 and scale
parameter b = 0.864 for assessing time-dependent elements
of disease progression
with:
/- infection
d - dose
t - time
ID50 = 11,000 spores
(95% CI = 7,200 to 17,000)
ID10 = 1,700 spores
(95% CI = 1,100 to 2,600)
IDi = 160 spores
(95% CI = 100 to 250)
67
-------
Published Study
Value or Associated Model (B. anthracis Strain)
U.S. Centers for Disease Control and Prevention (2009)
(Unknown strain[s])
LD50 = 4,100 to 10,000 inhaled spores
Basis for Value or Model Specification
Nonhuman primate data from Glassman (1966),
Peters and Hartley (2002), and Franz et al. (1997).
Note: Glassman (1966) referenced as Glassman
(1965) in U.S. Centers for Disease Control and
Prevention (2009).
U.S. Army Medical Research Institute of Infectious
Diseases (2011)
(Unknown strain)
ID = 8.000 to 50.000 spores
No studies identified, same ID range as identified
inFranz et al. (1997).
Watson and Keir (1994)
(Unknown strain)
6,000 inhaled spores as a'
dose to man"
worst" case inhalation critical
Brachman et al. (1960) NHP LD50 value identified
as the lowest single strain LD50 value of 6,000
spores, assumed direct applicability to the human.
Webb and Blaser (2002)
Logit equation describing probability of infection given
age (a) and dose (S), with a[n] = ID50 and b[n] = ID10 with
age-specific values identified below
(Unknown strain)
Used expert elicitation values for specific IDX
values as reported in Rickmeier et al. (2001) and
modified to develop age-adjusted distribution.
b[n]
Pr[n](S) =
(""(asr1)
Ui ')
1 + Z?[n] I exp
ID50 and ID10 values by age group:
Less than 25 years: 15,000 and 4,500 spores
25-44 years: 10,000 and 3,000 spores
45-65 years: 6,000 and 1,800 spores
Greater than 65 years: 1.500 and 450 spores
Wein and Craft (2005)
(Unknown strain[s])
Probit slope value of 1.82
Probit slope value of 0.7
Wein and Craft (2005) is an independent paper by
members of AMWG convened by DHHS, probit
slope value of 1.82 reportedly developed by Harper
and Kaufmann of the AMWG, no description or
formal citation for derivation, the source of the 0.7
value was Glassman (1966).
Wein et al. (2003) (Supporting Text)
Age-dependent probit slope model
(Unknown strain[s] in Glassman [1966])
P(s, a) = ® (a +(3 log(s) + y (a) + 5(a2)
Where s = dose of spores
a = age in years
® = standard normal distribution
Intercept (a) = -9.733
Probit dose slope ((3) = 1.025
Probit age slope (y ) = -0.016 year-1
Probit age quadratic (5) = 0.0006 year'2
Wein et al. (2003) is an independent paper by
members of AMWG, incorporated age-dependence
into the Glassman (1966) probit model using Webb
and Blaser (2002) infectious dose values (ID50 and
ID10) for the ages of 15, 35, 55, and 75 years with
parameter values estimated using least-squares
analysis.
68
-------
AMWG - Anthrax Modeling Working Group 1 or r - fitted parameter, potency estimate in
convened by U. S. Department of Health and Human exponential dose-response model
Services ID - infectious dose, infective dose
CBRN - chemical, biological, radiological, and IDX - infectious dose for x% of individuals
nuclear LDX - lethal dose for x% of individuals
CI - 95% confidence interval NHP - nonhuman primate
DHHS - U.S. Department of Health and Human Pdf - probability density function
Services USAMRIID - U.S. Army Medical Research Institute
EISD - Exposure - Infection - Symptomatic illness - of Infectious Diseases
Death
First reported citations for inhalation anthrax LD50 or ID50 values for a single dose (or less than
24-hour total exposure) ranged from 1,500 spores identified for those older than 65 years of age
(Webb and Blaser, 2002) to approximately 51,000 spores presumably appropriate for a general
population (Franz et al., 1997; Curling et al., 2010; U.S. Army Medical Research Institute of
Infectious Diseases, 2011). The ID50 value of 50,000 spores for the human reported in U.S. Army
Medical Research Institute of Infectious Diseases (2011) (and which was also reported in
previous editions) is generally consistent with nonhuman primate median lethality values
reported by authors with a U.S. Army Medical Research Institute of Infectious Diseases
(USAMRIID) affiliation during the 1990s (Friedlander et al., 1993; Ivins et al., 1996; Ivins et al.,
1998). It is also comparable to the LD50 value of 53,000 spores (single spore size) originally
reported for the nonhuman primate by Druett et al. (1953) and the range of LD50 values (c.
48,750 to 53,500 spores in a single spores cloud) that could be calculated from the Henderson et
al. (1956) control group data. The Druett et al. (1953) data were the basis for the Curling et al.
(2010) LD50 value of approximately 51,000 spores, which is the highest value identified in
Table 5-10.
The classic human LD50 range of 8,000 to 10,000 spores was first published by the Defense
Intelligence Agency (1986) and is commonly cited in the literature, but the original dose-
response data set(s) and study protocol(s) remain unpublished (Coleman et al., 2008). Using a
0
-------
slightly broader range of LD50 values, the U.S. Department of Health and Human Services
(DHHS) Aerosolized Anthrax Response Playbook (U.S. Centers for Disease Control and
Prevention, 2009) estimated that the human LD50 value for inhalation anthrax ranged between
4,100 and 10,000 based on the nonhuman primate study values reported in Glassman (1966),
Peters and Hartley (2002), and Franz et al. (1997). However, U.S. Centers for Disease Control
and Prevention (2009) acknowledged the uncertainty inherent in the range of values for the
human.
Only two studies reported values for response levels less than 50%, including an ID10 range of
1,000 to 2,000 spores derived from expert elicitation (Rickmeier et al., 2001) and an ID10 value
of 1,700 spores (95% confidence interval of 1,100 to 2,600 spores) based on the modeling of a
combination of nonhuman primate and human data (Toth et al., 2013). However, response levels
other than the median lethality value can easily be calculated from reported probit slope values
or empirical models, such as the exponential model.
Anthrax models developed to assess human populations, which incorporated dose-response
elements, are also a source of data for the modeling of human dose-response relationships for
inhalation anthrax. Prior to the 2001 anthrax letter event, the DHHS convened the Anthrax
Modeling Working Group (AMWG) to provide modeling support for recommendations on
medical countermeasures (Hupert et al., 2009). Members of the AMWG published a series of
papers, but noted that the papers were not representative of group consensus or final group
outputs as indicated in Craft et al. (2005) and Wein and Craft (2005). These papers presented
various models to predict necessary medical countermeasures during a disease event, with
human dose-response models or model parameter values (e.g., probit slope) embedded in the
69
-------
overall mathematical models. Two human dose-response models that predicted the probability of
infection as a function of dose and age were developed. Wein et al. (2003) combined a probit
slope model with a quadratic expression describing age-dependency in response, based on the
age-based infection distributions reported in Webb and Blaser (2002). Craft et al. (2005) then
developed a linear model of age dependency from the same data in Webb and Blaser (2002).
Interestingly, the base data for the inhalation anthrax dose-response relationship in these AMWG
members' models were derived from the expert elicitation values reported in Rickmeier et al.
(2001), not dose-response data from animal challenges.
Animal model data has been used as input to semi-quantitatively assess the dose-response
relationship for the human and to identify "threshold" dose levels where infection and disease
may be less likely in identified or general populations. Watson and Keir (1994) identified 6,000
spores as the critical dose for inhalation anthrax infection based on their identification of the
lowest single strain published LD50 value in the nonhuman primate of 6,000 spores (Brachman et
al., 1960). Cohen and Whalen (2007) reported that 600 spores "may not be sufficient to induce
disease" in those exposed unless they exhibited health issues associated with increased
susceptibility to inhalation anthrax. The 600 spore value was based on an estimation of human
exposure using aerosol sampling results from two goat hair mills reported by Dahlgren et al.
(1960) and Brachman et al. (1966). Ho and Duncan (2005) calculated a range of potential
exposure doses after the handling B. anlhracis-contaminated envelopes and reported that
modeled exposure doses associated with human mortality were between 30,000 and 170,000
spores.
70
-------
Qualitative data categorizing human dose-response relationships relative to that reported for
animals were also identified in the literature search. The relatively low overall incidence of
inhalation anthrax in the human in both occupational and general settings led those studying the
issue in England to assert that the human had an "inferior susceptibility" as early as the 1800s
(Gochenour, 1961). This was based on the recognition of many possible exposure sources for the
general public and the acknowledgement of relatively higher source exposure in workers without
universal illness (Gochenour, 1961). The World Health Organization (2008) classified the human
as "moderately resistant" to anthrax (presumably to infection) based on epidemiological data
derived from circumstantial and historical evidence for incidence in wildlife workers, and human
outbreaks. Exposure sources included a mixture of natural and occupational settings, as well as
accidental and intentional releases of manufactured spores. Given the mix of exposure sources,
the use of a single descriptor for human susceptibility implies a generally perceived equivalence
in World Health Organization (2008) in the hazard of infection posed by equivalent exposures of
manufactured or naturally occurring spore products.
In contrast to the "moderately resistant" determination of the World Health Organization (2008),
Lincoln et al. (1967) categorized the rabbit, rhesus monkey, and human as "susceptible" (versus
resistant) to the establishment of anthrax. The susceptible category was defined by relative
differences between susceptible and resistant animal models. Characteristics for placement in
the susceptible category of Lincoln et al. (1967) included lower parenteral and aerosol LD50
doses to establish anthrax, higher number of toxin units to cause lethality by intravenous
injection, higher terminal concentration of bacteremia, greater inhibition of phagocytes by toxin,
and differing rates of intracellular germination by spores in phagocytes in reported values
relative to the resistant group. Since challenge data are unavailable, the human was placed in the
71
-------
susceptible category based on consideration of in vitro results from human-derived cell lines and
the evaluation of limited availability epidemiology data (Lincoln et al., 1967). Examples of
animals identified as resistant to the establishment of anthrax included the rat, swine, and dog
(Lincoln etal., 1967).
With regard to the endpoint of the available dose-response data, all data reported either lethality
or modeled infection with the assumption that infection led to 100% lethality. Human survival of
inhalation anthrax after development of clinical symptoms was reported, but generally after the
use of antibiotics and aggressive medical treatment (Jernigan et al., 2001; Walsh et al., 2007;
Griffith et al., 2014). Survival increased to 55% for those infected with inhalation anthrax as a
consequence of the 2001 anthrax letter event (Inglesby et al., 2002). Historical reports of survival
after inhalation anthrax are relatively rare, though Albrink et al. (1960) reported one suspected
case of inhalation anthrax in the 1957 epidemic that resulted in survival of the individual. The
simplifying assumption that infection is equivalent to lethality has been identified through expert
elicitation (Rickmeier et al., 2001) and included in modeling for bioterrorism medical
preparedness (Hupert et al., 2009) as well as human dose-response modeling (Toth et al., 2013).
Given the scarcity of rigorous data regarding survival after inhalation anthrax infection in the
human, lethality will be used as the endpoint for human inhalation anthrax dose-response
modeling for this report.
5.4.2.4 Rabbit Inhalation Data
Two Key Studies for the rabbit animal model were identified through the literature search, the
single-dose U.S. Environmental Protection Agency (201 la) study and the multiple-dose U.S.
Environmental Protection Agency (2012b) study (Table 5-11). These studies used similar study
72
-------
designs and were categorized as Key Studies. Each study reported data for the endpoints of
infection and lethality, though dose-response calculations were evaluated for lethality only.
Table 5-11. Single- and Multiple-Dose Key Studies for the Rabbit
Study Citation, Rabbit
Breed, and Strain
Key Study Outputs
Modeled Data Identified for Key Study
Single Dose
U.S. Environmental
Protection Agency
(2011a)
New Zealand white
rabbit
Ames strain
Logistic regression model fit to dose
group level logio dose data
Inhaled dose LD50 = 51,800 CFU
(Fieller's CI = 6.14 x l03to 7.27 x 105
CFU)
U.S. Enviromnental Protection Agency
(2012b)
Benchmark dose analysis, dichotomous-
Hill model with individual animal doses
BMDso = 52,000 CFU
BMDLso = 13,000 CFU
BMDio = 5,700 CFU
BMDLio = 1,400 CFU
U.S. Enviromnental Protection Agency
(2014d)
Exponential model with individual animal
doses
r = 7.507 xlO-6
Multiple Dose (Number of Doses and Exposure Duration)
U.S. Enviromnental
Protection Agency
(2012b)
(15 doses over 19 days)
New Zealand white
rabbit
Ames strain
Logistic regression to fit logio
transformed geometric mean inhaled
dose for each individual animal using an
accumulated dose metric
LD50 = 8,100 CFU
(Fieller's CI = 2.3 x 103 to 3.6 x 107
CFU)
Benchmark dose analysis, loge logistic
model with average daily dose
BMDso = 6,800 CFU
BMDLso = 2,600 CFU
BMDio = 760 CFU
BMDLio = 290 CFU
Benclunark dose analysis, loge logistic
model with accumulated dose
BMDso = 120,000 CFU
BMDLso = 44,000 CFU
BMDio = 13,000 CFU
BMDLio = 4,900 CFU
U.S. Enviromnental Protection Agency
(2014d)
Exponential model with individual animal
accumulated daily doses
r=5.243xl0"6
BMDx - benchmark dose for response in x% of individuals
BMDLx - the 95% lower statistical confidence limit of the
BMDx when the 95% lower confidence limit is applied to
the estimated slope parameter value
CFU - colony forming unit(s)
CI - 95% confidence interval
LD50 - lethal dose for 50% of individuals
r - fitted parameter, potency estimate in exponential dose-
response model
73
-------
The U.S. Environmental Protection Agency (201 la) challenge doses ranged from an average
inhaled dose of 286 to 2.75 x 105 CFU. Using logistic regression to fit logio dose single-dose
data at the level of the individual animal, U.S. Environmental Protection Agency (201 la)
reported an LD50 value of 51,800 CFU, with a Fieller's 95% confidence interval that spanned
almost two orders of magnitude (6.14 x 103 to 7.27 x 105 CFU) (Table 5-11). A benchmark dose
(BMD) analysis of these same data in the U.S. Environmental Protection Agency (201 la) study
using a dichotomous-Hill model with individual animal doses was reported in U.S.
Environmental Protection Agency (2012b). The BMD value for response of 50% of the
population (BMD50) value was 52,000 CFU (U.S. Environmental Protection Agency, 2012b).
When using a lethality endpoint, the BMD50 corresponds to the LD50 of the population. The
benchmark dose limit value (BMDL) represents the 95% lower statistical confidence limit of the
BMD when the 95% lower confidence limit is applied to the estimated slope parameter value for
50% response (BMDL50) value. The BMDL50 for the U.S. Environmental Protection Agency
(201 la) was 13,000 CFU U.S. Environmental Protection Agency (2012b).
A re-analysis of the U.S. Environmental Protection Agency (201 la) data using individual animal
doses in the exponential model reported an r value (fitted potency parameter for the exponential
model) of 7.507 x 10"6 (U.S. Environmental Protection Agency, 2014d), which would calculate
an LD50 value of approximately 92,000 CFU. Gutting et al. (2013) analyzed a combination of
data sets [i.e., U.S. Environmental Protection Agency (201 la), Zaucha et al. (1998), and
previously unpublished data] and reported a fitted potency parameter value3 for the exponential
3 Depending on the modeler and/or cited publication, the fitted parameter value for the exponential model is
identified as an r, k, or 1 parameter. Regardless of the term used to identify the fitted potency parameter for the
exponential model, it represents the same value.
74
-------
model of 7.22 x 10"6, which was generally similar to that reported in U.S. Environmental
Protection Agency (2014d). One possible reason for the two-fold disparity LD50 values for the
same data set is the probable lower quality fit of the exponential model relative to the
dichotomous-Hill or logistic regression models. U.S. Environmental Protection Agency (2012b)
evaluated the exponential model in the suite of evaluated models and this model was not the best
fitting model of those assessed.
No single-dose data for the rabbit were categorized as Supporting Studies. Single-dose dose-
response data categorized as Additional Data for the rabbit are provided in Appendix D.
The most cited rabbit LD50 value of 1.05 x 105 originated from the Zaucha et al. (1998) study,
though the original dose-response data set was not published until Gutting et al. (2013). The
Zaucha et al. (1998) LD50 value is based on a challenge of 50 animals with mean group doses of
98 to 713,000 spores (Gutting et al., 2013). The Zaucha et al. (1998) value has been directly cited
or others have reported values that differ only by varying adjustments in the number of
significant figures (see Appendix D for the Additional Data Table). The Zaucha et al. (1998)
study was categorized as Additional Data due to: (1) the lack of response data in the range
between 1% and 49%, (2) particle size data were not associated with the study exposures for
which the LD50 value was derived, and (3) it was assumed that the inhalation rate was
determined via plethysmography but prior to the aerosol challenge. The dose spacing and the
lack of responses between 0 and 50% lethality are problematic because there are insufficient data
to differentiate between possible mathematical dose-response models based on the fit to the
observable data.
75
-------
One multiple-dose study in the rabbit was identified through the literature search. The U.S.
Environmental Protection Agency (2012b) multiple-dose study in the rabbit was categorized as a
Key Study. In this study, rabbits were challenged with 15 doses over 19 days (i.e., Monday
through Friday dosing, with no doses over the weekend). Using logistic regression to fit logio
transformed geometric mean inhaled dose data for each individual animal, U.S. Environmental
Protection Agency (2012b) reported an LD50 value for the accumulated dose metric of 8,100
CFU with a Fieller's 95% confidence interval that spanned approximately four orders of
magnitude (2.3 x 103 to 3.6 x 107 CFU). Using the U.S. Environmental Protection Agency
(2012b) data set and a calculated average daily dose derived using the exposure duration of the
challenge, a benchmark dose analysis identified the best fitting model as the loge logistic and
reported a BMD50 of 6,800 CFU and a BMDL50 of 2.60 x 103 CFU (U.S. Environmental
Protection Agency, 2012b). The same BMD analysis process using the loge logistic model and an
accumulated dose metric reported a BMD50 of 120,000 CFU and a BMDL50 of 44,000 CFU (U.S.
Environmental Protection Agency, 2012b).
The U.S. Environmental Protection Agency (2012b) data were reanalyzed using individual
animal accumulated doses and the exponential model; a r value of 5.243 x 10"6 was reported
(U.S. Environmental Protection Agency, 2014d). This r value would derive an LD50 value of
approximately 132,000 CFU. The LD50 value calculated by the U.S. Environmental Protection
Agency (2012b) was considerably lower than that reported for the BMD50 value in U.S.
Environmental Protection Agency (2012b) or the LD50 value calculated from the r value reported
in U.S. Environmental Protection Agency (2014d). No additional multiple-dose dose-response
data for the rabbit were identified.
76
-------
5.4.2.5 Inhalation Data for the Nonhuman Primate
One Key Study for the nonhuman primate was identified through the literature search
(Table 5-12). Lever et al. (2008) challenged a group of 12 male and female common marmoset
{Callithrix jacchus) with a range of inhaled doses from 1.4 x 101 to 1.9 x 105 CFU and a reported
LD50 value of 1.47 x 103 CFU (95% confidence interval of 7.19 x 105 to 2.95 x 105 CFU). The
marmoset animal model was evaluated as a small animal alternative in the nonhuman primate
animal model. The endpoint assessed was lethality between the exposure challenge and 10 days
after exposure. Infection was not reported.
The Lever et al. (2008) data set was categorized as a Key Study. The judgment was made that the
study was sufficiently close to meeting the requirement of having an overall lethality rate of less
than 50% (i.e., 6 of 12 monkeys died). Though a higher number of animals in the low-dose
region of exposure may have been preferred, the available data were sufficient to derive the
reported LD50 value.
Table 5-12. Single-Dose Key Study for the Nonhuman Primate
,
^¦
Single Dose
Lever et al. (2008)
Common marmoset
('Callithrix jacchus)
Ames strain
Geometric mean LD50 =
1.47 x 103 CFU
CI = 7.19 x 105 to 2.95 x
105 CFU
No additional data
No additional data
CFU - colony forming unit(s) LD50 - median lethality value
CI - 95% confidence interval
77
-------
Single dose data for the nonhuman primate characterized as Supportive Data or Additional Data
are provided in Appendix E.
An additional consideration for the use of the nonhuman primate data is that the identified LD50
values categorized as Additional Data must be carefully evaluated prior to use for informing risk
assessment. It is important to recognize that most values were derived from studies with the
primary purpose of evaluating pathology or medical countermeasures; the LD50 values were
generated with study designs that did not explicitly evaluate statistical considerations regarding
animal and dose range to generate a representative median value. With the exception of the
Vasconcelos et al. (2003) LD50 value, the remaining identified values in the 50,000 to 62,000
CFU range were cited as a personal communication or unpublished data from an author
associated with the USAMRIID laboratories. Examples of publications fitting this description
include Ivins et al. (1996), Vasconcelos et al. (2003), and Coleman et al. (2008). Other examples
include those directly cited by an author with USAMRIID affiliation as in the case of Henderson
et al. (1956) and Friedlander et al. (1993). It is possible that multiple published citations of
approximately the same LD50 value may not represent multiple independent studies that
corroborate the identified value, but may be the same study or a limited number of studies
repeatedly cited.
Two multiple-dose studies (Albrink and Goodlow, 1959; Brachman et al., 1966) and subsequent
re-analyses of these data were identified in the literature search (Table 5-13). Brachman et al.
(1966) reported selected dose data from three multiple-dose exposure challenges of nonhuman
primates to B. anthracis-coni&minated aerosols from a picking station at a goat hair processing
plant (Table 5-13). The total cumulative doses ranged from 947 to 16,962 B. anthracis-beax'mg
77
-------
particles. The exposure duration varied from a low of 31 hours for Run 5 to up to 47 days for the
first segment of Run 3. The endpoint reported was lethality as measured over an observation
period that varied for each run; with a low of two to five days for Run 3 to up to 25 days for the
first challenge in Run 5. Infection was not reported, though evidence of anthrax infection was
noted in individual animals at sacrifice.
Brachman et al. (1966) graphically reported daily cumulative dosing with an accompanying
identification of animal deaths from anthrax. The original raw dose-response data set was not
published and has not since become available. After interpolation of the graphical data to
identify values for modeling, the Brachman et al. (1966) data were reanalyzed by Haas (2002)
and Mayer et al. (2011). Mayer et al. (2011) and Haas (2002) reported fitted values for the
potency parameter in the exponential model that can be used to calculate LD50 values of 19,327
spores and 28,750 spores, respectively. The higher LD50 value of 28,750 likely resulted from an
error in the calculation of the average daily dose by Haas (2002) that was identified in Toth et al.
(2013).
Most reported studies identified were performed to determine the median lethality endpoint,
assess efficacy of medical countermeasures, or describe the pathology resulting from lethal
infection, but not to identify dose-response relationships for infection from low- or high-dose
nonhuman primate study data (Albrink and Goodlow, 1959; Ivins et al., 1996; Ivins et al., 1998;
Fellows et al., 2001; Rossi et al., 2008; Saile et al., 2011; Henning et al., 2012). However,
survival after anthrax bacteremia in animal models appears to be rare relative to lethality in the
dose ranges commonly tested. Published reports of survival after anthrax bacteremia were
identified during the literature search in the unvaccinated nonhuman primate (Albrink and
78
-------
Goodlow, 1959; Fellows et al., 2001; Saile et al., 2011; Henning et al., 2012) and the
unvaccinated rabbit (U.S. Environmental Protection Agency, 2012b). Given the lack of research
interest in the survival endpoint after infection, study designs did not incorporate statistical
sufficiency to estimate the likelihood of survival after bacteremia. This would likely entail the
need for significantly higher animal numbers to reliably measure prevalence. It is also unknown
if there is dose-dependence in survival after infection.
Table 5-13. Multiple-Dose Additional Data for the Nonhuman Primate
Study, Nonhuman Primate, and
Model Parameters or Other Outputs
Brachman et al. (1966)
Cynomolgus monkey
(Macaca fascicularis)
Reanalyzed by Haas (2002)
Exponential model
k = 2.6 x 10"5
CI = 1.3 to 1.6 x 10"5
Reanalyzed by Toth et al. (2013)
EISD model
Assumed fixed model parameters for clearance
where 0 = 0.07 day"1, shape parameter a = 5.43,
scale parameter b = 0.864, and then fit an r value
of 6.4 x lO 5, and T = 2.3 days
Reanalyzed by Mayer et al. (2011)
Exponential model
k = 3.57 x lO 5 when assuming a = 1.0
Also derived time-dependent modification for
exponential model, a = 0.9, y = 0.0097 h1, and s =
1.81 x 10-7 h"1 with s/y = 1.87 x 10~5 where s/y is
mathematically equivalent to the k potency
estimate in exponential equation
Other Data
Albrink and Goodlow (1959)
Chimpanzee
(Pan troglodytes Schwarz and Pan troglodytes troglodytes)
MelvinDose 1: 32,800 inhaled viable spores
Dose 2: 90,300 inhaled viable spores with survival after
Dose 2
John Dose 1: 34,350 inhaled viable spores
Dose 2: 112,000 inhaled viable spores with death after
Dose 2
Brachman et al. (1966)
Cynomolgus monkey
(Macaca fascicularis)
Daily doses not reported, 3 exposure runs of various
lengths < 47 days with reported exposure data, differing
exposure sources and concentrations
Run Three: 16,962 total B. anthracis particles over 47 days
Run Four: 4,959 total B. anthracis particles over 41 days
Run Five: 947 total B. anthracis particles over 55 hours +
1,347 total B. anthracis particles over 31 hours
Fatality rate of approximately 10% for exposure to
approximately 1,000 B. anthracis-bearing particles over 3
to 5 days, with fatality rate of 20 to 25% for exposure to
approximately 3,500 to 5.500 II anthracis-bearing particles
over a 5 days
79
-------
Study, Nonhuman Primate, and
Model Parameters or Other Outputs
Other Data
CI - 95% confidence interval
EISD - Expo sure-Infection-Symptomatic Illness-Death
IDX - Infectious Dose for x percent exposed
k or r - fitted parameter, potency estimate in exponential
dose-response model
a - shape parameter
a - shaping parameter for accumulation effects
b - scale parameter
y - net per pathogen clearance rate (lv1)
© - probability-per-time for clearance of spores from the lung
s - instantaneous risk to individual pathogen
T - delay between spore germination and initiation of symptoms
80
-------
5.4.2.6 Oral Data for Multiple Animal Models
Few published data are available for oral exposure to B. cmthrcicis spores or vegetative bacteria
(Table 5-14). Oral challenge dose-response data were identified for the guinea pig, rabbit, rhesus
monkey, cow, mouse, pig, and human (Table 5-14). Published LD50 values for oral exposure
generally range from 106 to 108 spores, and include data from animals that are viewed as very
susceptible to infection (World Health Organization, 2008). For example, Schlingman et al.
(1956) reported that a group of three cattle challenged with oral doses of 107 spores had one
survivor, and exhibited a longer time-to-death after exposure than 108 and 109 spore doses.
Table 5-14. Oral Dose-Response Data
Study
Animal Model
(Form of B. anthracis)
Dose-Response Data
Young Jr. et al. (1946)
Guinea pig (Spores)
Occasional deaths in tested guinea pigs after oral administration of 1 x 108 Detrick
25 strain spores, details not provided on death number or total tested.
Druett et al. (1953)
Guinea pig (Spores)
No infections after oral administration of 108 strain spores in unspecified number of
animals (assumed to be same M36 strain used in aerosol challenge).
Druett et al. (1953)
Rabbit (Spores)
No infections after oral administration of 108 strain spores in unspecified number of
animals (assumed to be same M36 strain used in aerosol challenge).
Lincoln et al. (1965)
Rhesus monkey (Spores)
Two monkeys each were orally challenged using infant feeding tubes at doses of
102, 104, 106, and 108 spores, all animals survived (assumed to be same Vlb strain
used in aerosol challenge).
Redmond et al. (1997)
Large White x Landrace
Crossbred pig
(Spores with feed and grit)
Two of 50 pigs died that were challenged with total doses (delivered in one to three
doses) of approximately 107 to 10UI CFU of Ames strain or reisolates from pigs
infected with same Ames strain, grit was added to feed to facilitate infection.
Schlingman et al. (1956)
Mixed dairy and Hereford
breeds of cattle
(Spores)
One cow administered 6 x 108 Vollum strain spores in gelatin capsule exhibited an
elevated temperature for days 5 through 9, then recovered.
Three of 4 cattle administered 109 V770-2-P strain spores in a feed pellet died, the
surviving cow was rechallenged with same dose (timing unknown) and promptly
died.
One cow was challenged with 109 Vollum strain spores, exhibited a slight febrile
reaction and recovered, was rechallenged with 109 V770-2-P strain spores and
exhibited no evidence of infection.
Three of 4 cattle administered 109 V770-2-P strain spores in a feed pellet died, the
survivor was rechallenged 7 days later and survived.
In a rechallenge taking place an unknown time after the first exposure, two of 2
cattle died that were administered 109 V770-2-P strain spores in a feed pellet.
Two of 3 cattle administered 108 V770-2-P strain spores in a feed pellet died, the
surviving cow was noted to not chew the pellet, was rechallenged 10 days later, and
survived.
Two of 3 cattle administered 107 V770-2-P strain spores in a feed pellet died, the
surviving cow had an elevated temperature days 2 through 6 and survived.
79
-------
Study
Animal Model
(Form of B. anthracis)
Dose-Response Data
Two of 3 cows administered 1.5 x io8 V770-2-P strain spores in a feed pellet
survived, the survivor was rechallenged after 7 days with the same dose of the
initial challenge and died.
One cow challenged with 109 V770-2-P strain spores died.
Schlingman et al. (1956)
Chester White pig (Spores)
No evidence of infection after oral administration of 106 V770-2-P strain spores in
pigs (number assumed to be 15).
Schlingman et al. (1956)
Chester White pig
(Likely mixture of
vegetative and spore
forms)
Two of 2 pigs that were fed 1 guinea pig recently dead of anthrax infection from
either V770-2-P or 1062 strain survived, with each exhibiting fever.
Of the 8 swine in the control groups, all swine survived ingestion of guinea pig
carcasses that died from anthrax infection with V770-2-P strain spores, though all
pigs exhibited symptoms of elevated temperature, with some pigs noted to exhibit
pharyngeal swelling and anorexia.
Xie et al. (2013)
A/J mouse (Vegetative
bacteria)
LD50 = 2.3 x io7 for Sterne strain authors noted that dose of 2.3 x 106vegetative
bacteria can cause lethal infection.
When designing testing, an oral dose of 1.5 x io8 spores was thought to prove fatal to most
unimmunized cattle (Schlingman et al., 1956). The pig is known to exhibit a high degree of
resistance to systemic anthrax infection from the inhalation and intraperitoneal challenge routes
(Walker et al., 1967). Accordingly, high oral dose levels ranging from 107 to IO10 CFU were
associated with very low levels of lethality (2/50) in the challenged swine even with the addition
of grit to the food source (Redmond et al., 1997). In two separate studies, an oral challenge dose
as high as 108 spores did not result in infection in the rabbit (Druett et al., 1953) or the nonhuman
primate (Lincoln et al., 1965) (Table 5-14).
Of the few oral challenge studies available, most have been conducted with spores. The relative
infectivity of spores versus vegetative bacteria has been characterized as unknown (World Health
Organization, 2008). However, relatively new data in the mouse animal model from Xie et al.
(2013) described infection lethality in doses as low as 2.3 x io6 CFU and reported an LD50 value
of 2.3 x 107 CFU. Noting increased infectivity of vegetative bacteria in subcutaneous challenge
in the same animal model, Xie et al. (2013) hypothesized that vegetative bacteria toxin
80
-------
production could contribute to breakdowns in the epithelial barrier and promote infection and
dissemination. Data are unavailable to draw conclusions on the relative human infectivity of the
spore versus vegetative bacterial forms in the human beyond that hypothesized by Inglesby et al.
(2002), that large oral doses of vegetative bacteria may be necessary to result in gastrointestinal
anthrax.
5.4.2.7 Conclusions for Dose-Response Data Literature Review
Few studies were identified as Key Studies for the rabbit and nonhuman primate; there were no
Key Studies or Supporting Studies identified for the human. The two Key Studies for the rabbit
were the single-dose U.S. Environmental Protection Agency (201 la) study and the multiple-dose
U.S. Environmental Protection Agency (2012b) study. No studies were categorized as
Supporting Studies. For the nonhuman primate, one single-dose Key Study (Lever et al., 2008)
and one single-dose Supporting Study (Druett et al., 1953) were identified.
Table 5-15. Summary of Number of Key Studies, Supporting Studies, and Additional Data
Sources for the Human, Rabbit, and Nonhuman Primate
Number of
Key Studies
(Table)
Study Citation
Number of Supporting
Studies
(Table)
Study Citation
Number of Sources of
Additional Data
(Table)
Human
0
0
15 single dose*
(Table 5-9)
1 single dose
(Table 5-10)
U.S. Enviromnental
Rabbit
Protection Agency (201 la)
1 multiple dose
(Table 5-10)
U.S. Enviromnental
Protection Agency (2012b)
0
15 Single Dose
(Table Appendix D-l)
Nonhuman primate
1 Single Dose
(Table 5-11)
Lever et al. (2008)
1 Single Dose
(Table E-l)
Druett et al. (1953)
5 Single Dose
(Table Appendix E-l)
4 Multiple Dose
(Table 5-12)
* For the human Additional Data, some of the input data reported as the basis for the published data were derived
using multiple dose data in full or in part. However, the Additional Data sources did not clearly specify whether the
modeled values should be applied to single or multiple dose exposures.
81
-------
The results of the literature search indicate that the lethality endpoint is the only endpoint that
can be supported with identified data for inhalation anthrax in the rabbit, nonhuman primate, and
human. This is identified for the following reasons: (1) the high concordance between infection
and death for challenge studies in animal models and human epidemiological reports, (2) very
few studies that report infection data, and (3) lack of appropriate study design to capture the
incidence of nonlethal infection.
Table 5-16 reviews the Twenhafel (2010) key human histopathological findings relative to the
pathology reported in the Rabbit and Nonhuman Primate Key Studies. There was not a good
concordance between the key human histopathological findings and identified Key Studies for
the rabbit and nonhuman primate, with 3/6 of the histopathological findings reported in the rabbit
and 2/6 of the histopathological findings reported in the nonhuman primate. The lack of
concordance may include differing protocols for the pathology evaluations and/or relatively
small animal numbers of the animals evaluated for pathology in the dose-response studies. The
Key Studies identified for the rabbit did not report findings for pathology in the spleen. Given
that the spleen is one of the earlier involved organs in the disease process, it is unexpected that
the spleen did not show early signs of lesions or other pathologies, even if it had not progressed
to splenic lymphoid depletion. However, the protocol triggered histopathology on those organs
that exhibited gross pathology at necropsy. This could explain the lack of even initial stages of
splenic pathology reported.
When comparing the results contained in Table 5-16 with
82
-------
Table 5-9, the Zaucha et al. (1998) study for the rabbit reported concordance with 4/6 of the
histopathology findings. However, the Zaucha et al. (1998) also had the largest animal number
examined for pathology (n=22 for aerosol challenged rabbits). This is less than the six nonhuman
primates that died of inhalation anthrax in Lever et al. (2008) and eleven and five rabbits in the
U.S. Environmental Protection Agency (201 la) U.S. Environmental Protection Agency (2012b)
that died of inhalation anthrax. It is possible that the low animal numbers evaluated may affect
the presentation of specific pathologies in inhalation anthrax, as noted by Lever et al. (2008) with
regard to the lack of meningitis with hemorrhage in the study. Given that some of the Twenhafel
(2010) histopathological findings may be infrequent in the human and having variability in
appearance in studies, the low animal numbers in the studies may be a compelling explanation.
Table 5-16. Identification of Twenhafel (2010) Key Human Histopathological Findings in
Rabbit and Nonhuman Primate Key Studies
Rabbit Key Studies
• U.S. Environmental
Nonhuman Primate Key
Evidence for Time-
Histopathology
Protection Agency (2011a)
Study
Dependency Indicated in
• U.S. Environmental
• Lever et al. (2008)
Table 5-9?
Protection Agency (2012b)
Pneumonia
Yes*
Not reported in Key Study
Yes - Progression to
pneumonia is associated
with inflammatory process,
lower incidence, and lesser
severity reported in rabbit.
Splenic
lymphoid
depletion
Not reported in Key Studies
Not reported in Key Study
No - Spleen is an early
disease target in inhalation
anthrax.
Meningitis
Yes
Yes
Yes - Hypothesized as
time-dependent in Zaucha
et al. (1998), not identified
in any of NHP serial
sacrifices reported in
Berdjis et al. (1962).
Hepatic
hemorrhage or
inflammation
Not reported in Key Studies
Yes
Yes - Reported as time-
dependent in NHP by
Vasconcelos et al. (2003).
Gastrointestinal
hemorrhage or
inflammation
Yes
Not Reported in Key Study
No - Hemorrhagic spread
to gastrointestinal tract
seems to occur early in the
disease process.
83
-------
Histopathology
Rabbit Key Studies
• U.S. Environmental
Protection Agency (2011a)
• U.S. Environmental
Protection Agency (2012b)
Nonhuman Primate Key
Study
• Lever et al. (2008)
Evidence for Time-
Dependency Indicated in
Table 5-9?
Urogenital
hemorrhage or
inflammation
Not reported in Key Studies
Not Reported in Key Study
Unknown - Evidence or
reports for time-
dependency are lacking.
* Reported suppurative inflammation in pulmonary interstitium
NHP - nonhuman primate
5.5 Model the Dose-Response Relationship
There are a number of considerations necessary to model a dose-response relationship for B.
cmthrcicis once determinations have been made regarding suitable animal models and available
dose-response data gathered. Table 5-17 indicates the key questions and associated report
sections in which available data and dose-response analysis processes are reviewed for B.
cmthrcicis. The following sections will consider identification of appropriate dose metrics,
empirical and mechanistic approaches for modeling dose-response relationships for B. cmthrcicis,
and mathematically modeling the dose-response relationship.
Table 5-17. Development of Microbial Dose-Response Relationships
Step in Microbial Dose-
Response Analysis
Key Questions
Report Section
Model the dose-response
relationship
(Section 5.5)
What dose metrics can be supported based
on available disease pathogenesis and other
dose-response data?
What assumptions are associated with a
given dose metric?
Section 5.5.1 Determination of Dose
Metric
What types of empirical and mechanistic
models may be suitable for B. anthracisl
Can mechanistic models be supported by
available dose-response data for B.
anthracisl
Section 5.5.2 Empirical and
Mechanistic Modeling Approaches
What approaches can be used to
mathematically model the dose-response
relationship and estimate the POD?
Section 5.5.3 Mathematically
Modeling the Microbial Dose-
Response Relationship
84
-------
5.5.1 Determination of Dose Metric
A dose metric is the mathematical description of the challenge study dose that is used to model
the dose-response relationship and conduct the interspecies extrapolation. The preferred dose
metric is the internal dose that can be most closely mechanistically or otherwise correlated with
the biological endpoint of interest (Jarabek et al., 2005). A dose metric is associated with a
specified exposure duration and can also be expressed as a time-normalized measurement (e.g.,
CFU/day) (U.S. Environmental Protection Agency, 2014b). Dose metrics may also include a
"biologically motivated" normalization factor that assesses the dose magnitude over an identified
tissue area or cell number (e.g., number of macrophages contacting the particle) (Jarabek et al.,
2005).
There are a range of potential dose metrics for inhalation B. cmthracis
exposure ranging from administered dose to differing measures of
internal dose (e.g., deposited dose, dose accessible by macrophages)
Summary of Findings for
Determination of Dose
Metric
(U.S. Environmental Protection Agency, 2010a). The dose metric
Environmental Protection Agency, 201 la; Gutting et al., 2013) and the
selected for the single-dose B. cmthracis dose-response studies (U.S.
• There is a lack of
mechanistic data relating
dose to the lethality
endpoint.
multiple-dose study (U.S. Environmental Protection Agency, 2012b)
• Uncertainty in the
initiation of infection adds
to the difficulty in dose
metric selection.
was an inhaled dose metric.
• There is uncertainty in the
selection of an appropriate
dose metric when
evaluating multiple-dose
exposure of microbial
pathogens, including B.
cmthracis.
Uncertainty in the most appropriate internal dose for the endpoint of
lethality poses a challenge in the selection of dose metrics. If it is
assumed that initiation of infection is the key event most closely
associated with the endpoint of lethality and that initiation of infection
takes place in the alveolar lung region (e.g., Trojan horse model), one
85
-------
appropriate measure of the internal tissue dose is the deposited dose in the alveolar region. If
infection is assumed to initiate across a variety of respiratory tract tissues and the likelihood of
initiation of infection across tissue or regions is unknown (e.g. Jailbreak model), multiple-dose
metrics may be appropriate for consideration. Though not evaluated to date for B. anthracis,
dose metric selection can incorporate a normalization factor. For evaluation of inhaled
particulate chemical hazards, normalization factors have described the magnitude of dose relative
to the number of contacting cells with potential to initiate infection (e.g., macrophage) or surface
area available for uptake of chemical (Jarabek et al., 2005).
There is also uncertainty in the selection of an appropriate dose metric when evaluating multiple-
dose exposure of microbial pathogens, including B. anthracis. The U.S. Environmental
Protection Agency (2012b) multiple-dose study reported dose-response relationship evaluations
using two dose metrics: accumulated inhaled dose and average daily inhaled dose. An
accumulated dose metric assumes an equivalent hazard whether the intake is in the form of one
dose or in many doses over that same time (i.e., the hazard assumed per spore is equivalent
regardless of the dosing schedule) (Mayer et al., 2011).4 The independent action hypothesis, also
termed the independent event hypothesis, may have relevance for the determination of dose
metrics for multiple-dose B. anthracis exposure studies (U.S. Environmental Protection Agency,
2014d). Independent action of pathogens was described by Druett (1952) as a constant
relationship between response and the product of administered dose (e.g., environmental
concentration) and exposure time. In the derivation of the independent action hypothesis, Druett
4 Druett (1952) independently described the microbial-equivalent of Haber's Law. Haber's Law, reported in the
early 1900s, also described a constant concentration-time relationship between exposure and mortality response for
exposure to inhalation exposure to volatile chemicals. Since that time, Haber's Law has been updated to include a
fitted exponent on the concentration term to better fit tested chemicals (ten Berge et al., 1986). Likewise, a fitted
exponent may also be found appropriate for the mathematical description of independent action.
86
-------
(1952) assumed the following: (1) a constant probability for each organism to cause the
identified response (i.e., mortality or infection) in the host, (2) independent action of each
organism (e.g., no immune system activation), (3) an LD50 value that can be determined, and (4)
a large homogenous experimental population (Druett, 1952). The independent action hypothesis
has been further defined to indicate that the probability of survival of each individual organism is
the same (Haas et al., 1999b) and that the probability of an individual organism causing infection
is independent of the number ingested or inhaled (Buchanan et al., 2009). Relevant to the
consideration of multiple-dose exposure, the definition of the independent action hypothesis has
been expanded to include a lack of effect of prior doses on subsequent dose (U.S. Environmental
Protection Agency, 2014d).
If the independent action hypothesis were correct, the accumulated (or total) dose would be an
appropriate dose metric for a B. anthracis and there is no biological justification for
consideration of a daily average dose. However, a limitation to the exposure duration over which
independent action could be assumed (e.g., short enough to preclude immune system activation)
was noted by Druett (1952) in the original formulation of the hypothesis. Though Druett (1952)
developed the hypothesis with single-dose data, the concept should be equally relevant to
multiple-dose assessments. The independent action hypothesis allows for the use of an aggregate
dose metric only if the exposure time over which the daily doses are aggregated does not exceed
the time duration associated with dose independence.
Potential dependencies by time, dose, or route of exposure may affect consistency with the
independent action hypothesis. The magnitude of exposure or exposure duration (Mayer et al.,
2011; U.S. Environmental Protection Agency, 2014d) where independent doses can be
delineated from dependent doses have not been explicitly evaluated to date. Dose-dependencies
87
-------
may be present in the expression of independent action if larger doses could affect response to
subsequent doses when overloading of clearance or other innate immune functions are affected
(Mayer et al., 2011). If overloading can occur, this implies that the presence of independent
action could vary by route of exposure if varying innate response levels are present (e.g., dermal
exposure versus inhalation). The timing of the exposures relative to the dose and clearance
capabilities is also a critical exposure consideration relative to the selection of dose metrics
(Mayer et al., 2011).
The determination of a theoretical time point separating independent and dependent doses may
be considerably more complicated for inhaled pathogens that have the potential to persist in the
lungs (U.S. Environmental Protection Agency, 2014d). For example, spore persistence in the
lung and subsequent inhalation anthrax has been reported in one nonhuman primate that died 58
days after exposure after initially receiving 30 days of antibiotic treatment starting on the
exposure day (Friedlander et al., 1993).
Though there is uncertainty in the identification of the most appropriate dose metric, this should
not limit the evaluation of dose-response relationships. Relevant dose metrics should be
identified and a justification provided for those that are evaluated. With regard to selection of the
regional deposition location(s) for the deposited dose, multiple-dose metrics can be evaluated.
Given that the differences in deposition may be small relative to other components of the
inhalation dose calculation, the actual difference in the modeled dose-response relationship may
be of limited magnitude. The documentation for the dose-response relationship should include a
transparent identification of the basis for selection of the dose metric(s) considered. There should
also be a qualitative discussion of the uncertainty associated with the dose metric selection in the
risk characterization element of the risk assessment.
88
-------
5.5.2 Empirical and Mechanistic Modeling Approaches
Two fundamental types of dose-response modeling approaches are
available to derive microbial dose-response relationships.
Empirical models, also termed fitted models (Gutting et al., 2008),
rely on statistical curve-fitting techniques to fit mathematical
models to dose-response data. Depending on the model, parameter
values fit by these models may not have biological meaning or bear
precise relationships to measurable biological parameters
(Andersen et al., 1999). It is recommended that microbial dose-
response models exhibit biological plausibility, which is defined as
a biological basis for the mathematical representation of the model
(Haas et al., 1999a). However, the ability to precisely describe
biological plausibility may be limited due to lack of basic
mechanistic data for microbial pathogenesis (Taft and Hines, 2012).
Empirical models have considerable utility in dose-response modeling because they allow for the
description of a wide variety of curve shapes, provide a general assessment of potency (or
virulence for pathogens), and can assess time-based elements of the test system (Andersen et al.,
1999). Empirical models can interpolate within the original range of the study data, but may
provide unreliable extrapolations to lower or higher doses (Buchanan et al., 2000; Gutting et al.,
2008). The primary value of empirical models is to provide a first step in the identification of
dose-response relationships when scarce mechanistic and parameter value data limit the use of
other approaches.
Summary of Findings for
Empirical and Mechanistic
Modeling Approach es
• There is insufficient
mechanistic data for
comprehensive
mechanistic dose-
response models for B.
anthrctcis.
• Parsimony in model
selection will lead to the
continued use of empirical
models and limited or
nominally mechanistic
models.
89
-------
Most of the microbial dose-response modeling conducted for B. anthracis to date has relied on
empirical modeling approaches. The probit slope and median lethality values reported by Druett
et al. (1953) are an early example of empirical dose-response modeling. Empirically derived
dose-response relationships using either inhaled or deposited dose metric have since been
reported for B. anthracis inhalation exposure in the nonhuman primate (Glassman, 1966; Haas,
2002; Bartrand et al., 2008; Weir and Haas, 2011; Taft and Hines, 2012) and rabbit (U.S.
Environmental Protection Agency, 201 la, 2012b). Hybrid models of empirically fit parameters
combined with expert elicited dose-response values were published in population-based anthrax
models for the human (Webb and Blaser, 2002; Wein et al., 2003; Wein and Craft, 2005).
Likewise, empirically fit models have been developed using a survival analysis framework to
incorporate time-dependencies in dosing and/or response (Mayer et al., 2011; U.S.
Environmental Protection Agency, 2014d).
In contrast to empirical models, mechanistic models incorporate known or hypothesized
biological mechanisms to derive an estimate of predicted response (U.S. Environmental
Protection Agency, 201 lc). Mechanistic models can be extremely data-intensive and rely upon
significant mechanistic knowledge of the microbial pathogen and host (Gutting et al., 2008).
However, mechanistic models offer a unique advantage over empirical models as they can allow
for more robust extrapolation across species and dose ranges of interest (Gutting et al., 2008).
The biologically based dose-response (BBDR) model is a mechanistic model, but has the
distinguishing trait where the probability of response to an administered dose is modeled as a
function of biological variable(s) that are mechanistically associated with the adverse response
(Crump et al., 2010).
90
-------
There are differences of technical opinion in the microbial dose-response community as to
whether the exponential and beta-Poisson dose-response models should be identified as
empirical or mechanistic (U.S. Environmental Protection Agency, 2010b). The determination
that exponential and beta-Poisson models are mechanistic has been used as the basis to exclude
consideration of empirical models. However, uncertainty in the basic pathogenesis of B.
anthracis and conflicting evidence for the presence of identified disease pathogenesis
characteristics used to define the model as mechanistic (e.g., independent action, no threshold,
assumed particle distribution) should prompt consideration of both empirical and mechanistic
dose-response modeling approaches to reduce the potential impact of model uncertainty (Taft
and Hines, 2012).
To facilitate clarity in the discussion of mechanistic models, a hierarchy of mechanistic models is
presented that defines models relative to the level of biological knowledge incorporated in the
model. The conceptual basis for the three-part delineation is based on the dose-response model
description described in Andersen et al. (1999). The hierarchy of mechanistic models,
terminology for category of model, and published models for each category are identified in
Figure 5-2 and summarized below:
(1) Nominally mechanistic models incorporate simple biological representations, but
biological measurements or modeling cannot inform parameter values; all parameter
values are derived through empirically fitting the dose-response data to a mathematical
model; an example is the exponential model described by Haas et al. (1999a),
(2) Limited mechanistic models, including BBDR models, incorporate mechanistic
assumptions and data that can be derived or informed by biological measurements,
91
-------
examples include the competing risk model of Gutting et al. (2008) and biokinetic model
of Huang and Haas (2011), and
(3) Comprehensive mechanistic models incorporate mechanistic assumptions and data to
fully describe biodynamic and biokinetic elements, the earliest conceptualization of a
microbial-equivalent to the physiologically based pharmacokinetic [PBPK] model
generated for chemical hazards was proposed by Coleman and Marks (1998), and a
compartmental and data description for a physiologically based biokinetic [PBBK] model
specific for B. anthracis was subsequently described by Gutting et al. (2008).
Type
Parameters
Complexity
Comprehensive Mechanistic Model
• Full incorporation of biodynamic and
biokinetic elements
Biological Representation
Reliance on Empirical
Curve Fitting
///////////////
• ••
Limited Mechanistic Model
• Simple biological representation of
mechanism
• Limited potential to incorporate selected
mechanistic information derived from
biological measurements
Biological Representation |
Reliance on Empirical WMHHMMMU
Curve Fitting 1
• •
Nominally Mechanistic Model
• Simple biological representation of
mechanism
• Parameter values derived through fitting
empirical data to the mathematical model
• Biological measurements or models do not
inform parameter values
Biological Representation
Reliance on Empirical
Curve Fitting
•
Figure 5-2. Comparison of mechanistic models relative to biological representation,
empirical curve-fitting, and complexity.
Nominally mechanistic dose-response models (i.e., exponential and beta-Poisson) were evaluated
for inhalation exposure to B. anthracis in the nonhuman primate (Haas, 2002; Bartrand et al.,
2008; Weir and Haas, 2011; Taft and Hines, 2012) and the rabbit (U.S. Environmental Protection
92
-------
Agency, 2012b, 2014d). Using the competing risk mathematical model to describe the likelihood
of successful spore germination versus clearance (Brookmeyer et al., 2003; Brookmeyer et al.,
2005), limited mechanistic BBDR models were generated for the rabbit (Gutting et al., 2013) and
the human (Toth et al., 2013) using a mixture of human and nonhuman primate sourced data.
Table 5-18 summarizes the types of mathematical models that have been reported for the rabbit,
nonhuman primate, or human by type of mechanistic model and presence of threshold.
Table 5-18. Examples of Mathematical Dose-Response Models for Inhalation Anthrax in
the Rabbit, Nonhuman Primate, or Human by Type of Model
Type of Model
Dose-Response Model
Does Model Exhibit
Threshold?
Reported Dose-Response
Relationship Using Model
Empirical
Probit or Log Probit
No, unless a background or
threshold parameter is
included.
Druett et al. (1953)
Glassman (1966)
Taft and Hines (2012)
Logistic or Log Logistic
Taft and Hines (2012)
U.S. Enviromnental Protection
Agency (2012b)
Weibull
Dichotomous Hill
Yes
Gamma
Yes
Survival Models
Varies
U.S. Enviromnental Protection
Agency (2014d)
Nominally
Mechanistic
Exponential
No
Haas (2002)
Bartrand et al. (2008)
Taft and Hines (2012)
Beta Poisson
No
Bartrand et al. (2008)
Taft and Hines (2012)
Limited
Mechanistic
Competing Risk Model
Yes
Gutting et al. (2008) and Gutting et
al. (2013), incorporating base
competing risk model of
Brookmeyer et al. (2003);
Brookmeyer et al. (2005)
Cumulative Dose Model
No
Pujol et al. (2009)
In-vivo Delivered Dose
Model
Depends on the model from
which the dose variable is
being revised to represent
delivered dose
Weir and Haas (2011)
Novel EISD Model
(Expansion of
Competing Risk Model)
No
Toth et al. (2013)
Time-Dependent Dose-
Response Model with
Survival Analysis Model
No
Mayer et al. (2011)
Comprehensive
Mechanistic
None to Date
N/A
None to Date
EISD - Exposure - Infection - Symptomatic Illness - Death
N/A not applicable
93
-------
Mechanistic models have been reported to exhibit greater validity than empirical models (Haas et
al., 1999a). However, mechanistic models only consistently outperform empirical models to the
extent that the mathematical and statistical assumptions correctly and sufficiently capture the
modeled biological setting (Portier and Lyles, 1996). Not all mechanistic models are sufficiently
rigorous relative to the actual biology that they can be reliably assumed to outperform empirical
models, especially when there are insufficient data to support the development of a mechanistic
model (Taft and Hines, 2012). There are also the twin concerns of scarce and uncertain
mechanistic data. The lack of specific mechanistic data (i.e., quantitative impacts of dose-
dependency, time dynamics of response) has been identified as a rationale for the selection of
simpler models, including the exponential model (Toth et al., 2013). There is also the potential
tradeoff from increasing the complexity of mechanistic models where any potential advantages
of introducing more biological or mechanistic realism then has the potential "to be lost in a sea
of statistical uncertainty" of the model outputs (Crump et al., 2010).
With regard to selection of models specifically to support risk-based decision-making, the term
"mechanistic-enough" models has been coined to acknowledge that there is utility in models
other than comprehensive mechanistic models for use in risk-based decision-making for
microbial pathogens (U.S. Environmental Protection Agency, 2010b). Models only need to be
sufficiently mechanistic to allow for confidence in the decisions made using its outputs (U.S.
Environmental Protection Agency, 2010b). The lack of necessary mechanistic data for
comprehensive mechanistic dose-response models for B. anthracis and a preference for
parsimony in model selection will continue to favor the types of models currently in use:
empirical models, nominally mechanistic models, and possibly limited mechanistic models.
94
-------
5.5.3 Mathematically Modeling the Microbial Dose-Response Relationship
Benchmark dose (BMD) analysis empirically fits models to dose-response data and identifies the
dose associated with a specific response level for an identified endpoint
(U.S. Environmental Protection Agency, 2012a). The dose-response
models selected for evaluation must be appropriate based on the type of
data. In the case of B. anthracis, the lethality endpoint will be used for
dose-response analysis and is typically reported as a percentage or fraction
of the individuals that die at a given dose. This allows for the use of
dichotomous dose-response models (e.g., exponential, probit).
BMD analysis is distinguished from other approaches for empirical curve
fitting due to its clear terminology to describe the overall process and
associated reporting of results. The POD is then generated based on an
identified BMD that has associated with it a specified response level and
has an identified lower limit of the BMD value at the specified response
level. The discussion of mathematical modeling will focus on the BMD
approach for empirical modeling of dose-response relationships because it
adds necessary structure to the reporting of empirical modeling results.
Benchmark dose analysis estimates the dose, termed a BMD, for an identified response level, the
benchmark dose-response (BMR) (U.S. Environmental Protection Agency, 2012a). The BMD is
the model's best estimate of the dose that produces a response at the level of the BMR. A BMR
of 10% would be equivalent to a 10% increase in the response rate of the endpoint of interest
(i.e., extra risk) (U.S. Environmental Protection Agency, 2012a). Ideally, the response level of
interest is within or near the lowest end of the observable range of the dose-response data set
95
Summary of Findings for
Mathematically Modeling
the Microbial Dose-
Response Relationship
• Benchmark dose
analysis empirically fits
models to dose-
response data.
• A science policy gap
for the use of
benchmark dose
analysis is guidance on
the selection of the
BMR and POD for a
given data set.
• Selection of the BMR
for B. anthracis is
challenged by the
reliance on lethality
endpoints in most data
sets.
-------
(U.S. Environmental Protection Agency, 2012a). The POD is then determined using the
identified BMR value. The POD is the dose-response point from where the low-dose
extrapolation can be performed when necessary.
When modeling dichotomous data for chemical hazards, BMR values of 0.50, 0.10, and 0.01 are
identified as standardized reporting values. When using a lethality endpoint, these values
correspond to BMD estimates of 50% lethality (i.e., LD50), 10% lethality, and 1% lethality, with
the resulting BMDs written as BMD50, BMD10, and BMD01, respectively. An identified BMR
value, or a range of BMR values, specific for microbial data to support risk-informed decision-
making from BMD outputs or for standardized reporting is not available. The lack of BMR
guidance limits the ability to define a statistically-based POD from the fitted dose-response
model.
However, the determination of the appropriate BMR values may require a unique evaluation
relative to the values for chemical agents due to the reliance on lethality endpoints in B.
anthracis dose-response data sets, high lethality levels associated with exposure levels of
concern, and limited statistical power of most dose-response data sets. The selection of the BMR
value is data-dependent, but also incorporates science policy determinations when setting the
value. The identification of the BMR range of values or guidance for their selection is a science
policy gap for microbial dose-response analysis.
When empirically fitting models, there are many methods to fit the models to a data set (e.g.,
methods of maximum likelihood, nonlinear least squares, and generalized estimating equations
[GEE]) (U.S. Environmental Protection Agency, 2012a). U.S. Environmental Protection Agency
(2012a) should be consulted for more details on how best to evaluate the method of fitting the
96
-------
model. Empirical curve-fitting is appropriate for microbial dose-response analysis of empirical
models, nominally mechanistic models, and some limited mechanistic models depending on the
form of the model.
To capture the statistical variability associated with the calculated BMD value, the benchmark
dose limit (BMDL) value identified. The BMDL is the 95% lower statistical confidence limit of
the BMD when the 95% lower confidence limit is applied to the estimated slope parameter value
(U.S. Environmental Protection Agency, 2012a). The BMDL is the lowest dose that is
supportable from the modeling when the BMR is within or near the lower end of the observable
range of dose-response data. The modeled BMDL values are then evaluated to select the POD(s)
as a starting dose value for an interspecies or low-dose extrapolation (U.S. Environmental
Protection Agency, 2012a).
Appendix F identifies the process to perform BMD, available software, and potential
considerations when modeling dose-response relationships of microbial pathogens.
5.6 Conduct Interspecies Extrapolation
The purpose of the interspecies extrapolation process is to account for potential differences in
kinetics and dynamics between the human and the animal models from which the dose-response
data were obtained. Specifically, the POD is converted to a HED via this process. Table 5-19
identifies the key questions that must be assessed as part of the interspecies extrapolation
process.
97
-------
Table 5-19. Conduct Interspecies Extrapolation
Steps in Microbial
Dose-Response
Analysis
Key Questions
Report Section
Conduct Interspecies
Extrapolation to a HED
(Section 5.6)
What is a general framework that
can be used for interspecies
extrapolation of B. anthracis?
Section 5.6.3. Proposed Framework for Interspecies
Extrapolation for B. anthracis
What data for the rabbit,
nonhuman primate, and human are
available to evaluate the kinetics
and dynamics of B. anthracis
pathogenesis?
Section 5.6.4 Available Kinetic Data
Section 5.6.5 Available Dynamic Data
How can available data be
incorporated in the extrapolation
process?
Section 5.6.6 Summary of Extrapolation Framework
for B. anthracis
HED — human equivalent dose
98
-------
5.6.1 Review of Interspecies Extrapolation Approaches for Chemical Agents
The interspecies extrapolation process estimates a HED by accounting for differences in
response between the animal model and the human to the same level of external exposure. The
HED is derived to have the same "magnitude of effect" as the POD of the animal model (U.S.
Environmental Protection Agency, 201 lc). Comprehensive guidance for interspecies
extrapolation of chemical dose-response data is available and is routinely applied in the
generation of toxicity values for chemicals with sufficient data.
However, the development of microbial dose-response approaches to
address interspecies extrapolation lags significantly behind that of the
chemical agents.
The interspecies extrapolation process for microbial dose-response
analysis lacks a framework, defined terminology, and published
approaches to comprehensively describe an interspecies extrapolation
process. The lack of accepted interspecies extrapolation approaches
has been widely identified as a knowledge gap to be addressed (U.S.
Environmental Protection Agency, 1994a.; International Life Sciences
Institute (ILSI), 2000; U.S. Environmental Protection Agency, 2014c).
Given the progress made for interspecies extrapolation of chemical
dose-response analysis, these frameworks should be evaluated for
applicability to microbial agents.
The interspecies extrapolation process for chemical agents identifies
two factors that contribute to variability in response between the
Summary of Findings for
Conduct Interspecies
Extrapolation
• Interspecies extrapolation
process for microbial
dose-response analysis
lacks a framework,
defined terminology, and
published frameworks.
• The interspecies
extrapolation process for
chemical agents is an
appropriate starting
framework for
interspecies extrapolation
process of biological
agents.
• There are sufficient data
and available approaches
to conduct the dosimetric
adjustment element of the
interspecies extrapolation
process for inhaled and
deposited dose metrics.
• Knowledge gaps that
currently limit the
quantitative assessment of
dynamic differences
between the animal model
and the human.
99
-------
animal model and the human: kinetics and dynamics. Kinetics considers the dosimetry associated
with the movement and transformation of the administered dose to an internal dose, whereas
dynamics evaluates how differences in concentration at the identified target tissue may be
associated with the same level of response in both the test animal and human (U.S.
Environmental Protection Agency, 2014b). The common element of kinetics and dynamics is the
focus on the internal dose: the factors determining the internal dose from an administered dose
that dominate kinetics or the factors that define the response from a given internal dose level that
are describing dynamics (U.S. Environmental Protection Agency, 2014b).
Kinetics is the "determination and quantification of the time course and dose-dependency of
adsorption, distribution, metabolism, and excretion (ADME) of chemicals" (U.S. Environmental
Protection Agency, 2014b). One adjustment for the kinetics of inhalation exposure is a
categorical dosimetric adjustment factor (DAF) that explicitly considers differences (i.e.,
anatomical, physiological) between species, physical differences between particles and gases,
and whether the toxicity is anticipated to be limited to the portal-of-entry or will have a systemic
presentation (U.S. Environmental Protection Agency, 2014b). The process for development of
reference concentration values detailed the derivation and application of DAF values (U.S.
Environmental Protection Agency, 1994a).
Dynamics is the "determination and quantification of the sequence of cellular and molecular
events leading to a toxic response" (U.S. Environmental Protection Agency, 2014b). Dynamics
evaluates the interaction of the "biologically active chemical" with the target site and subsequent
events that are associated with toxicity (U.S. Environmental Protection Agency, 2014b). The
measure of the "biologically active chemical" at the target site is termed the internal dose. The
internal dose should be the measurement at a specified tissue location that is most closely
100
-------
associated with the response endpoint of (Jarabek et al., 2005). The evaluation of dynamics
requires some level of mechanistic knowledge, including key events and mode of action leading
to the toxicological endpoint of interest (U.S. Environmental Protection Agency, 2014b).
U.S. Environmental Protection Agency (2014b) identifies a hierarchy of extrapolation techniques
to model kinetics and dynamics. The hierarchy ranges from data-intensive modeling to default
values consisting of PBPK modeling, data-derived extrapolation factors (DDEF), and default
factors (U.S. Environmental Protection Agency, 2014b). The recommended approach for
extrapolation is based upon the availability of data and supporting models (U.S. Environmental
Protection Agency, 2014b).
The least data-intensive approach is the use of default values, such as Uncertainty Factors (UF)
(e.g., 10-fold UF values used in toxicity values for chemical hazards). The UF values are used
when there are very limited or no chemical-specific data (U.S. Environmental Protection
Agency, 2014b). The following UF values are identified: interspecies UF, intraspecies UF,
lowest observable adverse effect level (LOAEL) to lowest observable adverse effect level
(NOAEL) UF, and Database UF, and Subchronic to Chronic UF, with the recommendation that
the total UF should not exceed 3,000 (U.S. Environmental Protection Agency, 2002). The UF
values account for both uncertainty and variability (U.S. Environmental Protection Agency,
2014b). The maximum UF value is 10 (i.e., one order of magnitude), with a half-power value
(10°-5) of approximately 3. A UF factor of up to 10 is assigned to interspecies differences, with V2
of 10 (i.e., 10°-5) applied for interspecies kinetic differences and V2 of 10 (i.e., 100 5) assigned for
dynamics differences (U.S. Environmental Protection Agency, 2014b). The UF values were
defined specifically for chemical agents, with evolution in their interpretation over time and data
generated showing the values could be supported through evaluation of chemical-specific animal
101
-------
and human data (Renwick, 1993). Renwick (1993) evaluated chemical-specific ADME data for
the animal and human relative to the UF value of 10 and found the value to be generally
appropriate for that element of an interspecies extrapolation. The use of the UF value of 10
combined for kinetics and dynamics for interspecies extrapolation has not been assessed for
microbial pathogens.
The most data-intensive approach for extrapolation involves the use of PBPK modeling. U.S.
Environmental Protection Agency (2014b) identifies this as the preferred approach if sufficient
chemical-specific mechanistic data and models are available. The PBPK model is a type of
compartment model that incorporates consideration of both tissue volume and blood flow
information. Models are individually developed for the animal model and the human to predict
internal doses and responses (U.S. Environmental Protection Agency, 2006).
The remaining method of extrapolation in the hierarchy is the use of DDEF values. A DDEF
approach is based on two fundamental assumptions: (1) the endpoint of interest results from the
interplay of kinetic and dynamic elements, and (2) relevant kinetic and dynamic elements can be
quantified in animals and humans (U.S. Environmental Protection Agency, 2014b). In contrast to
UFs, DDEF values address variability only (U.S. Environmental Protection Agency, 2014b).
They may reduce uncertainty through the incorporation of chemical-specific data, but they do not
explicitly include an uncertainty component (U.S. Environmental Protection Agency, 2014b).
U.S. Environmental Protection Agency (2014b) identifies three forms of data necessary to derive
the DDEF value: (1) mode of action, including key events through endpoint of interest and
identification of "toxicologically active chemical species," (2) target tissue, and (3) an
appropriate dose metric for measurement of exposure (U.S. Environmental Protection Agency,
2014b). For chemicals with some kinetic and dynamic data, the DDEF values provide a data-
102
-------
driven middle ground between comprehensive PBPK models and default approaches for
extrapolation of chemical data.
5.6.2 Published Approaches for Interspecies Extrapolation of B. anthracis
A partial interspecies extrapolation for B. anthracis was conducted using a "dosimetric
adjustment" to evaluate species differences in inhalation and deposition for the nonhuman
primate and the human (U.S. Environmental Protection Agency, 2010a). The "microbial
equivalent of dynamics" was identified as a component of an interspecies extrapolation, but was
noted to be beyond the scope of that particular assessment (U.S. Environmental Protection
Agency, 2010a). The same dosimetric adjustment approach was later applied in the rabbit animal
model using an average daily dose metric for the multiple-dose B. anthracis data set (U.S.
Environmental Protection Agency, 2012b).
Stochastic mass balance modeling of inhalation and particle deposition rates was used to
evaluate species differences between identified animal models (i.e., guinea pig, nonhuman
primate) and the human for B. anthracis inhalation exposure as described in Weir and Haas
(2011). An alternative approach for interspecies extrapolation of a different microbial pathogen,
Legionella spp. was the preferential selection of animal models to maximize similarity for a
subset of host immune responses in the human (Armstrong and Haas, 2007). Modeling results
were then compared with human epidemiological data to evaluate model outputs suitability for
the human (Armstrong and Haas, 2007). However, the extremely low incidence of human
inhalation anthrax and lack of epidemiological data would preclude use of this approach for B.
anthracis.
5.6.3 Proposed Framework for Interspecies Extrapolation for B. anthracis
103
-------
The interspecies extrapolation process for chemical agents is an appropriate starting framework
to begin development of an interspecies extrapolation process for inhalation exposure to
B. anthracis spores. This general framework is consistent with and will build upon the approach
initially described in U.S. Environmental Protection Agency (2010a) and U.S. Environmental
Protection Agency (2012b). These approaches incorporated EPA exposure assessment practices
and some terminology from the interspecies extrapolation framework for chemical agents (e.g.,
dosimetric adjustment).
To address the dosimetric elements of kinetics, U.S. Environmental Protection Agency (2010a)
and U.S. Environmental Protection Agency (2012b) evaluated inhalation rate and deposition rate
to derive an internal dose for the animal model. This is equivalent to the use of the DAF
described in U.S. Environmental Protection Agency (1994a) in guidance for the development of
the inhalation reference concentration. Due to a lack of dynamic data, it was assumed that an
equivalent internal dose was associated with the level of response in the animal model and the
human. While this assumption was made to simplify the previous assessment, the potential to
assess dynamics for B. anthracis requires further evaluation. For example, significant differences
in species sensitivity were reported across a variety of animal models to intravenous challenge
with anthrax toxin Lincoln et al. (1967). Additionally, population variation in cellular sensitivity
to anthrax toxin was reported from in vitro studies of human cells (Martchenko et al., 2012). The
key challenge will be sufficient mechanistic knowledge to quantitatively link these various
measures to both dose and endpoints of interest.
For inhalation of B. anthracis spores, the internal dose evaluation, at a minimum, should
consider both an inhaled dose and deposited dose(s) to the region(s) associated with initiation of
infection. Though it was not explicitly stated, U.S. Environmental Protection Agency (2010a)
104
-------
and U.S. Environmental Protection Agency (2012b) assumed that the B. anthracis spore is the
biologically active form of the pathogen. It can be argued that the vegetative bacterium should be
considered the biologically active form as the spore is not pathologically active until it
germinates. However, the initial host-pathogen interaction takes place between the spore and the
host tissue (e.g., phagocyte, epithelial cell, lymphoid tissue). It is this first contact that is the
opportunity point for the spore to germinate or to lose viability based on the action of the host
immune system (e.g., phagocytosis by macrophage).
For this proposed framework, an initial point of delineation between kinetic and dynamic
processes is the interface of the spore and the environment associated with initiation of infection.
Assumptions must be made regarding the host tissue most closely associated with initiation of
infection to select appropriate dose metrics for dose-response relationship development and the
interspecies extrapolation. It was identified in Section 0 that multiple-dose metrics should be
evaluated for the development of dose-response relationships. If an internal dose was not used as
part of the dose-response modeling, it is reasonable to assess multiple internal doses as part of
the interspecies extrapolation process to see if there is a substantial difference in outputs.
For microbial dose-response analysis of B. anthracis spores, the kinetics process can be
described in two parts. The first part represents host contact with the administered dose (e.g., air
concentration) or delivered dose (e.g., inhaled dose) through the spore transport to the target
internal tissue where germination may first take place. However, U.S. Environmental Protection
Agency (2012b) notes the challenge in a clear delineation between kinetics and dynamics
because of the interplay between the two processes. Given this reasoning, a second conceivable
kinetics element for interspecies differences might be the proliferation rate of vegetative bacteria
in the blood based on recently reported species differences among the human, nonhuman
105
-------
primate, and rabbit for spore germination and vegetative proliferation rates (Bensman et al.,
2012).
To address the dynamic elements for the interspecies extrapolation, the interaction between the
host and the biologically active B. anthracis form must be described through the key events
leading to the endpoint of the assessment (U.S. Environmental Protection Agency, 2014b). There
must be sufficient data on the key events and quantitative mechanistic information to link them
with internal dose and the endpoint (U.S. Environmental Protection Agency, 2014b). The
dynamic element of the extrapolation may be appropriately modeled with BBDR or other
dynamic models (U.S. Environmental Protection Agency, 2014b). This is the area of greatest
challenge for the development of a microbial interspecies extrapolation process. For
B. anthracis, there is currently an insufficient mechanistic understanding of the key events from
initiation of infection through bacteremia and toxemia to conduct a full dynamic evaluation.
However, it may be possible to begin to evaluate initial host-pathogen interactions to develop a
better understanding of dynamics associated with initiation of infection as a starting point for
species differences.
There are elements of the hierarchy used with chemical dose-response analysis that are difficult
to implement with microbial dose-response data. It would entail considerable effort to develop
comprehensive default values (e.g., UF values), which may not be appropriate across the diverse
group of microbial pathogens of interest. There would be considerable effort associated with the
development of UF values for microbial pathogens as there is not an equivalent set of data for
microbial pathogens relative to chemical agents to support selection of UF values. There is a lack
of general data describing variability in response for microbial pathogens as a group, and B.
anthracis specifically, to support development of interspecies and intraspecies UF values. The
106
-------
concept of uncertainty has not been considered outside of general qualitative statements. It is
unknown if a pathogen-wide default value is biologically appropriate given potential differences
among pathogens. The initial dose-response modeling approach to use BMD models in lieu of
identification of NOAEL and LOAEL values negates the use of that UF value. The adjustment
for subchronic studies that is applied for chronic values also does not have available the same
body of data that was used for chemical dose-response (i.e., the vast majority of B. anthracis
challenge studies are single-dose).
5.6.4 A vailable Kinetic Data
The two categories of kinetics data relevant for inhalation of B. anthracis spores are inhalation
rate and deposition rate. Most currently performed animal challenge studies with B. anthracis
use plethysmographic data to determine the inhalation rate (e.g., volume/time) during the
challenge study. However, care should be taken if allometric equations are used to derive the
animal model inhalation rate if plethysmographic data are not available. When allometric
equations are used to estimate minute volume, they do not consider the physiological state of the
test animal (e.g., stress, tranquilizers) and may not accurately reflect the actual inhalation rate
during the challenge (Taft and Hines, 2012). However, human inhalation data for a variety of
activity levels are readily available from the Exposure Factors Handbook (U.S. Environmental
Protection Agency, 201 lb) and will not be further considered here.
5.6.4.1 Experimental Sources of Deposition Data
For the rabbit, published particle deposition data and modeled data are available describing
whole or lung region-specific values (Raabe et al., 1988; Gutting et al., 2012; Gutting et al.,
2013) (Table 5-20). However, the reliability and precision of the measurement techniques raise
potential issues for their application in modeling. Potential biases in measurement approaches are
107
-------
described in Table 5-20. For example, historical data derived from inhalation of radiological
aerosols is compromised by the lack of real-time inhalation data to estimate dose (Raabe et al.,
1988) and bronchoalveolar lavage may undercount deposited particles due to potential
translocation within the lungs (Gutting et al., 2012).
Table 5-20. Summary Table of B. anthracis Deposition Data for the Rabbit
Study
Reported Value
Measurement
Potential Bias
Gutting et al.
(2013)
Pooled value of
4.63% from two
data set values:
4.33% (+2.2%)
and 4.93 %
(+0.8%),
represents whole
lung deposition
Homogenization of New Zealand
white rabbit lung tissue and
extrapolation to the whole lung after
inhalation exposure to B. anthracis
spores, particle size MMADa of 1.0
jim + 0.3 jim.
Potential for underestimation of
deposition if epithelial cell
internalization of deposited particles is
rapid, see Jenkins and Xu (2013) data
for mouse animal model.
Gutting et al.
(2012)
3.07% + 0.9% and
1.33%+ 0.2%,
represents whole
lung deposition
Bronchoalveolar lavage to wash out
deposited B. anthracis spores in New
Zealand white rabbit, particle size
MMAD of 1.0 jim + 0.3 jim.
Deposited doses reported from
bronchoalveolar lavage may be biased
low if inability to wash out all deposited
spores or rapid transport across
epithelial cell lining takes place (Gutting
et al., 2012).
Raabe et al.
(1988)a
Ranging from 6.6
±0.6% at 0.97
junto 1.1 ±0.2 %
at 4.86 jun,a
pulmonary
deposition only
Measurement of deposition to
pulmonary region of the rabbit after
inhalation of monodisperse lo9Yb
aluminosilicate aerosol with
aerodynamic resistance diameters of
particles ranging from 0.18 to 8.65
jim.
Use of Guyton's formula to estimate
minute volume for calculation of
deposition would bias results if actual
animal inhalation rate differed (Raabe et
al., 1988).
MMAD - Mass Median Aerodynamic Diameter
a Raabe et al. (1988) data were the basis for U.S. EPA's RDDR model as described in U.S. Environmental Protection Agency
(1994b)
5.6.4.2 RDDR Modeling
The EPA's Regionally Deposited Dose Ratio (RDDR) model (U.S. Environmental Protection
Agency, 1994b) provides estimates for the fractional regional depositional efficiency in the lung
for inhalation of particulates for laboratory animal species and the human (U.S. Environmental
Protection Agency, 1994a). The model output, the RDDR, is the "ratio of the deposited dose in a
respiratory tract region (r) for the laboratory animal species of interest (RDDa) to that of humans
(RDDh)" (U.S. Environmental Protection Agency, 1994a). This ratio can be used as a DAF for
108
-------
the kinetics portion of an interspecies extrapolation (U.S. Environmental Protection Agency,
1994a). At a minimum, the inputs include the particle air concentration, the Mass Median
Aerodynamic Diameter (MMAD) and geometric standard deviation (GSD) value for the particle
distribution, and the animal model body weight from the challenge study. One caveat to the use
of the RDDR model is that the animal deposition modeling incorporates data from Raabe et al.
(1988), which relied on an allometric equation to determine the inhalation rate necessary to
determine deposition.
However, the RDDR model can be used with supplied inhalation rate data (e.g.,
plethysmographic data) to generate regional surface area (SAr) and regional fraction deposition
(Fr) values specific to the lung region of interest. If this change is not made, allometric body
weight equations will be used in the model to generate the minute volume (Ve). For distributions
of particle sizes with a known MMAD and GSD, one advantage of the RDDR software is that
the software scales the Fr value to the specified Ve (U.S. Environmental Protection Agency,
1994a).
The equations of the RDDR model can be used with study-specific data to generate a type of
dosimetric adjustment factor, the RDDR value. Figure 5-3 shows the calculation of the RDDR
DAF that can be used to account for interspecies differences in inhalation and deposition for
inhaled particles. As shown in Figure 5-3, the DAF can be multiplied by the POD from the
animal study to derive a HED that accounts for interspecies differences in inhalation and
deposition.
109
-------
RDDR - * a,,in,al
(^rxfv)
\3rtr 'h
Dosimetric Adjusted Dose for Human: POD x RDDR
Where:
POD = Point of Departure from animal study data
RDDR = Regional Deposited Dose Ratio of the animal to the human
(Also termed a dosimetric adjustment factor [DAF])
Ve = Minute Volume (ml,)
SAr = Surface Area of Region,, (cm2)
Fr = Fractional Deposition of Region,. (Unitless)
Figure 5-3. Calculation of an RDDR-based dosimetric adjustment factor.
5.6.4.3 CFD Modeling
Due to advances in computational modeling, particulate deposition models for the lung have
become highly developed (Kleinstreuer et al., 2008) and provide the ability to track patterns of
deposition through the pulmonary system as a function of the morphology, breathing parameters,
and particle characteristics. Building on these advances, Kabilan et al. (2015) developed the first
particle deposition model using physiologically realistic, image-based 3D airway geometries of
the human and rabbit with computational fluid dynamics (CFD) airflow modeling coupled with
Lagrangian particle tracking methods. The CFD model was developed using particle size
distributions, concentrations, and rabbit plethysmography data from the EPA single-dose
challenge study for the rabbit (U.S. Environmental Protection Agency, 2011a). The CFD model
predicts the inhalation and deposition of B. anthracis spores during transient breathing. Table
110
-------
5-21, originally from Kabilan et al. (2015), reports the modeled deposition efficiencies for the
respiratory tract regions in the rabbit and the human. The modeled deposition values for the deep
lung are considerably higher for both the rabbit and the human than previous deposition
measurements or modeled results; further corroboration may be appropriate prior to use.
Table 5-21. Deposition Efficiencies for Different Annotated Regions in the Rabbit and the
Human
Modeling Case
MMAD
(lira)
Concentration
(Spores/m3)
Location
% Deposition Based
on Exposure
Rabbit
Human
Nose
12.61
3.21
Pharynx
0.03
0.12
Case 1
1.12
3.97E+11
Larynx
0.13
0.33
Trachea
0.07
0.01
Bronchi & Bronchioles
1.44
5.70
Deep Lung
54.34
62.08
Nose
7.05
-
Pharynx
0.01
-
Case 2
0.92
1.18E+08
Larynx
0.16
-
Trachea
0.06
-
Bronchi & Bronchioles
1.49
-
Deep Lung
58.94
-
*The total particle deposition for the rabbit and the human was 68.62% and 71.45%, respectively for Case 1.
5.6.5 Available Dynamic Data
Though the pathophysiology of anthrax in the human has been deemed "well characterized" for
over one hundred years (Ioannidis, 2012), these data were not generated to mechanistically
describe the origin and magnitude of potential response differences between the test animal and
the human. Currently, there are insufficient mechanistic knowledge and associated modeling
approaches to assess dynamic contributions to potential interspecies differences. The evaluation
of dynamics requires mechanistic knowledge of key events associated with the host-pathogen
interactions at a quantitative level and in association with internal dose. B. cmthrcicis is not
unique in lacking these data, as sufficiently detailed mechanistic knowledge for an interspecies
111
-------
extrapolation is likely lacking for most if not all microbial pathogens for which microbial risk
assessment is conducted. Given these data challenges, it is not recommended that a generic
default value be developed for use.
As a first step in developing a framework for dynamics of B. anthracis response, a conceptual
mapping of contributors to potential species differences in response should be conducted.
Though their approach employed a qualitative evaluation for Legionella spp., Armstrong and
Haas (2007) described a systematic approach to compare early immune system response between
an animal model and the human. The initiation of infection of Legionella spp. is associated with
inhalation and uptake by the alveolar macrophage. Armstrong and Haas (2007) identified
mechanisms that could be associated with species differences (e.g., macrophage uptake and
replication, macrophage "bactericidal mechanism responses") and compared responses for the
human and guinea pig. However, Armstrong and Haas (2007) used the assessment qualitatively
to determine sufficient similarity between the animal model and the human. This approach could
be easily applied to B. anthracis to map potential host-pathogen interactions with the goal of
identifying potential contributors to species differences in response and gathering of potentially
relevant data. The ultimate goal would be development of a quantitative assessment factor.
5.6.6 Summary of Extrapolation Framework for B. anthracis
An interspecies extrapolation framework that considers both kinetic and dynamic elements as
potential contributors to species differences is a viable approach for microbial pathogens,
including B. anthracis. For the kinetics element of the process, the dosimetric adjustment process
for assessment previously described in U.S. Environmental Protection Agency (2010a) and U.S.
Environmental Protection Agency (2012b) provides a good starting foundation. The availability
of new CFD data (Kabilan et al., 2015) modeled with the U.S. Environmental Protection Agency
112
-------
(2010a) adds the knowledge base for deposition of spore particles in the rabbit. The dosimetric
adjustment factor equation (Figure 5-3) as used in the RDDR model provides a mathematical
approach that can use currently available data for general species-specific elements (e.g., particle
deposition) and study-specific data (e.g., animal-specific inhalation rate during the challenge) to
conduct the kinetics portion of the interspecies extrapolation.
However, there are knowledge gaps that currently limit the quantitative assessment of dynamic
differences between the animal model and the human. One starting recommendation is to map
host-pathogen interactions associated with initiation of infection for B. anthracis with the goal of
identifying potential contributors to species differences in response. Available data can then be
evaluated relative to the sufficiency to quantitatively evaluate species differences in the context
of key events and endpoints of interest.
113
-------
6 Conclusion
The primary purpose of this report is to provide open source data and analysis approaches that
can be used to develop a site-specific HHRA for B. anthracis. The report presents the results of
an agent-specific planning activity for B. anthracis that evaluated published dose-response data,
identified data and process gaps for microbial dose-response analysis of the agent, and identified
science policy decisions that may be necessary to conduct a HHRA for this agent. The results of
the report are summarized by answering the science questions posed in Section 2 as shown in
Figure 6-1.
• What natural history data for B. anthracis are available to inform development of a site-
specific CSMfor the identified exposure scenario?
Source materials associated with potential exposure to B. anthracis spores include contaminated
animal products, cross-contamination of materials by contaminated animal products, or
manufactured spore products that are intentionally or unintentionally released. With the
exception of the deliberate release of manufactured spores, anthrax illness is relatively rare in
developed countries and most often results from contact with infected animals or contaminated
animal products (Passalacqua and Bergman, 2006). Published reports of anthrax infection
support the potential for the released B. anthracis spores to result in inhalation, ingestion, and
dermal exposure with potential disease transmission associated with these routes of exposure.
Inhalation anthrax is associated with severe life-threatening illness and a quantitative HHRA
could be developed with existing data. However, there is the potential for high levels of
uncertainty associated with the quantitative HHRA outputs from limitations in dose-response
data. The ingestion and dermal pathways are also likely to be complete but there are insufficient
114
-------
data to conduct a quantitative HHRA. As a result, a qualitative assessment is recommended for
these exposure pathways.
The available natural history data are sufficient to generate a site-specific CSM with regard to
identification of potential sources of B. anthracis exposure, fate and transport mechanisms,
potential exposure pathways, the likelihood of completed exposure pathways, and the ability to
perform a quantitative or qualitative assessment.
115
-------
Effects Assessment
• Hazard Identification
• Dose-Response
Risk Characterization
Problem Formulation
• Conceptual Model
• Analysis Plan
Planning & Scoping
Exposure Assessment
Risk Assessment
What natural history data are available to
inform development of a site-specific
Conceptual Site Model (CSM) for the
identified exposure scenario?
What data support the use of the rabbit and nonhuman primate animal models
for development of dose-response modeling of B. antbracis?
What data are available to support the development of the hazard
identification, including disease pathogenesis data?
What dose-response data are available for inhalation and oral exposure in the
rabbit, nonhuman primate, and human that may be appropriate for
development of a microbial dose-response relationship?
What are available approaches to model a microbial dose-response
relationship?
How might an animal-to-human extrapolation be conducted with dose-
response data and what data are available?
Figure 6-1. Science questions and associated elements of the U.S. Environmental Protection Agency (2014a) human health risk
assessment framework.
116
-------
• What data are available to support the development of the hazard identification, including
disease pathogenesis data?
The hazard posed by exposure to B. anthracis spores is documented by published reports,
including the transmission of inhalation anthrax from contaminated animal products or the
intentional or accidental release of spores. Though each type of anthrax illness can progress to a
fulminant infection, inhalation anthrax poses the greatest threat of lethality because it is difficult
to diagnose during early stages of illness and becomes rapidly lethal after development of severe
symptoms (Inglesby et al., 2002). Even with modern medical treatment and early diagnosis, the
case fatality rate of those with inhalation anthrax during the 2001 anthrax letter event was 45%
(Inglesby et al., 2002). However, the fatality rate is generally estimated to be almost twice as
high without antibiotics or intensive medical treatment (Inglesby et al., 2002; Hilmas et al.,
2009).
The disease pathogenesis process for inhalation anthrax is well described relative to key events.
However, there is still considerable uncertainty in the mechanistic details of the disease process.
There is not a clear link between mechanistic pathway(s) or tissue dose(s) associated with the
lethality endpoint. There is also uncertainty regarding the mechanistic process for the initiation
of the infection. There are two models that currently describe the initiation of infection using
slightly different assumptions regarding the role of identified tissues and B. anthracis toxin in the
initial stages of infection: the Trojan horse model of Guidi-Rontani (2002) and the jailbreak
model of Weiner and Glomski (2012). Knowledge of the pathway(s) by which infection is
initiated is critical for many aspects of the dose-response modeling process.
117
-------
There are sufficient natural history data to generate the hazard identification element of an
HHRA. However, there are significant data gaps associated with disease pathogenesis
knowledge. This uncertainty has ramifications for multiple areas in the HHRA including the
selection of dose metric(s) for generation of dose-response relationships and the interspecies
extrapolation process.
• What data support the use of the rabbit and nonhuman primate animal models for
development of dose-response modeling ofB. anthracis?
Animal model suitability for development of a B. anthracis dose-response relationship was
determined by an assessment of general concordance in anthrax pathology between the human
and the rabbit and nonhuman primate animal models. Twenhafel (2010) evaluated human
pathology data from Sverdlovsk (Abramova et al., 1993; Grinberg et al., 2001) and the 2001
anthrax letter event (Jernigan et al., 2001) to generate the list of key human pathological
findings. The Twenhafel (2010) list was used to assess anthrax pathology of the rabbit and
nonhuman primate relative to that of the human.
The rabbit and nonhuman primate exhibit many commonalities in the type of lesions and tissues
associated with inhalation anthrax pathology in the human. The principal anthrax lesions of
edema, hemorrhage, and necrosis are present in a variety of common tissues in the rabbit,
nonhuman primate, and human. However, this constellation of pathology is generally consistent
with descriptions of animal models susceptible to fulminant inhalation anthrax infection (Gleiser
et al., 1963) and is not unique to the rabbit and nonhuman primate animal models. Lesion
differences among susceptible animals are manifested by differing levels of inflammation and
infiltration of leukocytic elements into existing lesions (U.S. Food and Drug Administration,
118
-------
2002), whereby less susceptible animals exhibit greater inflammation and leukocytic infiltration
than more susceptible animals, which rapidly succumb to illness.
There were no identified differences between the rabbit and the nonhuman primate animal
models for elements of anthrax pathology that do not have a time-dependency for incidence or
severity in presentation. However, there are preliminary indications that time-dependency may
be contributing to the identified differences in pathology.
The results of this pathology assessment support the continued use of the rabbit and nonhuman
primate animal models for development of dose-response data for B. anthracis.
• What dose-response data are available for inhalation and oral exposure in the rabbit,
nonhuman primate, and human that may be appropriate for development of a microbial
dose-response relationship for B. anthracis?
Dose-response data were categorized into three categories: Key Data, Supporting Data, and
Additional Data. Key Studies were defined as representative of the highest quality dose-response
studies that met criteria for selection during the literature search. Supporting Studies had
identifiable limitations in assessment quality indicators relative to Key Studies, yet were found to
have potential in bounding the potential dose-response relationship(s) as described by Key
Studies. Additional Data were defined by the lack of data critical to assessing dose-response
relationships (e.g., original dose and response data set) or study design elements that limit utility
for development of low-dose dose-response relationships.
A literature search was conducted for the inhalation route of exposure for each animal model and
dose-response data were categorized. Few inhalation challenge studies were identified as Key
Studies for the rabbit and nonhuman primate; there were no Key Studies or Supporting Studies
119
-------
identified for the human. The two Key Studies for the rabbit were the single dose U.S.
Environmental Protection Agency (201 la) study and the multiple dose U.S. Environmental
Protection Agency (2012b) study. No studies were categorized as Supporting Studies. For the
nonhuman primate, one single dose Key Study (Lever et al., 2008) and one single dose
Supporting Study (Druett et al., 1953) were identified.
One area of particular concern is the limited number single or multiple dose challenge studies
using low doses. Most animal dose-response data identified through the literature search
originated from single dose studies at very high doses, sometimes as high as 200 times an
identified LD50 value. Single high-dose studies have limited value for the assessment of repeated
low-dose exposure (U.S. Environmental Protection Agency, 2012c). Few studies that reported
dose-response data were designed to derive data for dose-response analysis. Study purposes for
recent data sets included evaluation of the pathology, pathophysiology, or assessment of the
efficacy of medical countermeasures. These studies were often conducted using a single high-
dose challenge to ensure a high likelihood of systemic anthrax infection in the challenge animals.
Historical data were often developed to report an LD50 value for use in military applications or
early anthrax research with little representation of low doses.
Dose-response data are available for the rabbit and nonhuman primate that may be suitable for
development of a human dose-response relationship. However, the uncertainty associated with
the use of these data may be high and is associated with a lack of corroborative data to increase
confidence in their use. Depending on the level of acceptable uncertainty in the analysis outputs,
there may be limitations on how these data may be used in decision-making. There may be value
in conducting additional dose-response challenge studies that are designed with appropriate
120
-------
statistical power for modeling and gather necessary data to inform the animal-to-human
extrapolation process.
• What are available approaches to model a microbial dose-response relationship for B.
anthracis?
Empirical and Mechanistic Models
Empirical and mechanistic models have been used for microbial dose-response modeling of B.
anthracis. To aid in model evaluation, a hierarchy of mechanistic models was proposed to
describe the relative level of biological representation and complexity in the models. The
simplest models are nominally mechanistic models that incorporate simple biological
representations, but biological measurements or modeling cannot inform parameter values. All
parameters are estimated empirically. Limited mechanistic models are the next level of model;
they incorporate mechanistic assumptions and data that can be derived or informed by biological
measurements. The most complex models are comprehensive mechanistic models that
incorporate mechanistic assumptions and data to fully describe biodynamic and biokinetic
elements. The lack of necessary mechanistic data for comprehensive mechanistic dose-response
models for B. anthracis and a preference for parsimony in model selection (i.e., models with as
few parameters as necessary) will lead to the continued use of empirical models and limited or
nominally mechanistic models.
Determination of Dose Metric and Other Modeling Assumptions
A dose metric is the mathematical description of the challenge study dose that is used to model
the dose-response relationship and conduct the interspecies extrapolation. The preferred dose
metric is the internal dose that can be most closely mechanistically or otherwise correlated with
121
-------
the biological endpoint of interest (Jarabek et al., 2005). A dose metric is associated with a
specified exposure duration and can also be expressed as a time-normalized measurement (e.g.,
CFU/day) (U.S. Environmental Protection Agency, 2014b).
The selection of dose metrics for multiple-dose exposure of B. anthracis introduces questions
regarding the time duration to which the dose should be applied. The U.S. Environmental
Protection Agency (2012b) multiple-dose study reported dose-response relationship evaluations
using two dose metrics: accumulated inhaled dose and average daily inhaled dose. An
accumulated dose metric assumes an equivalent hazard whether the intake is in the form of one
dose or in many doses over that same time (Mayer et al., 2011). The independent action
hypothesis may have relevance for the determination of dose metrics for multiple-dose B.
anthracis exposure studies (U.S. Environmental Protection Agency, 2014d). Potential
dependencies by time, dose, or route of exposure may affect consistency with the independent
action hypothesis. The magnitude of exposure or exposure duration (Mayer et al., 2011; U.S.
Environmental Protection Agency, 2014d) where independent doses can be delineated from
dependent doses have not been explicitly evaluated to date.
Though there is uncertainty in the identification of the most appropriate dose metric, this should
not limit the evaluation of dose-response relationships. Relevant dose metrics should be
identified and a justification provided for those that are evaluated. With regard to selection of the
regional deposition location(s) for the deposited dose, multiple-dose metrics can be evaluated.
Given that the differences in deposition may be small relative to other components of the
inhalation dose calculation, the actual difference in the modeled dose-response relationship may
be of limited magnitude. The documentation for the dose-response relationship should include a
transparent identification of the basis for selection of the dose metric(s) considered. There should
122
-------
also be a qualitative discussion of the uncertainty associated with the dose metric selection in the
risk characterization element of the risk assessment.
Benchmark Dose Modeling
Benchmark dose modeling can be used to fit dose-response data to mathematical models.
However, one science policy gap in the use of BMD for microbial pathogens is the lack of
guidance on the selection of a BMR for microbial data. The determination of a BMR should be
based upon the intended use of the BMD outputs, the statistical features of the data set, and
biological basis of the modeled disease process (U.S. Environmental Protection Agency, 2012a).
A BMR value (or range of BMR values) to standardize reporting or to support BMD decision-
making using microbial data is not available. However, the determination of a suggested range of
appropriate BMR values may require a unique evaluation relative to the values used for chemical
agents. This is due to the reliance on lethality endpoints in B. anthracis dose-response data sets,
high lethality levels associated with exposure levels of concern, and limited statistical power of
most dose-response data sets.
• How might an animal-to-human extrapolation be conducted with B. anthracis dose-response
data and what data are available?
The interspecies extrapolation process is designed to account for differences between the animal
model and the human that could affect the human response to environmental exposures.
However, the development of microbial dose-response approaches to address interspecies
extrapolation lags significantly behind that of chemical dose-response analysis. The interspecies
extrapolation process for microbial dose-response analysis lacks a framework, defined
terminology, and published approaches that comprehensively describe an interspecies
123
-------
extrapolation process. Using the interspecies extrapolation process for chemical agents as a
starting framework, an interspecies extrapolation framework that considers both kinetic and
dynamic elements as potential contributors to species differences should be a viable approach for
microbial pathogens, including B. anthracis. The use of a dosimetric adjustment process to
assess the initial elements of kinetics for B. anthracis has been described previously in U.S.
Environmental Protection Agency (2010a) and U.S. Environmental Protection Agency (2012b).
There are sufficient data and available approaches to conduct the dosimetric adjustment element
of the interspecies extrapolation process for inhaled and deposited dose metrics.
However, there are knowledge gaps that currently limit the quantitative assessment of dynamic
differences between the animal model and the human. One starting recommendation is to map
host-pathogen interactions associated with initiation of infection for B. anthracis with the goal of
identifying potential contributors to species differences in response. Available data can then be
evaluated relative to the potential to quantitatively evaluate species differences in the context of
key events and associated endpoints. While there do not appear to be sufficient mechanistic
knowledge and quantitative data to fully evaluate dynamic elements of the extrapolation at
present, the approach should be increasingly attainable over time with continued evaluation and
directed data generation.
Summary
A considerable body of knowledge is now available for the development of a site-specific HHRA
for B. anthracis. There are sufficient data to develop the CSM, generate the hazard identification,
data and methods to generate a dose-response relationship for B. anthracis, and conduct a partial
interspecies extrapolation. While there are sufficient data to generate a quantitative HHRA, data
124
-------
quality and the presence of data gaps may contribute to potentially high levels of uncertainty in
the risk assessment outputs. Depending on the intended use of the risk assessment outputs, these
data may not be acceptable for all types of risk-based decision-making. Microbial risk assessors
who are assisting in the initial planning and scoping element of the HHRA should take care to
communicate these potential data limitations to decision-makers early in the process.
Table 6-1 summarizes the identified data gaps and science policy gaps by risk assessment
element. The most significant data gap relates to the lack of high quality dose-response data,
defined as possessing sufficient quality to be categorized as Key Data. This clearly affects the
rigor of the risk assessment. An additional data gap is the lack of basic mechanistic data for the
initiation of infection and dynamics of the early infection process. These mechanistic data would
greatly assist in the confirmation of appropriate dose metrics and inform the interspecies
extrapolation process. However, alternative dose metrics can be assessed for substantive
differences in outputs and the interspecies extrapolation process can be conducted in part to
address kinetic elements.
Science policy gaps also affect current readiness to generate a site-specific HHRA for B.
anthracis inhalation exposure. The selection of appropriate BMR targets for reporting and risk-
based decision-making for microbial pathogens is a current policy gap. While technical
knowledge may inform BMR selection relative to known data set characteristics for BMD
modeling, selection of values for reporting and risk-based decision-making may incorporate
numerous policy considerations. An additional science policy gap is the management of
uncertainty in the interspecies extrapolation given the current inability to address dynamic
differences between the animal model and the human. In addition to a statement of this
125
-------
uncertainty in the risk characterization, a default adjustment factor could be considered for use
until further data or methodologies are available.
Table 6-1. Summary Table for Data Gaps and Science Policy Gaps
Use of Microbial
Dose-Response
Data
Data Gaps
Science Policy Gaps
Hazard
Identification
including
Disease
Pathogenesis
• Identification of BMR values or
ranges
• Mechanistic data for the initiation
of infection and dynamics of the
early infection process necessary
for dose metric selection
Evaluation of
Microbial Dose-
Response Data
• High quality dose-response data
for the rabbit and nonhuman
primate
• Mechanistic data for the initiation
of infection and dynamics of the
early infection process necessary
for dose metric selection
• Identification of BMR values
or ranges to select POD for
microbial pathogens
Conduct
Interspecies
Extrapolation
• Lack of data to support inter-
species and intra-species UF
values
• Management of uncertainty in
the interspecies extrapolation
given the current inability to
address dynamic differences
between the animal model and
the human
BMR - benchmark dose response
POD - point of departure
UF - uncertainty factor
126
-------
7 References
Abramova, F. A., L. M. Grinberg, O. V. Yampolskaya and D. H. Walker (1993). Pathology of
Inhalational Anthrax in 42 Cases from the Sverdlovsk Outbreak of 1979. Proceedings of
the National Academy of Sciences of the United States of America 90(6): 2291-2294.
Albrink, W. S. (1961). Pathogenesis of Inhalation Anthrax. Bacteriological Reviews 25(3): 268-
273.
Albrink, W. S., S. M. Brooks, R. E. Biron and M. Kopel (1960). Human Inhalation Anthrax. A
Report of Three Fatal Cases. The American Journal of Pathology 36(4): 457-471.
Albrink, W. S. and R. J. Goodlow (1959). Experimental Inhalation Anthrax in the Chimpanzee.
The American Journal of Pathology 35(5): 1055-1065.
Andersen, M. E., R. B. Conolly, E. M. Faustman, R. J. Kavlock, C. J. Portier, D. M. Sheehan, P.
J. Weir and L. Ziese (1999). Quantitative Mechanistically Based Dose-Response
Modeling with Endocrine-Active Compounds. Environmental Health Perspectives
107(Supplement 4): 631-638.
Armstrong, T. W. and C. N. Haas (2007). A Quantitative Microbial Risk Assessment Model for
Legionnaires' Disease: Animal Model Selection and Dose-Response Modeling. Risk
Analysis 27(6): 1581-1596.
Bales, M. E., A. L. Dannenberg, P. S. Brachman, A. F. Kaufmann, P. C. Klatsky and D. A.
Ashford (2002). Epidemiological Response to Anthrax Outbreaks: Field Investigations,
1950-2001. Emerging Infectious Diseases 8(10): 1163-1174.
Barakat, L. A., H. L. Quentzel, J. A. Jernigan, D. L. Kirschke, K. Griffith, S. M. Spear, K.
Kelley, D. Barden, D. Mayo, D. S. Stephens, T. Popovic, C. Marston, S. R. Zaki, J.
Guarner, W. J. Shieh, H. W. Carver, 2nd, R. F. Meyer, D. L. Swerdlow, E. E. Mast, J. L.
Hadler and Anthrax Bioterrorism Investigation Team (2002). Fatal Inhalational Anthrax
in a 94-Year-Old Connecticut Woman. Journal of the American Medical Association
287(7): 863-868.
Bartrand, T. A., M. H. Weir and C. N. Haas (2008). Dose-Response Models for Inhalation of
Bacillus anthracis Spores: Interspecies Comparisons. Risk Analysis 28(4): 1115-1124.
127
-------
Beatty, M. E., D. A. Ashford, P. M. Griffin, R. V. Tauxe and J. Sobel (2003). Gastrointestinal
Anthrax: Review of the Literature. Archives of Internal Medicine 163(20): 2527-2531.
Bensman, M. D., R. S. Mackie, Z. A. Minter and B. W. Gutting (2012). Effect of Animal Sera on
Bacillus anthracis Sterne Spore Germination and Vegetative Cell Growth. Journal of
Applied Microbiology 113(2): 276-283.
Berdjis, C. C., C. A. Gleiser, H. A. Hartmen, R. W. Kuehne and W. S. Gochenour (1962).
Pathogenesis of Respiratory Anthrax in Macaca mulatta. British Journal of Experimental
Pathology 43: 515-524.
Borio, L., D. Frank, V. Mani, C. Chiriboga, M. Pollanen, M. Ripple, S. Ali, C. DiAngelo, J. Lee,
J. Arden, J. Titus, D. Fowler, T. O'Toole, H. Masur, J. Bartlett and T. Inglesby (2001).
Death Due to Bioterrorism-Related Inhalational Anthrax: Report of 2 Patients. JAMA :
Journal of the American Medical Association 286(20): 2554-2559.
Brachman, P. S. (1984). Chapter 87. Anthrax. Tropical and Geographic Medicine. K. S. Warren
and A. F. Mahmoud. (pp. 826-830). New York. McGraw-Hill.
Brachman, P. S., A. F. Kaufman and F. G. Dalldorf (1966). Industrial Inhalation Anthrax.
Bacteriological Reviews 30(3): 646-659.
Brachman, P. S., J. S. Pagano and W. S. Albrink (1961). Two Cases of Fatal Inhalation Anthrax,
One Associated with Sarcoidosis. New England Journal of Medicine 265(5): 203-208.
Brachman, P. S., S. A. Plotkin, F. H. Bumford and M. M. Atchison (1960). An Epidemic of
Inhalation Anthrax: The First in the Twentieth Century. II. Epidemiology. American
Journal of Hygiene 72(1): 6-23.
Bravata, D. M., G. S. Zaric, J. E. C. Holty, M. L. Brandeau, E. R. Wilhelm, K. M. McDonald and
D. K. Owens (2006). Reducing Mortality from Anthrax Bioterrorism: Strategies for
Stockpiling and Dispensing Medical and Pharmaceutical Supplies. Biosecurity and
Bioterrorism: Biodefense Strategy, Practice, and Science 4(3): 244-262.
Brittingham, K. C., G. Ruthel, R. G. Panchal, C. L. Fuller, W. J. Ribot, T. A. Hoover, H. A.
Young, A. O. Anderson and S. Bavari (2005). Dendritic Cells Endocytose Bacillus
anthracis Spores: Implications for Anthrax Pathogenesis. The Journal of Immunology
174: 5545-5552.
128
-------
Brookmeyer, R., E. Johnson and S. Barry (2005). Modelling the Incubation Period of Anthrax.
Statistics in Medicine 24(4): 531-542.
Brookmeyer, R., E. Johnson and R. Bollinger (2003). Modeling the Optimum Duration of
Antibiotic Prophylaxis in an Anthrax Outbreak. Proceedings of the National Academy of
Sciences of the United States of America 100(17): 10129-10132.
Buchanan, R. L., A. H. Havelaar, M. A. Smith, R. C. Whiting and E. Julien (2009). The Key
Events Dose-Response Framework: Its Potential for Application to Foodborne
Pathogenic Microorganisms. Critical Reviews in Food Science and Nutrition 49(8): 718-
728.
Buchanan, R. L., J. L. Smith and W. Long (2000). Microbial Risk Assessment: Dose-Response
Relations and Risk Characterization. International Journal of Food Microbiology 58(3):
159-172.
Bush, L. M., B. H. Abrams, A. Beall and C. C. Johnson (2001). Index Case of Fatal Inhalational
Anthrax Due to Bioterrorism in the United States. New England Journal of Medicine
345(22): 1607-1610.
Canter, D. A. (2005). Addressing Residual Risk Issues at Anthrax Cleanups: How Clean Is Safe?
Journal of Toxicology and Environmental Health Part A, 68(11-12): 1017-1032.
Coggeshall, K. M., F. Lupu, J. Ballard, J. P. Metcalf, J. A. James, D. Farris and S. Kurosawa
(2013). The Sepsis Model: An Emerging Hypothesis for the Lethality of Inhalation
Anthrax. Journal of Cellular and Molecular Medicine 17(7): 914-920.
Cohen, M. L. and T. Whalen (2007). Implications of Low Level Human Exposure to Respirable
B. anthracis. Applied Biosafety 12(2): 109-115.
Coleman, M. and H. Marks (1998). Topics in Dose-Response Modeling. Journal of Food
Protection 61(11): 1550-1559.
Coleman, M. E., B. Thran, S. S. Morse, M. Hugh-Jones and S. Massulik (2008). Inhalation
Anthrax: Dose Response and Risk Analysis. Biosecurity and Bioterrorism: Biodefense
Strategy, Practice, and Science 6(2): 147-160.
Cote, C. K., S. L. Welkos and J. Bozue (2011). Key Aspects of the Molecular and Cellular Basis
of Inhalational Anthrax. Microbes and Infection 13(14): 1146-1155.
129
-------
Craft, D. L., L. M. Wein and A. H. Wilkins (2005). Analyzing Bioterror Response Logistics: The
Case of Anthrax. Management Science 51(5): 679-694.
Crump, K., C. Chen, W. A. Chiu, T. A. Louis, C. J. Portier, R. P. Subramaniam and P. D. White
(2010). What Role for Biologically Based Dose-Response Models in Estimating Low-
Dose Risk? Environmental Health Perspectives 118(5): 585-588.
Curling, C. A., J. K. Burr, L. Danakian, D. S. Disraelly, L. A. LaViolet, T. J. Walsh and R. A.
Zirkle (2010). Technical Reference Manual: NATO Planning Guide for the Estimation of
Chemical, Biological, Radiological, and Nuclear (CBRN) Casualities, Allied Medical
Publication-8(C). Alexandria, VA: Institute for Defense Analyses. IDA-D-4082
Dahlgren, C. M., L. M. Buchanan, H. M. Decker, S. W. Freed, C. R. Phillips and P. S. Brachman
(1960). Bacillus anthracis Aerosols in Goat Hair Processing Mills. American Journal of
Hygiene 72(1): 24-31.
Dalldorf, F. G. and A. Kaufman (1966). Studies on Inhalation Anthrax III. Morphologic Studies
Technical Manuscript 290. United States Army Biological Center Fort Detrick.
CBRNIAC AD482650.
Dalldorf, F. G., A. F. Kaufman and P. S. Brachman (1971). Woolsorters' Disease. An
Experimental Model. Archives of Pathology 92(6): 418-426.
Defense Intelligence Agency (1986). Soviet Biological Warfare Threat\J.S. Department of
Defense. DST-1610F-057-86.
Dixon, T. C., A. A. Fadl, T. M. Koehler, J. A. Swanson and P. C. Hanna (2000). Early Bacillus
anthracis-Macrophage Interactions: Intracellular Survival and Escape. Cellular
Microbiology 2(6): 453-463.
Druett, H. A. (1952). Bacterial Invasion. Nature 170(4320): 288.
Druett, H. A., D. W. Henderson, L. Packman and S. Peacock (1953). Studies on Respiratory
Infection: I. The Influence of Particle Size on Respiratory Infection with Anthrax Spores.
The Journal of Hygiene (London) 51(3): 359-371.
Dull, P. M., K. E. Wilson, B. Kournikakis, E. A. S. Whitney, C. A. Boulet, J. Y. W. Ho, J.
Ogston, M. R. Spence, M. M. McKenzie, M. A. Phelan, T. Popovic and D. Ashford
130
-------
(2002). Bacillus anthracis Aerosolization Associated with a Contaminated Mail Sorting
Machine. Emerging Infectious Diseases 8(10): 1044-1047.
Dumetz, F., G. Jouvion, H. Khun, I. J. Glomski, J.-P. Corre, C. Rougeaux, W.-J. Tang, M. Mock,
M. Huerre and P. L. Goossens (2011). Noninvasive Imaging Technologies Reveal Edema
Toxin as a Key Virulence Factor in Anthrax. The American Journal of Pathology 178(6):
2523-2535.
Ezzell, J. W., T. G. Abshire, R. Panchal, D. Chabot, S. Bavari, E. K. Leffel, B. Purcell, A. M.
Friedlander and W. J. Ribot (2009). Association of Bacillus anthracis Capsule with
Lethal Toxin During Experimental Infection. Infection and Immunity 77(2): 749-755.
Fellows, P. F., M. K. Linscott, B. E. Ivins, M. L. M. Pitt, C. A. Rossi, P. H. Gibbs and A. M.
Friedlander (2001). Efficacy of a Human Anthrax Vaccine in Guinea Pigs, Rabbits, and
Rhesus Macaques against Challenge by Bacillus anthracis Isolates of Diverse
Geographical Origin. Vaccine 19(23-24): 3241-3247.
Fisher, N., K. A. Carr, J. Giebel and P. C. Hanna (2011). Chapter 3: Anthrax Spore Germination.
Bacillus anthracis and Anthrax. N. H. Bergman, (pp. 39-51). Hoboken, NJ. John Wiley
& Sons, Inc.
Food and Agriculture Organization and World Health Organization (FAO and WHO) (2003).
Hazard Characterization for Pathogens in Food and Water, Guidelines. Rome, Italy and
Geneva, Switzerland: World Health Organization. Microbiological Risk Assessment
Series No. 3.
Franz, D. R., P. B. Jahrling, A. M. Friedlander, D. J. McClain, D. L. Hoover, W. R. Bryne, J. A.
Pavlin, G. W. Christopher and E. M. Eitzen, Jr. (1997). Clinical Recognition and
Management of Patients Exposed to Biological Warfare Agents. Journal of the American
Medical Association 278(5): 399-411.
Freedman, A., O. Afonja, M. W. Chang, F. Mostashari, M. Blaser, G. Perez-Perez, H. Lazarus,
R. Schacht, J. Guttenberg, M. Traister and W. Borkowsky (2002). Cutaneous Anthrax
Associated with Microangiopathic Hemolytic Anemia and Coagulopathy in a 7-Month-
Old Infant. Journal of the American Medical Association 287(7): 869-874.
Friedlander, A. M. (2001). Tackling Anthrax. Nature 414(6860): 160-161.
Friedlander, A. M., S. L. Welkos, M. L. M. Pitt, J. W. Ezzell, P. L. Worsham, K. J. Rose, B. E.
Ivins, J. R. Lowe, G. B. Howe, P. Mikesell and W. B. Lawrence (1993). Postexposure
131
-------
Prophylaxis against Experimental Inhalation Anthrax. Journal of Infectious Diseases
167(5): 1239-1243.
Fritz, D. L., N. K. Jaax, W. B. Lawrence, K. J. Davis, M. L. Pitt, J. W. Ezzell and A. M.
Friedlander (1995). Pathology of Experimental Inhalation Anthrax in the Rhesus
Monkey. Laboratory Investigation 73(5): 691-702.
Gill, J. R. and J. Melinek (2002). Inhalation of Anthrax: Gross Autopsy Findings. Archives of
Pathology and Laboratory Medicine 126(8): 993-994.
Glassman, H. N. (1966). Industrial Inhalation Anthrax - Discussion. Bacteriological Reviews
30(3): 657-659.
Gleiser, C. A. (1967). Pathology of Anthrax Infection in Animal Hosts. Federation Proceedings
26(5): 1518-1521.
Gleiser, C. A., C. C. Berdjis, H. A. Hartman and W. S. Gochenour, Jr., (1963). Pathology of
Experimental Respiratory Anthrax in Macaca mulatta. British Journal of Experimental
Pathology 44: 416-426.
Gleiser, C. A., W. S. Gochenour, Jr., and M. K. Ward (1968). Pulmonary Lesions in Dogs and
Pigs Exposed to a Cloud of Anthrax Spores. Journal of Comparative Pathology 78(4):
445-448.
Glomski, I. J., A. Piris-Gimenez, M. Huerre, M. Mock and P. L. Goossens (2007). Primary
Involvement of Pharynx and Peyer's Patch in Inhalational and Intestinal Anthrax. PloS
Pathogens 3(6): e76.
Gochenour, W. S., Jr. et al., (1961). Studies on Bacillus anthracis, Parts 1 -7. Army Medical
Unit, Frederick MD. CB-023805.
Gochenour, W. S., Jr., C. A. Gleiser and W. D. Tigertt (1962). Observations on Penicillin
Prophylaxis of Experimental Inhalation Anthrax in the Monkey. Journal of Hygiene 60:
29-33.
Gold, H. (1955). Anthrax: A Report of One Hundred Seventeen Cases. AMA Archives of
Internal Medicine 96(3): 387-396.
132
-------
Goodlow, R. J. and F. A. Leonard (1961). Viability and Infectivity of Microorganisms in
Experiemental Airborne Infection. Bacteriological Reviews 25(3): 182-187.
Goossens, P. L. (2009). Animal Models of Human Anthrax: The Quest for the Holy Grail.
Molecular Aspects of Medicine 30(6): 467-480.
Griffith, J., D. Blaney, S. Shadomy, M. Lehman, N. Pesik, S. Tostenson, L. Delaney, R. Tiller,
A. DeVries, T. Gomez, M. Sullivan, C. Blackmore, D. Stanek, R. Lynfield and the
Anthrax Investigation Team (2014). Investigation of Inhalation Anthrax Case, United
States. Emerging Infectious Diseases 20(2): 280-283.
Grinberg, L. M., F. A. Abramova, O. V. Yampolskaya, D. H. Walker and J. H. Smith (2001).
Quantitative Pathology of Inhalational Anthrax I: Quantitative Microscopic Findings.
Modern Pathology 14(5): 482-495.
Grunow, R., S. R. Klee, W. Beyer, M. George, D. Grunow, A. Barduhn, S. Klar, D. Jacob, M.
Elschner, P. Sandven, A. Kjerulf, J. S. Jensen, W. Cai, R. Zimmerman and L. Schaade
(2013). Anthrax among Heroin Users in Europe Possibly Caused by Same Bacillus
anthracis Strain since 2000. Eurosurveillance 18(13): pii=20437.
Guarner, J. and C. del Rio (2011). Pathology, Diagnosis, and Treatment of Anthrax in Humans.
Bacillus anthracis and Anthrax. N. H. Bergman, (pp. 251-268). Hoboken, New Jersey.
John Wiley & Sons, Inc.
Guarner, J., J. A. Jernigan, W.-J. Shieh, K. Tatti, L. M. Flannagan, D. S. Stephens, T. Popovic,
D. A. Ashford, B. A. Perkins, S. R. Zaki and the Inhalational Anthrax Pathology Working
Group (2003). Pathology and Pathogenesis of Bioterrorism-Related Inhalational Anthrax.
The American Journal of Pathology 163(2): 701-709.
Guichard, A., V. Nizet and E. Bier (2012). New Insights into the Biological Effects of Anthrax
Toxins: Linking Cellular to Organismal Responses. Microbes and Infection 14(2): 97-
118.
Guidi-Rontani, C. (2002). The Alveolar Macrophage: The Trojan Horse of Bacillus anthracis.
Trends in Microbiology 10(9): 405-409.
Guidi-Rontani, C., Y. Pereira, S. Ruffle, J.-C. Sirard, M. Weber-Levy and M. Mock (1999).
Identification and Characterization of a Germination Operon on the Virulence Plasmid
pXOl of Bacillus anthracis. Molecular Microbiology 33(2): 407-414.
133
-------
Gutting, B. W., S. R. Channel, A. E. Berger, J. M. Gearhart, G. A. Andrews, R. L. Sherwood and
T. L. Nichols (2008). Mathematically Modeling Inhalational Anthrax. Microbe 3(2): 78-
85.
Gutting, B. W., D. Marchette, R. Sherwood, G. A. Andrews, A. Director-Myska, S. R. Channel,
D. Wolfe, A. E. Berger, R. S. Mackie, B. J. Watson and A. Rukhin (2013). Modeling
Low-Dose Mortality and Disease Incubation Period of Inhalational Anthrax in the Rabbit.
Journal of Theoretical Biology 329: 20-31.
Gutting, B. W., T. L. Nichols, S. R. Channel, J. M. Gearhart, G. A. Andrews, A. E. Berger, R. S.
Mackie, B. J. Watson, S. C. Taft, K. A. Overheim and R. L. Sherwood (2012)
"Inhalational Anthrax (Ames Aerosol) in Naive and Vaccinated New Zealand Rabbits:
Characterizing the Spread of Bacteria from Lung Deposition to Bacteremia." Frontiers in
Cellular and Infection Microbiology 2: 87. doi: 10.3389/fcimb.2012.00087.
Haas, C. N. (2002). On the Risk of Mortality to Primates Exposed to Anthrax Spores. Risk
Analysis 22(2): 189-193.
Haas, C. N., J. B. Rose and C. P. Gerba (1999a). Chapter 7. Conducting the Dose-Response
Assessment. Quantitative Microbial Risk Assessment, (pp. 260-319). New York. John
Wiley & Sons, Inc.
Haas, C. N., J. B. Rose and C. P. Gerba (1999b). Quantitative Microbial Risk Assessment. New
York, John Wiley & Sons, Inc.
Hanna, P. C., D. Acosta and R. J. Collier (1993). On the Role of Macrophages in Anthrax.
Proceedings of the National Academy of Sciences of the United States of America
90(21): 10198-10201.
Henderson, D. W., S. Peacock and F. C. Belton (1956). Observations on the Prophylaxis of
Experimental Pulmonary Anthrax in the Monkey. Journal of Hygiene (London) 54(1):
28-36.
Henning, L. N., J. E. Comer, G. V. Stark, B. D. Ray, K. P. Tordoff, K. A. B. Knostman and G. T.
Meister (2012). Development of an Inhalational Bacillus anthracis Exposure Therapeutic
Model in Cynomolgus Macaques. Clinical and Vaccine Immunology 19(11): 1765-1775.
134
-------
Hilmas, C. J., A. M. Katos, P. T. Williams and J. Anderson (2009). Chapter 31. Anthrax.
Handbook of Toxicology of Chemical Warfare Agents. R. C. Gupta, (pp. 433-462).
Boston. Academic Press.
Hines, S. A. and J. E. Comer (2012). Identification of Critical Data Gaps in the Development of
a Key Events Dose-Response Framework for Bacillus anthracis. 2012 Annual Meeting of
the Society for Risk Analysis, Hyatt Regency San Francisco, December 9-12, 2012.
Ho, J. and S. Duncan (2005). Estimating Aerosol Hazards from an Anthrax Letter. Journal of
Aerosol Science 36(5-6): 701-719.
Holty, J.-E. C., D. M. Bravata, H. Liu, R. A. Olshen, K. M. McDonald and D. K. Owens (2006).
Systematic Review: A Century of Inhalational Anthrax Cases from 1900 to 2005.
Annuals of Internal Medicine 144(4): 270-280.
Huang, Y. and C. N. Haas (2011). Quantification of the Relationship between Bacterial Kinetics
and Host Response for Monkeys Exposed to Aerosolized Francisella tularensis. Applied
and Environmental Microbiology 77(2): 485-490.
Hupert, N., D. Wattson, J. Cuomo, E. Hollingsworth, K. Neukermans and W. Xiong (2009).
Predicting Hospital Surge after a Large-Scale Anthrax Attack: A Model-Based Analysis
of CDC's Cities Readiness Initiative Prophylaxis Recommendations. Medical Decision
Making 29(4): 424-437.
Inglesby, T. V., D. A. Henderson, J. G. Bartlett, M. S. Ascher, E. Eitzen, A. M. Friedlander, J.
Hauer, J. McDade, M. T. Osterholm, T. O'Toole, G. Parker, T. M. Perl, P. K. Russell and
K. Tonat (1999). Anthrax as a Biological Weapon: Medical and Public Health
Management. Working Group on Civilian Biodefense. Journal of the American Medical
Association 281(18): 1735-1745.
Inglesby, T. V., T. O'Toole, D. A. Henderson, J. G. Bartlett, M. S. Ascher, E. Eitzen, A. M.
Friedlander, J. Gerberding, J. Hauer, J. M. Hughes, J. McDade, M. T. Osterholm, G.
Parker, T. M. Perl, P. K. Russell, and K. Tonat for the Working Group on Civilian
Biodefense (2002). Anthrax as a Biological Weapon, 2002. Updated Recommendations
for Management. Journal of the American Medical Association 287(17): 2236-2252.
International Life Sciences Institute (ILSI) (2000). Revised Framework for Microbial Risk
Assessment. Washington, D.C.: ILSI Press.
135
-------
Ioannidis, J. P. A. (2012). Extrapolating from Animals to Humans. Science Translational
Medicine 4(151): 15 lpsl 15.
Ivins, B. E., P. F. Fellows, M. L. M. Pitt, J. E. Estep, S. L. Welkos, P. L. Worsham and A. M.
Friedlander (1996). Efficacy of a Standard Human Anthrax Vaccine against Bacillus
anthracis Aerosol Spore Challenge in Rhesus Monkeys. Salisbury Medical Bulletin
Special Supplement Number 87: 125-126.
Ivins, B. E., M. L. M. Pitt, P. F. Fellows, J. W. Farchaus, G. E. Benner, D. M. Waag, S. F. Little,
G. W. Anderson, Jr., P. H. Gibbs and A. M. Friedlander (1998). Comparative Efficacy of
Experimental Anthrax Vaccine Candidates against Inhalation Anthrax in Rhesus
Monkeys. Vaccine 16(11-12): 1141-1148.
Jackson, P. J., M. E. Hugh-Jones, D. M. Adair, G. Green, K. K. Hill, C. R. Kuske, L. M.
Grinberg, F. A. Abramova and P. Keim (1998). PCR Analysis of Tissue Samples from
the 1979 Sverdlovsk Anthrax Victims: The Presence of Multiple Bacillus anthracis
Strains in Different Victims. Proceedings of the National Academy of Sciences of the
United States of America 95(3): 1224-1229.
Jarabek, A. M., B. Asgharian and F. J. Miller (2005). Dosimetric Adjustments for Interspecies
Extrapolation of Inhaled Poorly Soluble Particles (PSP). Inhalation Toxicology 17(7-8):
317-334.
Jenkins, S. A. and Y. Xu (2013). Characterization of Bacillus anthracis Persistence In Vivo.
PLoS One 8(6): e66177.
Jernigan, J. A., D. S. Stephens, D. A. Ashford, C. Omenaca, M. S. Topiel, M. Galbraith, M.
Tapper, T. L. Fisk, S. Zaki, T. Popovic, R. F. Meyer, C. P. Quinn, S. A. Harper, S. K.
Fridkin, J. J. Sejvar, C. W. Shepard, M. McConnell, J. Guarner, W.-J. Shieh, J. M.
Malecki, J. L. Gerberding, J. M. Hughes, B. A. Perkins and Members of the Anthrax
Bioterrorism Investigation Team (2001). Bioterrorism-Related Inhalational Anthrax: The
First 10 Cases Reported in the United States. Emerging Infectious Diseases 7(6): 933-
944.
Julien, E., A. R. Boobis, S. S. Olin and and the ILSI Research Foundation Threshold Working
Group (2009). The Key Events Dose-Response Framework: A Cross-Disciplinary Mode-
of-Action Based Approach to Examining Dose-Response and Thresholds. Critical
Reviews in Food and Science Nutrition 49(8): 682-689.
Kabilan, S., S. R. Suffield, K. P. Recknagle, R. E. Jacob, D. R. Einstein, A. P. Kuprat, J. P.
Carson, S. M. Colby, J. H. Saunders, S. A. Hines, J. G. Teeguarden, T. M. Straub, M.
136
-------
Moe, S. C. Taft and R. A. Corley (2015). Computational Fluid Dynamics Modeling of
Bacillus anthracis Spore Deposition in Rabbit and Human Respiratory Airways. In
Preparation.
Kleinstreuer, C., Z. Zhang and Z. Li (2008). Modeling Airflow and Particle Transport/Deposition
in Pulmonary Airways. Respiratory Physiology & Neurobiology 163(1-30): 128-138.
LaForce, F. M., F. H. Bumford, J. C. Feeley, S. L. Stokes and D. B. Snow (1969). Epidemiologic
Study of a Fatal Case of Inhalation Anthrax. Archives of Environmental Health 18(5):
798-805.
Layshock, J. A., B. Pearson, K. Crockett, M. J. Brown, S. Van Cuyk, W. B. Daniel and K. M.
Omberg (2012). Reaerosolization of Bacillus spp. in Outdoor Environments: A Review
of the Experimental Literature. Biosecurity and Bioterrorism: Biodefense Strategy,
Practice, and Science 10(3): 299-303.
Leffel, E. and L. M. Pitt (2006). Chapter 6. Anthrax. Biodefense: Research Methodology and
Animal Models. J. R. Swearengen. (pp. 77-94). CRC Press.
Lever, M. S., A. J. Stagg, M. Nelson, P. Pearce, D. J. Stevens, E. A. M. Scott, A. J. H. Simpson
and M. J. Fulop (2008). Experimental Respiratory Anthrax Infection in the Common
Marmoset (Callithrix jacchus). International Journal of Experimental Pathology 89(3):
171-179.
Levy, H., I. Glinert, S. Weiss, A. Sittner, J. Schlomovitz, Z. Altboum and D. Kobiler (2014).
Toxin-Independent Virulence of Bacillus anthracis in Rabbits. PLoS One 9(1): e84947.
Lincoln, R. E., D. R. Hodges, F. Klein, B. G. Mahlandt, W. I. Jones, Jr., B. W. Haines, M. A.
Rhian and J. S. Walker (1965). Role of the Lymphatics in the Pathogenesis of Anthrax.
Journal of Infectious Diseases 115(5): 481-494.
Lincoln, R. E., J. S. Walker, F. Klein, A. J. Rosenwald and W. I. Jones, Jr. (1967). Value of Field
Data for Extrapolation in Anthrax. Federation Proceedings 26(5): 1558-1562.
Lovchik, J. A., M. Drysdale, T. M. Koehler, J. A. Hutt and C. R. Lyons (2012). Expression of
Either Lethal Toxin or Edema Toxin by Bacillus anthracis Is Sufficient for Virulence in a
Rabbit Model of Inhalational Anthrax. Infection and Immunity 80(7): 2414-2425.
137
-------
Lowe, D. E., S. M. C. Ernst, C. Zito, J. Ya and I. J. Glomski (2013). Bacillus anthracis Has Two
Independent Bottlenecks That Are Dependent on the Portal of Entry in an Intranasal
Model of Inhalational Infection. Infection and Immunity 81(12): 4408-4420.
Lowe, D. E. and I. J. Glomski (2012) "Cellular and Physiological Effects of Anthrax Exotoxin
and Its Relevance to Disease." Frontiers in Cellular and Infection Microbiology 2: 76.
doi: 10.3389/fcimb.2012.00076.
MacDonald, W. D. (1942). Anthrax: Report of a Fatal Case Involving the Cutaneous and
Gastrointestinal Systems. The New England Journal of Medicine 226(24): 949-951.
Marston, C. K., C. A. Allen, J. Beaudry, E. P. Price, S. R. Wolken, T. Pearson, P. Keim and A.
R. Hoffmaster (2011). Molecular Epidemiology of Anthrax Cases Associated with
Recreational Use of Animal Hides and Yarn in the United States. PLoS One 6(12):
e28274.
Martchenko, M., S. I. Candille, H. Tang and S. N. Cohen (2012). Human Genetic Variation
Altering Anthrax Toxin Sensitivity. Proceedings of the National Academy of Sciences of
the United States of America 109(8): 2972-2977.
Mayer, B. T., J. S. Koopman, E. L. Ionides, J. M. Pujol and J. N. S. Eisenberg (2011). A
Dynamic Dose-Response Model to Account for Exposure Patterns in Risk Assessment: A
Case Study in Inhalation Anthrax. Journal of the Royal Society Interface 8(57): 506-517.
Meaney-Delman, D., M. E. Zotti, S. A. Rasmussen, S. Strasser, S. Shadomy, R. M. Turcios-
Ruiz, G. D. Wendel, Jr., T. A. Treadwell and D. J. Jamieson (2012). Anthrax Cases in
Pregnant and Postpartum Women: A Systematic Review. Obstetrics and Gynecology
120(6): 1439-1449.
Meek, M. E., A. Boobis, I. Cote, V. Dellarco, G. Fotakis, S. Munn, J. Seed and C. Vickers
(2014). New Developments in the Evolution and Application of the WHO/IPCS
Framework on Mode of Action/Species Concordance Analysis. Journal of Applied
Toxicology 34(1): 1-18.
Middleton, G. K. and A. C. Standen (1961). The Electrocardiogram in Fatal Anthrax Bacteremia.
The Journal of Infectious Diseases 108: 85-89.
Mina, B., J. P. Dym, F. Kuepper, R. Tso, C. Arrastia, I. Kaplounova, H. Faraj, A. Kwapniewski,
C. M. Krol, M. Grosser, J. Glick, S. Fochios, A. Remolina, L. Vasovic, J. Moses, T.
Robin, M. DeVita and M. L. Tapper (2002). Fatal Inhalational Anthrax with Unknown
138
-------
Source of Exposure in a 61-Year-Old Woman in New York City. Journal of the
American Medical Association 287(7): 858-862.
Moayeri, M. and S. H. Leppla (2009). Cellular and Systemic Effects of Anthrax Lethal Toxin
and Edema Toxin. Molecular Aspects of Medicine 30(6): 439-455.
Morens, D. M. (2002). Epidemic Anthrax in the Eighteenth Century, the Americas. Emerging
Infectious Diseases 8(10): 1160-1162.
Nordberg, B. K., C. G. Schmiterlow and H. J. Hansen (1961). Pathophysiological Investigations
into the Terminal Course of Experimental Anthrax in the Rabbit. Acta Pathologica et
Microbiologica Scandinavica 53: 295-318.
Passalacqua, K. D. and N. H. Bergman (2006). Bacillus anthracis: Interactions with the Host and
Establishment of Inhalational Anthrax. Future Microbiology 1(4): 397-415.
Peters, C. J. and D. M. Hartley (2002). Anthrax Inhalation and Lethal Human Infection. The
Lancet 359(9307): 710-711.
Peterson, J. W., J. E. Comer, W. B. Baze, D. M. Noffsinger, A. Wenglikowski, K. G. Walberg, J.
Hardcastle, J. Pawlik, K. Bush, J. Taormina, S. Moen, J. Thomas, B. M. Chatuev, L.
Sower, A. K. Chopra, L. R. Stanberry, R. Sawada, W. W. Scholz and J. Sircar (2007).
Human Monoclonal Antibody AVP-21D9 to Protective Antigen Reduces Dissemination
of the Bacillus anthracis Ames Strain from the Lungs in a Rabbit Model. Infection and
Immunity 75(7): 3414-3424.
Pitt, M. L. M. and R. D. LeClaire (2005). Pathogenesis by Aerosol. Biological Weapons
Defense: Infectious Diseases and Counterterrorism. L. E. Lindler, F. J. Lebeda and G. W.
Korch. (pp. 65-78). Totowa, NJ. Humana Press. Inc.
Plotkin, S. A., P. S. Brachman, M. Utell, F. H. Bumford and M. M. Atchison (2002). An
Epidemic of Inhalation Anthrax, the First in the Twentieth Century: I. Clinical Features.
(Reprint). The American Journal of Medicine 112(1): 4-12; Discussion 12-13.
Portier, C. J. and C. M. Lyles (1996). Practicing Safe Modeling: GLP for Biologically Based
Mechanistic Models. Environmental Health Perspectives 104(8): 806.
139
-------
Pujol, J. M., J. E. Eisenberg, C. N. Haas and J. S. Koopman (2009). The Effect of Ongoing
Exposure Dynamics in Dose Response Relationships. PLoS Computational Biology 5(6):
el000399.
Raabe, O. G., M. A. Al-Bayati, S. V. Teague and A. Rasolt (1988). Regional Deposition of
Inhaled Monodisperse Coarse and Fine Aerosol Particles in Small Laboratory Animals.
Annuals of Occupational Hygiene 32(Supplement 1): 53-63.
Rayamajhi, M., C. Delgado, T. V. Condon, D. W. Riches and L. L. Lenz (2012). Lung B Cells
Promote Early Pathogen Dissemination and Hasten Death from Inhalation Anthrax.
Mucosal Immunology 5(4): 444-454.
Redmond, C., G. A. Hall, P. C. Turnbull and J. S. Gillgan (1997). Experimentally Assessed
Public Health Risks Associated with Pigs from Farms Experiencing Anthrax. The
Veterinary Record 141(10): 244-247.
Reissman, D. B., E. A. S. Whitney, T. H. Taylor, Jr., J. A. Hayslett, P. M. Dull, I. Arias, D. A.
Ashford, E. A. Bresnitz, C. Tan, N. Rosenstein and B. A. Perkins (2004). One-Year
Health Assessment of Adult Survivors of Bacillus anthracis Infection. Journal of the
American Medical Association 291(16): 1994-1998.
Remy, K. E., P. Qiu, Y. Li, X. Cui and P. Q. Eichacker (2013). B. anthracis Associated
Cardiovascular Dysfunction and Shock: The Potential Contribution of Both Non-Toxin
and Toxin Components. BMC Med 11(1): 217.
Renwick, A. G. (1993). Data-Derived Safety Factors for the Evaluation of Food Additives and
Environmental Contaminants. Food Additives and Contaminants 10(3): 275-305.
Rickmeier, G. L., G. E. McClellan and G. A. Anno (2001). Biological Warfare Human Response
Modeling. Military Operations Research 6(3): 35-57.
Rossi, C. A., M. Ulrich, S. Norris, D. S. Reed, L. M. Pitt and E. K. Leffel (2008). Identification
of a Surrogate Marker for Infection in the African Green Monkey Model of Inhalation
Anthrax. Infection and Immunity 76(12): 5790-5801.
Russell, B. H., Q. Liu, S. A. Jenkins, M. J. Tuvim, B. F. Dickey and Y. Xu (2008). In Vivo
Demonstration and Quantification of Intracellular Bacillus anthracis in Lung Epithelial
Cells. Infection and Immunity 76(9): 3975-3983.
140
-------
Saile, E., G. J. Boons, T. Buskas, R. W. Carlson, E. L. Kannenberg, J. R. Barr, A. E. Boyer, M.
Gallegos-Candela and C. P. Quinn (2011). Antibody Responses to a Spore Carbohydrate
Antigen as a Marker of Nonfatal Inhalation Anthrax in Rhesus Macaques. Clinical and
Vaccine Immunology 18(5): 743-748.
Sanz, P., L. D. Teel, F. Alem, H. M. Carvalho, S. C. Darnell and A. D. O'Brien (2008). Detection
of Bacillus anthracis Spore Germination in Vivo by Bioluminescence Imaging. Infection
and Immunity 76(3): 1036-1047.
Schlingman, A. S., H. B. Devlin, G. G. Wright, R. J. Maine and M. C. Manning (1956).
Immunizing Activity of Alum-Precipitated Protective Antigen of Bacillus anthracis in
Cattle, Sheep, and Swine. American Journal of Veterinary Research 17(63): 256-261.
Shadomy, S. V. and T. L. Smith (2008). Zoonosis Update. Anthrax. Journal of the American
Veterinary Medical Association 233(1): 63-72.
Shafa, F., B. J. Moberly and P. Gerhardt (1966). Cytological Features of Anthrax Spores
Phagocytized in Vitro by Rabbit Alveolar Macrophages. The Journal of Infectious
Diseases 116(4): 401-413.
Sirisanthana, T. and A. E. Brown (2002). Anthrax of the Gastrointestinal Tract. Emerging
Infectious Diseases 8(7): 649-651.
Stearns-Kurosawa, D. J., F. Lupu, F. B. Taylor, Jr., G. Kinasewitz and S. Kurosawa (2006).
Sepsis and Pathophysiology of Anthrax in aNonhuman Primate Model. The American
Journal of Pathology 169(2): 433-444.
Suffin, S. C., W. H. Carnes and A. F. Kaufmann (1978). Inhalation Anthrax in a Home
Craftsman. Human Pathology 9(5): 594-597.
Taft, S. C. and S. A. Hines (2012). Benchmark Dose Analysis for Bacillus anthracis Inhalation
Exposures in the Nonhuman Primate. Risk Analysis 32(10): 1750-1768.
Taketoh, J., S. Komatsu, K. Adachi, R. Takai and K. Asanuma (2009). Application of an
Indwelling Vascular Access Port for Intravenous Administration in a Repeated and
Intermittent Dose Toxicity Study in Rats. The Journal of Toxicological Sciences 34(1):
39-52.
141
-------
ten Berge, W. F., A. Zwart and L. M. Appelman (1986). Concentration-Time Mortality Response
Relationship of Irritant and Systemically Acting Vapours and Gases. Journal of
Hazardous Materials 13(3): 301-309.
Thomas, R., C. Davies, A. Nunez, S. Hibbs, H. Flick-Smith, L. Eastaugh, S. Smither, A. Gates,
P. Oyston, T. Atkins and S. Eley (2010). Influence of Particle Size on the Pathology and
Efficacy of Vaccination in a Murine Model of Inhalational Anthrax. Journal of Medical
Microbiology 59(12): 1415-1427.
Thomas, R. J. (2013). Particle Size and Pathogenicity in the Respiratory Tract. Virulence 4(8):
847-858.
Toth, D. J. A., A. V. Gundlapalli, W. A. Schell, K. Bulmahn, T. E. Walton, C. W. Woods, C.
Coghill, F. Gallegos, M. H. Samore and F. R. Adler (2013). Quantitative Models of the
Dose-Response and Time Course of Inhalational Anthrax in Humans. PloS Pathogens
9(8): el003555.
Tournier, J.-N., A. Quesnel-Hellmann, A. Cleret and D. R. Vidal (2007). Contribution of Toxins
to the Pathogenesis of Inhalational Anthrax. Cellular Microbiology 9(3): 555-565.
Tournier, J.-N., A. Quesnel-Hellmann, J. Mathieu, C. Montecucco, W.-J. Tang, M. Mock, D. R.
Vidal and P. L. Goossens (2005). Anthrax Edema Toxin Cooperates with Lethal Toxin to
Impair Cytokine Secretion During Infection of Dendritic Cells. The Journal of
Immunology 174(8): 4934-4941.
Twenhafel, N. A. (2010). Pathology of Inhalational Anthrax Animal Models. Veterinary
Pathology 47(5): 819-830.
Twenhafel, N. A., E. Leffel and M. L. M. Pitt (2007). Pathology of Inhalational Anthrax
Infection in the African Green Monkey. Veterinary Pathology 44(5): 716-721.
U.S. Army Medical Research Institute of Infectious Diseases (2011). USAMRIID's Medical
Management of Biological Casualties Handbook. Washington, DC: U.S. Government
Printing Office.
U.S. Centers for Disease Control and Prevention (2000). Human Ingestion of Bacillus anthracis-
Contaminated Meat - Minnesota, August 2000. MMWR. Morbidity and Mortality
Weekly Report 49(36): 813-816.
142
-------
U.S. Centers for Disease Control and Prevention (2006). Inhalation Anthrax Associated with
Dried Animal Hides - Pennsylvania and New York City, 2006. MMWR. Morbidity and
Mortality Weekly Report 55(10): 280-282.
U.S. Centers for Disease Control and Prevention (2009). ESF-8 Aerosolized Anthrax Playbook,
U.S. Centers for Disease Control and Prevention.
U.S. Centers for Disease Control and Prevention (2010). Gastrointestinal Anthrax after an
Animal-Hide Drumming Event - New Hampshire and Massachusetts, 2009 MMWR.
Morbidity and Mortality Weekly Report 59(28): 872-877.
U.S. Communicable Disease Center (1961). Epidemiological Reports. MMWR. Morbidity and
Mortality Weekly Report 10(28): 2.
U.S. Department of Homeland Security and U.S. Environmental Protection Agency (2009).
Planning Guidance for Recovery Following Biological Incidents (Draft). Washington
DC: Biological Decontamination Standards Working Group, National Science and
Technology Council, U.S. Department of Homeland Security.
U.S. Environmental Protection Agency (1994a). Methods for the Derivation of Inhalation
Reference Concentrations and Application of Inhalation Dosimetry Washington, DC:
Office of Research and Development, U.S. Environmental Protection Agency.
EPA/600/8-90/066F.
U.S. Environmental Protection Agency (1994b). Regionally Deposited Dose Ratio (RDDR)
Model Software. Version 2.3.
U.S. Environmental Protection Agency (2002). A Review of the Reference Dose and Reference
Concentration Processes. Washington, DC: Risk Assessment Forum, U.S. Environmental
Protection Agency. EPA/630/P-02/002F.
U.S. Environmental Protection Agency (2003). A Summary of General Assessment Factors for
Evaluating the Quality of Scientific and Technical Information. Washington, DC: Science
Policy Council, U.S. Environmental Protection Agency. EPA/100/B-03/001.
U.S. Environmental Protection Agency (2006). Approaches for the Application of
Physiologically Based Pharmacokinetic (PBPK) Models and Supporting Data in Risk
Assessment. Washington, DC: National Center for Environmental Assessment Office of
Research and Development. EPA/600/R-05/043F.
143
-------
U.S. Environmental Protection Agency (2010a). Benchmark Dose Analysis for Bacillus
anthracis Inhalation Exposures in the Nonhuman Primate and Application to Risk-Based
Decision Making. Washington, DC: National Homeland Security Research Center Office
of Research and Development, U.S. Environmental Protection Agency. EPA 600/R-
10/138.
U.S. Environmental Protection Agency (2010b). Workshop Report: State-of-the-Science for the
Determination and Application of Dose-Response Relationships in Microbial Risk
Assessment. Washington, DC: Office of Research and Development National Homeland
Security Research Center. EPA/600/R-10/022.
U.S. Environmental Protection Agency (201 la). Acute Low Dose Bacillus anthracis Ames
Inhalation Exposures in the Rabbit. Cincinnati, OH: National Homeland Security
Research Center, U.S. Environmental Protection Agency. EPA/600/R-11/075.
U.S. Environmental Protection Agency (201 lb). Exposure Factors Handbook: 2011 Edition.
Washington, DC: Office of Research and Development, U.S. Environmental Protection
Agency. EPA/600/R-09/052F.
U.S. Environmental Protection Agency. (2011c). "IRIS Glossary." from
http://www.epa.gov/ncea/iris/index.html.
U.S. Environmental Protection Agency (2012a). Benchmark Dose Technical Guidance.
Washington DC: Office of the Science Adviser Risk Assessment Forum, U.S.
Environmental Protection Agency. EPA/100/R-12/001.
U.S. Environmental Protection Agency (2012b). Multiple Daily Low-Dose Bacillus anthracis
Ames Inhalation Exposures in the Rabbit. Cincinnati, OH: Office of Research and
Development National Homeland Security Research Center, U.S. Environmental
Protection Agency. EPA/600/R-11/145.
U.S. Environmental Protection Agency (2012c). Technical Brief: Review of Bacillus anthracis
(Anthrax) Studies for Dose-Response Modeling to Estimate Risk. Washington, DC: Office
of Research and Development National Homeland Security Research Center, U.S.
Environmental Protection Agency. EPA/600/S-12/681.
U.S. Environmental Protection Agency (2014a). Framework for Human Health Risk Assessment
to Inform Decision Making. Washington, DC: Office of the Science Adviser Risk
Assessment Forum, U.S. Environmental Protection Agency. EPA/100/R-14/001.
144
-------
U.S. Environmental Protection Agency (2014b). Guidance for Applying Quantitative Data to
Develop Data-Derived Extrapolation Factors for Interspecies and Intraspecies
Extrapolation. Washington, DC: Office of the Science Advisor Risk Assessment Forum,
U.S. Environmental Protection Agency.
U.S. Environmental Protection Agency (2014c). Microbiological Risk Assessment (MRA) Tools,
Methods, and Approaches for Water Media. Washington DC: Office of Science and
Technology Office of Water, U.S. Environmental Protection Agency. EPA-820-R-14-
009.
U.S. Environmental Protection Agency (2014d). Modeling Rabbit Responses to Acute and
Multiple Inhalation Exposures of Bacillus anthracis Spores. Cincinnati, Ohio: National
Homeland Security Research Center, U.S. Environmental Protection Agency.
EPA/600/R-14/058.
U.S. Food and Drug Administration (2002). Anthrax Vaccines: Efficacy Testing and Surrogate
Markers of Immunity Workshop. Bethesda Maryland: U. S. University.
U.S. Food and Drug Administration (2007). Anthrax Vaccines: Bridging Correlates of
Protection in Animals to Immunogenicity in Humans. Hilton Washington North,
Gaithersburg, Maryland.
Van Cuyk, S., A. Deshpande, A. Hollander, D. O. Franco, N. P. Teclemariam, J. A. Layshock, L.
O. Ticknor, M. J. Brown and K. M. Omberg (2012). Transport of Bacillus thuringiensis
var. kurstaki from an Outdoor Release into Buildings: Pathways of Infiltration and a
Rapid Method to Identify Contaminated Buildings. Biosecurity and Bioterrorism:
Biodefense Strategy, Practice, and Science 10(2): 215-227.
Van Cuyk, S., L. A. B. Veal, B. Simpson and K. M. Omberg (2011). Transport of Bacillus
thuringiensis var. kurstaki via Fomites. Biosecurity and Bioterrorism: Biodefense
Strategy, Practice, and Science 9(3): 288-300.
Vasconcelos, D., R. Barnewall, M. Babin, R. Hunt, J. Estep, C. Nielsen, R. Carnes and J. Carney
(2003). Pathology of Inhalation Anthrax in Cynomolgus Monkeys (Macaca fascicularis).
Laboratory Investigation 83(8): 1201-1209.
Vaswami, R. (1955). Anthrax-Agricultural and Industrial Infection, Treatment, and Prevention.
Industrial Medicine and Surgery 24(9): 408-412.
145
-------
Walker, J. S., F. Klein, R. E. Lincoln and A. L. Fernelius (1967). A Unique Defense Mechanism
against Anthrax Demonstrated in Dwarf Swine. Journal of Bacteriology 93(6): 2031-
2032.
Walsh, J. J., N. Pesik, C. P. Quinn, V. Urdaneta, C. A. Dykewicz, A. E. Boyer, J. Guarner, P.
Wilkins, K. J. Norville, J. R. Barr, S. R. Zaki, J. B. Patel, S. P. Reagan, J. L. Pirkle, T. A.
Treadwell, N. R. Messonnier, L. D. Rotz, R. F. Meyer and D. S. Stephens (2007). A Case
of Naturally Acquired Inhalation Anthrax: Clinical Care and Analyses of Anti-Protective
Antigen Immunoglobulin G and Lethal Factor. Clinical Infectious Diseases 44(7): 968-
971.
Watson, A. and D. Keir (1994). Information on Which to Base Assessments of Risk from
Environments Contaminated with Anthrax Spores. Epidemiology and Infection 113(3):
479-490.
Webb, G. F. and M. J. Blaser (2002). Mailborne Transmission of Anthrax: Modeling and
Implications. Proceedings of the National Academy of Sciences of the United States of
America 99(10): 7027-7032.
Weber, D. J. and W. A. Rutala (2001). Risks and Prevention of Nosocomial Transmission of
Rare Zoonotic Diseases. Clinical Infectious Diseases 32(3): 446-456.
Wein, L. M. and D. L. Craft (2005). Evaluation of Public Health Interventions for Anthrax: A
Report to the Secretary's Council on Public Health Preparedness. Biosecurity and
Bioterrorism: Biodefense Strategy, Practice, and Science 3(4): 348-356.
Wein, L. M., D. L. Craft and E. H. Kaplan (2003). Emergency Response to an Anthrax Attack.
Proceedings of the National Academy of Sciences of the United States of America
100(7): 4346-4351.
Weiner, Z. P. and I. J. Glomski (2012). Updating Perspectives on the Initiation of Bacillus
anthracis Growth and Dissemination through Its Host. Infection and Immunity 80(5):
1626-1633.
Weir, M. H. and C. N. Haas (2011). A Model for in-Vivo Delivered Dose Estimation for Inhaled
Bacillus anthracis Spores in Humans with Interspecies Extrapolation. Environmental
Science and Technology 45(13): 5828-5833.
146
-------
Weis, C. P., A. J. Intrepido, A. K. Miller, P. G. Cowin, M. A. Durno, J. S. Gebhardt and R. Bull
(2002). Secondary Aerosolization of Viable Bacillus anthracis Spores in Contaminated
US Senate Office. Journal of the American Medical Association 288(22): 2853-2858.
Wilkening, D. A. (2006). Sverdlovsk Revisited: Modeling Human Inhalation Anthrax.
Proceedings of the National Academy of Sciences of the United States of America
103(20): 7589-7594.
World Health Organization (2008). Anthrax in Humans and Animals (Fourth Edition). Geneva,
Switzerland: P. Turnbull, World Health Organization.
Xie, T., C. Sun, K. Uslu, R. D. Auth, H. Fang, W. Ouyang and D. M. Frucht (2013). A New
Murine Model for Gastrointestinal Anthrax Infection. PLoS One 8(6): e66943.
Yee, S. B., J. M. Hatkin, D. N. Dyer, S. A. Orr and M. L. M. Pitt (2010). Aerosolized Bacillus
anthracis Infection in New Zealand White Rabbits: Natural History and Intravenous
Levofloxacin Treatment. Comparative Medicine 60(6): 461-468.
Young Jr., G. A., M. R. Zelle and R. E. Lincoln (1946). Respiratory Pathogenicity of Bacillus
anthracis Spores I. Methods of Study and Observations on Pathogenesis. . Journal of
Infectious Diseases 79(3): 233-246.
Zaucha, G. M., M. L. M. Pitt, J. Estep, B. E. Ivins and A. M. Friedlander (1998). The Pathology
of Experimental Anthrax in Rabbits Exposed by Inhalation and Subcutaneous
Inoculation. Archives of Pathology and Laboratory Medicine 122(11): 982-992.
147
-------
Disclaimer: This text is a draft that has not been reviewed for technical accuracy or
adherence to EPA policy; do not quote or cite.
Report: Review of Bacillus anthracis Dose-Response Data for Human Health Risk
Assessment
(Rev. 3 Draft, October 2015)
Appendices
Appendix A - Transmission and Pathogenesis Considerations for Biological Threat Agents
Appendix B - Historical Approaches to Microbial Dose-Response Relationship Development for
Bacillus anthracis
Appendix C - Data Summary Table for End-stage Inhalation Anthrax Pathology of the Human,
Nonhuman Primate, and Rabbit
Appendix D - Bacillus anthracis Dose-Response Data for the Rabbit Characterized as
Supportive Data or Additional Data
Appendix E - Bacillus anthracis Dose-Response Data for the Nonhuman Primate Characterized
as Supportive Data or Additional Data
Appendix F - Conducting Benchmark Dose Analysis for Microbial Pathogens
-------
Appendix A - Transmission and Pathogenesis Considerations for
Biological Threat Agents
Introduction
Interest in the development of microbial dose-response relationships for biological threat agents
(BTA[s]) is currently high (U.S. Department of Homeland Security and U.S. Environmental
Protection Agency, 2009). The BTAs are a group of microbial pathogens that are capable of
producing significant illness, death, or incapacitation in people or animals when they are released
in a manner to facilitate specific types of exposure. Numerous dose-response relationships for
individual BTAs have been published (Haas, 2002; Bartrand et al., 2008; Weir and Haas, 2009;
Tamrakar et al., 2011; Teske et al., 2011; Weir and Haas, 2011; Taft and Hines, 2012). However,
further progress is challenged by the lack of an overarching methodology for microbial dose-
response analysis or alternatively, a dose-response modeling methodology specifically developed
to facilitate progress for the BTA group.
Current microbial risk assessment protocols, frameworks, or other publications have identified
transmission and pathogenesis considerations recommended for inclusion in microbial risk
assessment and dose-response modeling (Haas et al., 1999a; International Life Sciences Institute
[ILSI], 2000; Food and Agriculture Organization and World Health Organization [FAO and
WHO], 2003; Parkin, 2008; Interagency Microbiological Risk Assessment Guideline
Workgroup, 2011; U.S. Environmental Protection Agency, 2014).
A-l
-------
The transmission and pathogenesis considerations that have been identified include:
• secondary transmission (Parkin, 2008; Interagency Microbiological Risk Assessment
Guideline Workgroup, 2011; U.S. Environmental Protection Agency, 2014),
• propagation of the pathogen in the host (Haas et al., 1999a; Interagency Microbiological
Risk Assessment Guideline Workgroup, 2011; U.S. Environmental Protection Agency,
2014),
• immunity and susceptibility of the exposed population (Parkin, 2008; Interagency
Microbiological Risk Assessment Guideline Workgroup, 2011; U.S. Environmental
Protection Agency, 2014),
• use of threshold versus non-threshold models (International Life Sciences Institute
[ILSI], 2000; Food and Agriculture Organization and World Health Organization [FAO
and WHO], 2003; Interagency Microbiological Risk Assessment Guideline Workgroup,
2011), and
• potential variation in virulence exhibited by individual strains, variants, or isolates
(International Life Sciences Institute [ILSI], 2000; Parkin, 2008; Interagency
Microbiological Risk Assessment Guideline Workgroup, 2011).
Addressing these transmission and pathogenesis considerations in a manner suitable for all
microbial pathogens and across potential end uses of the risk assessment outputs represents a
significant technical challenge. Microbial risk assessment frameworks, including Haas et al.
(1999b) and the International Life Sciences Institute [ILSI] (2000), have been available for 15
years. The difficulty in addressing these considerations may help to explain the relatively slow
A-2
-------
progress in the development of microbial dose-response methodologies even though microbial
risk frameworks have been available for over 10 years.
Microbial pathogens are recognized to exhibit significant diversity in transmission and
pathogenesis characteristics. However, little attention has been focused on the collective
identification of BTAs that was initially based on a unique assemblage of transmission and
pathogenesis characteristics. The characteristics provide an intentional aerosol release of agent to
drive atypical routes of transmission relative to natural disease transmission (i.e., inhalation) and
greater severity of outcomes than typical natural routes of exposure (Roy et al., 2010). More
recently, changes in the desired end use of BTA dose-response relationship data also introduces
unique elements into the microbial dose-response analysis for these pathogens relative to
traditional pathogens. For example, dose-response relationships for BTAs may be used in the
development of clearance goals after an intentional or accidental release (U.S. Department of
Homeland Security and U.S. Environmental Protection Agency, 2009). This development of
clearance goals could drive the need for low-dose evaluations either on the outer boundaries of
the exposure areas or in areas where remedial technologies have been applied and residual levels
may remain.
This appendix will evaluate a defined set of BTAs relative to the transmission and pathogenesis
considerations that have been identified and consider the relevance for microbial dose-response
modeling when using empirical or mechanistic modeling approaches.
Characteristics of Biological Threat Agents
The infectivity of BTAs can be characterized as an opportunistic airborne transmission
capability, with enhanced virulence resulting from the inhalation route of exposure when
A-3
-------
compared to typical routes of exposure associated with natural exposure (e.g., inhalational versus
cutaneous anthrax) (Roy et al., 2010). The BTA group exhibits a unique capability for
persistence when released as respirable microbial aerosols (Eitzen, 2007) and subsequent
infectivity from inhalation exposure relative to most microbial bacteria and viruses (Roy et al.,
2010). It is hypothesized that commonalities in the biological mechanisms allowing for aerosol
persistence and infectivity may also mediate similarities in the early interactions between host
and the pathogen. Interestingly, the macrophage or other phagocytic cells are associated with
initiation of infection and/or preferential replication sites for a number of bacterial BTAs (e.g.,
Bacillus anthracis (Inglesby et al., 2002); Burkholderia spp. (Whitlock et al., 2007);
Franscisella tularensis (Ketavarapu et al., 2008)). In natural environments, many of these same
bacterial BTAs also utilize an amoebic niche which may be a training ground for successful
invasion of host phagocytic cells. Interestingly, reliance on the amoebic niche is not unique to
BTAs and has been associated with traits that also facilitate the successful invasion of host
phagocytic cells, including the macrophage, by Legionella spp as described by Swanson and
Hammer (2000).
Evaluation of Transmission and Pathogenesis Considerations for Dose-
Response Modeling of Biological Threat Agents
The BTA group for the evaluation is the "traditional" BTAs identified in the U.S. Centers for
Disease Control and Prevention (CDC) historic Select Agent Category A and B lists (Rotz et al.,
2002). The BTAs were defined to include bacterial agents (i.e., B. anthracis, B. mallei, B.
pseudomallei, F. tularensis, Yersiniapestis, Coxiella burnetii) and viral agents (i.e., filovirus and
arenavirus hemorrhagic fever viruses, Variola major virus [smallpox virus]).
A-4
-------
The identification of modeling considerations necessary for microbial dose-response analysis
was based on fundamental elements of infectious disease transmission and illness, with the initial
focus on points of difference between chemical toxicity and microbial pathogenesis.
The following modeling considerations were evaluated for the group of BTAs that have been
identified:
• secondary transmission,
• propagation of the pathogen in the host,
• immunity and susceptibility of the exposed population,
• determination of threshold in response, and
• potential variation in virulence exhibited by individual strains, variants, or isolates.
Secondary Transmission
For many infectious diseases, transmission has been modeled as a dynamic process where
infected individuals become the source of pathogens to which others can be exposed (Eisenberg
et al., 2002), either directly or indirectly. Secondary transmission has been defined in various
ways in the literature; this paper defines direct secondary transmission as the communicability,
or transmission, of disease directly from a primary to secondary case. Direct secondary
transmission can occur from person-to-person airborne transmission or direct contact with
infectious bodily fluids. Indirect secondary transmission is defined as the transmission of disease
through indirect means following a human-environment-human pathway, as occurs when contact
with a fomite contaminated by the primary case transmits infection to a secondary case(s) (U.S.
Environmental Protection Agency, 2007).
A-5
-------
For pathogens that exhibit secondary transmission, a population-based microbial dose-response
estimate based solely on the first transmission of disease can be biased low relative to the actual
response due to the potential "multiplier" effect of initial cases not explicitly included in the
model (i.e., successive cases that originate from transmission of the first case). This multiplier
effect has led to the assertion that infectious disease risk is appropriately assessed as a
population-based risk using a dynamic process for these pathogens (Eisenberg et al., 2002).
Dynamic models contrast with the use of static modeling approaches such as the empirical dose-
response models that are commonly used for chemical dose-response analysis.
However, most BTAs in this evaluation do not exhibit direct secondary transmission. Traditional
BTAs were preferentially selected to minimize the potential for direct person-to-person spread to
allow for containment of the disease spread by those releasing the agents (Eitzen, 2007).
Differences exist in the communicability of the bacterial and viral BTAs that have been
identified. A number of viral BTAs are considered communicable: the smallpox virus
(Henderson et al., 1999) and hemorrhagic fever viruses (e.g., arenaviruses, filoviruses, Lassa
viruses) (Borio et al., 2002). With the exception of the communicable pneumonic form of Y.
pestis (Inglesby et al., 2000), the remaining bacterial BTAs are noncommunicable or rarely
communicable.
In summary, the following BTAs are identified as (1) noncommunicable: B. anthracis (Inglesby
et al., 2002), F. tularensis (Dennis et al., 2001), and C. burnetii (Azad, 2007), or (2) rarely
communicable by humans: B. mallei (Whitlock et al., 2007) and B. pseudomallei (Cheng and
Currie, 2005). Some viral hemorrhagic fevers have been described as communicable
"predominantly" by physical contact with bodily fluids, and there is less compelling evidence
A-6
-------
that person-to-person airborne transmission has occurred for others absent contact (e.g.,
filoviruses) (Borio et al., 2002). However, it is recommended that new literature from the 2014
Ebola outbreak continue to be evaluated to ensure current data are incorporated into assumptions
regarding this pathogen, especially for the potential for person-to-person transmission absent
intense and/or aerosol exposure to contaminated bodily fluids.
Fomites are the primary concern for indirect secondary transmission of illness. However,
contamination from bioaerosols produced by infected individuals is constrained by concentration
limits imposed by the natural disease process (Roy et al., 2010), and the chain of transmission is
fairly limited for bacterial BTAs that are not communicable in their natural disease process. Most
traditional BTAs are zoonotic pathogens where humans are not the primary infectious target (i.e.,
humans as an incidental or dead-end host) (Eitzen, 2007).
Therefore, human illness may result from the high exposure concentration associated with the
intentional or accidental release of BTAs, but the potential for secondary transmission potential
then returns to the potentially normally exhibited during natural infections. There may be
variability in indirect secondary transmission among the viral BTAs. Indirect secondary
transmission has been documented for the smallpox virus; this includes transmission from books
as reported by Ferson (2001) and letters as identified by Ambrose (2005). During an Ebola
outbreak in 2000, there was limited evidence of secondary transmission and a lack of measurable
contamination on common fomite surfaces tested in a hospital setting during the 2000 Ebola
virus outbreak (Bausch et al., 2007).
For BTAs identified as noncommunicable or rarely communicable, traditional static dose-
response mathematical models are appropriate. Some viral BTAs identified as potentially
A-7
-------
communicable may require a fairly significant level of contact with infected individuals (e.g.,
intimate contact [Bausch et al., 2007]) or bodily fluids (e.g., blood, vomit in health care settings
[Bausch et al., 2007]) to produce transmission. Further evaluation of the applicability of assumed
secondary transmission may be appropriate for these viral BTAs, especially if infectivity
endpoints are used to derive the original dose-response estimates.
Pathogenic Propagation in the Host
The propagation of pathogens in the host is a key process in disease pathogenesis and can signal
the transition from infection to illness for some pathogens. As a differentiator between chemical
and microbial risk assessment, the multiplication of the pathogen is noted as a distinct
characteristic of microbial risk assessment as toxicants are not assumed to increase in
concentration or reproduce (U.S. Environmental Protection Agency, 2014). Pathogenic
propagation for microbial dose-response analysis may confound the relationship between the
exposure dose and response due to multiplication of pathogens in the host. The multiplication of
pathogens could result in a higher exposure dose to the target tissue associated with illness than
if no multiplication took place. Chemicals may form toxic metabolites and the metabolites
responsible for toxicity may increase in concentration over time. However, toxic metabolite
formation can be predicted from the chemical dose when kinetic relationships between the
chemical, enzyme, and metabolite are known.
A complication in the assessment of microbial dose-response relationships is the recognition that
larger doses of pathogens are not always associated with a higher probability of response or
severity of illness (U.S. Department of Homeland Security and U.S. Environmental Protection
A-8
-------
Agency, 2009). Dose-dependency in incubation periods has been preliminarily identified for
some microbial pathogens, including BTAs (e.g., B. anthracis [Wilkening, 2006]).
A-9
-------
Immunity and Susceptibility in Population
Immunity and susceptibility result from host characteristics that affect the host-pathogen
interaction. Susceptibility, inclusive of all host-related contributors1 to variability in response, is
defined as "the extent to which a host is vulnerable to infection, taking into account a host's
intrinsic and/or acquired traits that influence infection" (U.S. Environmental Protection Agency,
2007). Immunity results from immunization, previous exposure, or other host-related
characteristics and can provide partial or complete protection from exposure (U.S.
Environmental Protection Agency, 2007). Variability in susceptibility can modify response
through prevention of infection or illness or enhancement of susceptibility due to variation or
suboptimal functioning of the immune system. Susceptibility may also include variation in
response to toxins produced by pathogens that are toxico-infectious. For example, variation in
response to anthrax toxin has been identified for B. anthracis (Inglesby et al., 2002).
Variation in susceptibility has been identified as a critical element in the modeling of microbial
dose-response relationships (Food and Agriculture Organization and World Health Organization
[FAO and WHO], 2003; Interagency Microbiological Risk Assessment Guideline Workgroup,
2011; U.S. Environmental Protection Agency, 2014). Susceptibility considerations in dose-
response analysis are important to ensure that dose-response modeling allows for evaluation of
interindividual variability in response, including potentially sensitive subpopulations (e.g., health
compromises) or life stages (e.g., elderly). An additional concern for the modeling of infectious
disease is the potential for transmission to result from the interaction of susceptible and infected
1 However, susceptibility as defined in this paper does not extend to differential exposure as contributing to response
variation (e.g., Section 3.5.1 of U.S. Environmental Protection Agency [2004]).
A-10
-------
individuals. Susceptibility differences resulting from immunity can be exhibited as individuals
shift from susceptible to immune after transmission of illness, with the result that dynamic dose-
response modeling approaches may become necessary (Eisenberg et al., 2002).
Susceptibility in infection and/or illness is known to vary across populations for microbial
pathogens (e.g., Cryptosporidium sp. in Teunis et al. [2002], Norwalk virus in Teunis et al.
[2008]). There are preliminary indications that infectivity and illness exhibit greater variability
than the variability described for chemicals when compared on an absolute scale (Hattis, 1997).
General factors such as age, immune status, or co-existing health conditions have been identified
as contributing to susceptibility differences (Teunis et al., 2002). While data are emerging on
potential associations of genetic variation and modified susceptibility for some well-studied
pathogens (e.g., allelic variation and associated tuberculosis susceptibility across Canadian
Aboriginal populations in Larcombe et al. [2008]), the mechanistic incorporation into a dose-
response model has not been described. Susceptibility may also be expressed in a dose-
dependent manner whereby pathogens act as frank pathogens at higher doses but opportunistic
pathogens in more susceptible populations at lower doses (e.g., 2001 Connecticut anthrax case as
evaluated by Cohen and Whalen [2007]). Additionally, population variation in the sensitivity at
the cellular level to pathogenic toxins (e.g., anthrax toxin in Martchenko et al. [2012]) has also
been demonstrated in recently published in-vitro studies. However, there are critical knowledge
gaps for mechanistic process and associated quantitative data that limit the current capability to
model variation in susceptibility.
Historically, the selection of BTAs incorporated a preference for pathogens for which the
targeted population exhibits a lack of immunity (Fothergill, 1960; Eitzen, 2007) and presents
A-ll
-------
uniformity in susceptibility. There has also been a desire for use of BTAs that have vaccines
available, but where the population is not routinely vaccinated (Eitzen, 2007). For these reasons,
BTAs can be modeled without the assumption of immunity. However, BTAs are not unique
among the microbial pathogens in the potential for the host to exhibit variation in susceptibility.
While it can be hypothesized that the variation in susceptibility may be limited in expression at
higher dose levels, the evaluation of low level dose-response relationships will benefit from
consideration of the susceptibility differences in individuals. Potential approaches to evaluate
variation in susceptibility include development of data from animal models selected for their
ability to mimic susceptible subpopulation conditions (e.g., disease, age, immunosuppressant
drugs). General approaches to apply uncertainty factors to account for this variability have been
suggested for microbial dose-response analysis (U.S. Environmental Protection Agency, 2008,
2010), but have yet to be described and published. Modeling to include variation in response has
utilized tolerance-based dose-response models (e.g., probit slope in Wein et al. [2003]) where all
contributions to variation in response are mathematically aggregated into one normally
distributed value.
Threshold Versus Non-Threshold Models
A threshold model incorporates the assumption that there is a "dose or exposure below which no
deleterious effect is expected to occur" (U.S. Environmental Protection Agency, 2011). A non-
threshold model assumes that, even with the dose of one microorganism, there is a small nonzero
probability of infection and subsequent illness (Food and Agriculture Organization and World
Health Organization [FAO and WHO], 2003). From a practical perspective, the presence of a
pathogenic threshold cannot be determined experimentally or empirically (Food and Agriculture
A-12
-------
Organization and World Health Organization [FAO and WHO], 2003). It has been suggested that
nonthreshold mathematical models should be preferentially evaluated, but these models should
have sufficient inherent flexibility to allow high or low curvature at low doses allowing for the
mimicking of a "threshold-like" or sublinear response (Food and Agriculture Organization and
World Health Organization [FAO and WHO], 2003). However, a full range of models (e.g.,
threshold, non-threshold) should be considered to avoid the uncertainty introduced with selection
of one specific model assumption (Coleman and Marks, 2000).
Potential Strain, Allelic, or Variant Differences in Virulence
Virulence is defined as "the degree of intensity of the disease produced by a microorganism as
indicated by its ability to invade the tissues of the host and the ensuing severity of illness"
(International Life Sciences Institute [ILSI], 2000). Strain, allelic, or variant differences in
virulence for BTAs are relevant because of the potential for a mismatch between the virulence of
the BTA for which the dose-response relationship was derived versus the virulence of the BTA
to which the relationship is applied.
High variability in strain virulence has been described for common bacterial pathogens,
including Salmonella sp. (Coleman et al., 2004) and Campylobacter jejuni (Coleman et al., 2004)
and animal studies for BTAs, including B. anthracis (Fellows et al., 2001). Pathogenic virulence
can also be modified, either decreased or increased, in response to passage through multiple
A-13
-------
hosts (Roy et al., 2010). However, quantification of the variation in virulence is not well
characterized.2
BTAs do not differ from the larger group of pathogens with regard to this consideration.
However, there has been a preference for BTA selection based on a demonstration of greatest
virulence (Eitzen, 2007), whether the endpoint is lethality (e.g., inhalation anthrax) or
incapacitation (e.g., Q fever). If the concern regarding the exhibited variability is primarily
related to the possibility of underestimating virulence as part of the dose-response process, one
approach could include modeled strains with the presumed greatest virulence (i.e., a dose-
response equivalent of the Kuhn et al. [2011] approach).
Summary of Modeling Considerations for Biological Threat Agents
The lack of secondary transmission exhibited by bacterial BTAs and some viral BTAs allows for
the use of static dose-response models for these microbial pathogens (Table A-l). For the
remaining modeling process considerations, each can be addressed to varying degrees within
currently available dose-response models. While the remaining considerations can be properly
viewed as mechanistic, approaches are available to include these elements as part of empirical or
mechanistic models. Processes can be defined that allow for a modification of the dose-response
outputs (e.g., uncertainty factor, data-derived extrapolation factor) of empirical, nominally
mechanistic, or limited mechanistic models. Likewise, considerations can also be explicitly
modeled in increasingly mechanistic models as data are available. These considerations involve
content areas for which there is acknowledged high uncertainty and very limited data, as well as
2 Product formulation and associated practices may also affect the virulence. Further information on a Bayesian
assessment conducted for the guinea pig is available in Mitchell-Blackwood et al. (2012)
A-14
-------
the potential for extremes in variability to be exhibited. Chemical dose-response modelers
struggled with similar data and methodological challenges (e.g., interindividual variability in
susceptibility), and the chemical dose-response approaches may be leveraged for addressing data
gaps, variability, and uncertainty.
Table A-l. Summary of Transmission and Pathogenesis Considerations and Relevance for
Modeling
Transmission and
Pathogenesis
Consideration
Universal for Microbial
Pathogens or Limited
Relevance for BTAs
Mechanistic Modeling
Consideration and
Potential Means to
Address in Dose-
Response Modeling
Potential Means to Address in
Dose-Response Modeling in
Empirical Modeling
Immunity and
Susceptibility in
Population
Immunity not relevant for
BTAs, noting limited
immunity as defining
characteristic of BTAs
Yes, consider modeling
element mechanistically as
interindividual variability
in susceptibility
Address as animal model or
human dose-response input data
decision, use of data derived
extrapolation factor or uncertainty
factor for adjustment after
development of dose-response
relationship
Variation in susceptibility
universal for microbial
pathogens
Secondary
Transmission
Not relevant for bacterial
BTAs; relevant for some
viral BTAs
Yes for viral BTAs,
dynamic dose-response
models or multiplier
adjustment to static
estimate of response to
reflect additional
transmission may be
potential means to address
Dynamic dose-response models or
multiplier adjustment to static
estimate of dose-response
relationship to reflect additional
transmission may be potential
means to address
Pathogen
Propagation
Universal for microbial
pathogens
Yes, incorporate bacterial
kinetics of identified
compartment or other target
tissues
Not an element of an empirical
model
Strain, Allelic, or
Variant Differences
in Virulence
Universal for microbial
pathogens
Possibly, as more data are
available may be able to
mechanistically link
identified virulence
differences with known
elements of strains, alleles,
or variants
Differences in virulence may be
addressed by selection of target
strain, allele, or variant for dose-
response data set
Threshold or
Nonthreshold
Determination
Universal for microbial
pathogens
No, structure of
mathematical model pre-
determines whether
threshold or non-threshold
is modeled
Consider evaluation of
mathematical models that vary in
the assumption of threshold to
address uncertainty resulting from
model selection
Table A-2 identifies mathematical dose-response models and published examples of BTA dose-
response relationships using the identified model. Existing dose-response models can be used or
A-15
-------
new models developed. It is recommended that a variety of dose-response models be evaluated
with regard to incorporation of mechanistic elements and the presence or absence of a threshold.
A-16
-------
Table A-2. Mathematical Dose-Response Models by Type of Model and Referencing
Publication
Type of Model
Examples of Mathematical Dose-
Response Model
Published BTA Dose-Response
Relationship Using Model
Empirical
Probit or Log Probit
Taft and Hines (2012)
Logistic or Log Logistic
Weibull
Dichotomous Hill
Gamma
Nominally Mechanistic
Exponential
Bartrand et al. (2008)
Teske et al. (2011)
Taft and Hines (2012)
Beta Poisson
Limited Mechanistic
Competing Risk Model
Gutting et al. (2008)
Time-Dose-Response Model
Huang and Haas (2009)
In-vivo Growth Model
Huang and Haas (2011)
Time-Dependent Dose-Response Model
Mayer etal. (2011)
Cumulative Dose Model
Pujol et al. (2009)
In-vivo Delivered Dose Model
Weir and Haas (2011)
Age -Dose-Response Model
Weir and Haas (2009)
Comprehensive
Mechanistic
None to Date
None to Date
Applicability to Microbial Pathogens Other than Biological Threat Agents
There is wide potential applicability of this microbial dose-response methodology for microbial
pathogens other than BTAs. The following methodology is most appropriate for non-BTA
pathogens that do not exhibit secondary transmission and exhibit initiation of infection through
the inhalation route of exposure.
However, it is important to recognize that this evaluation may be appropriately applied to
pathogens described as BTAs (e.g., emerging BTAs) that do not exhibit pathogenesis and
transmission characteristics similar to those described for the traditional BTAs. The traditional
BTAs were selected from the larger universe of pathogens based on the recognized potential
infectivity for a large number of individuals when released and maximization of lethality or
A-17
-------
incapacitation to those exposed. While pathogens identified to be of greatest concern for
bioterrorism have typically been assumed to be the same as the traditional BTAs, the goal to
maximize adverse health effects may be overtaken by the achievement of other ends (e.g., salad
bar tampering with Salmonella in Oregon to disrupt elections [Torok et al., 1997]). For these
emerging BTAs, unusual exposure scenarios identified uniquely for bioterrorism (versus
traditional BTAs and the inhalation route of exposure) warrant further review before routine
applicability of this methodology.
A-18
-------
Bibliography
Ambrose, C. T. (2005). Osier and the Infected Letter. Emerging Infectious Diseases 11(5): 689-
693.
Azad, A. F. (2007). Pathogenic Rickettsiae as Bioterrorism Agents. Clinical Infectious Diseases
45 (Supplement 1): S52-S55.
Bartrand, T. A., M. H. Weir and C. N. Haas (2008). Dose-Response Models for Inhalation of
Bacillus anthracis Spores: Interspecies Comparisons. Risk Analysis 28(4): 1115-1124.
Bausch, D. G., J. S. Towner, S. F. Dowell, F. Kaducu, M. Lukwiya, A. Sanchez, S. T. Nichol, T.
G. Ksiazek and P. E. Rollin (2007). Assessment of the Risk of Ebola Virus Transmission
from Bodily Fluids and Fomites. The Journal of Infectious Diseases 196 (Supplement 2):
S142-S147.
Borio, L., T. Inglesby, C. J. Peters, A. L. Schmaljohn, J. M. Hughes, P. B. Jahrling, T. Ksiaszek,
K. M. Johnson, A. Meyerhoff, T. O'Toole, M. S. Ascher, J. Bartlett, J. G. Breman, E. M.
Eitzen, Jr., M. Hamberg, J. Hauer, D. A. Henderson, R. T. Johnson, G. Kwik, M. Layton,
S. Lillibridge, G. J. Nabel, M. T. Osterholm, T. M. Perl, P. Russell, K. Tonat for the
Working Group on Civilian Biodefense (2002). Hemorrhagic Fever Viruses: Medical and
Public Health Management. Journal of the American Medical Association 287(18):
2391-2405.
Cheng, A. C. and B. J. Currie (2005). Melioidosis: Epidemiology, Pathophysiology, and
Management. Clinical Microbiology Reviews 18(2): 383-416.
Cohen, M. L. and T. Whalen (2007). Implications of Low Level Human Exposure to Respirable
B. anthracis. Applied Biosafety 12(2): 109-115.
Coleman, M. E. and H. M. Marks (2000). Mechanistic Modeling of Salmonellosis. Quantitative
Microbiology 2(3): 227-247.
Coleman, M. E., H. M. Marks, N. J. Golden and H. K. Latimer (2004). Discerning Strain Effects
in Microbial Dose-Response Data. Journal of Toxicology and Environment Health, Part
A 67(8-10): 667-685.
A-19
-------
Dennis, D. T., T. V. Inglesby, D. A. Henderson, J. G. Bartlett, M. S. Ascher, E. Eitzen, A. D.
Fine, A. M. Friedlander, J. Hauer, M. Layton, S. R. Lillibridge, J. E. McDade, M. T.
Osterholm, T. O'Toole, G. Parker, T. M. Perl, P. K. Russell, K. Tonat for the Working
Group on Civilian Biodefense (2001). Tularemia as a Biological Weapon: Medical and
Public Health Management. Journal of the American Medical Association 285(21): 2763-
2773.
Eisenberg, J. N. S., M. A. Brookhart, G. Rice, M. Brown and J. M. Colford, Jr. (2002). Disease
Transmission Models for Public Health Decision Making: Analysis of Epidemic and
Endemic Conditions Caused by Waterborne Pathogens. Environmental Health
Perspectives 110(8): 783-790.
Eitzen, E. M., Jr. (2007). Chapter 20: Use of Biological Weapons. Medical Aspects of Biological
Warfare. Z. F. Dembek. Falls Church, VA and Washington, DC, Office of the Surgeon
General and Borden Institute: 437-450.
Fellows, P. F., M. K. Linscott, B. E. Ivins, M. L. M. Pitt, C. A. Rossi, P. H. Gibbs and A. M.
Friedlander (2001). Efficacy of a Human Anthrax Vaccine in Guinea Pigs, Rabbits, and
Rhesus Macaques against Challenge by Bacillus anthracis Isolates of Diverse
Geographical Origin. Vaccine 19(23-24): 3241-3247.
Ferson, M. J. (2001). Books as Carriers of Disease. The Medical Journal of Australia. 175(11):
663-664.
Food and Agriculture Organization and World Health Organization (FAO and WHO) (2003).
Hazard Characterization for Pathogens in Food and Water, Guidelines .W orld Health
Organization: Rome, Italy and Geneva, Switzerland. Microbiological Risk Assessment
Series No. 3.
Fothergill, L. D. (1960). Chapter 4. The Biological Warfare Threat. Nonmilitary Defense:
Chemcial and Biological Defenses in Perspective. Advances in Chemistry Series.
Volume 26. pp.23-33.
Gutting, B. W., S. R. Channel, A. E. Berger, J. M. Gearhart, G. A. Andrews, R. L. Sherwood and
T. L. Nichols (2008). Mathematically Modeling Inhalational Anthrax. Microbe 3(2): 78-
85.
Haas, C. N. (2002). On the Risk of Mortality to Primates Exposed to Anthrax Spores. Risk
Analysis 22(2): 189-193.
A-20
-------
Haas, C. N., J. B. Rose and C. P. Gerba (1999a). Chapter 7. Conducting the Dose-Response
Assessment. Quantitative Microbial Risk Assessment, (pp. 260-319). New York: John
Wiley & Sons, Inc.
Haas, C. N., J. B. Rose and C. P. Gerba (1999b). Quantitative Microbial Risk Assessment. New
York: John Wiley & Sons, Inc.
Hattis, D. (1997). Human Variability in Susceptibility: How Big, How Often, for What
Responses to What Agents? Environmental Toxicology and Pharmacology 4(3-4): 195-
208.
Henderson, D. A., T. V. Inglesby, J. G. Bartlett, M. S. Ascher, E. Eitzen, P. B. Jahrling, J. Hauer,
M. Layton, J. McDade, M. T. Osterholm, T. O'Toole, G. Parker, T. Perl, P. K. Russell, K.
Tonat for the Working Group on Civilian Biodefense (1999). Smallpox as a Biological
Weapon: Medical and Public Health Management. Journal of the American Medical
Association 281(22): 2127-2137.
Huang, Y. and C. N. Haas (2009). Time-Dose-Response Models for Microbial Risk Assessment.
Risk Analysis 29(5): 648-661.
Huang, Y. and C. N. Haas (2011). Quantification of the Relationship between Bacterial Kinetics
and Host Response for Monkeys Exposed to Aerosolized Francisella tularensis. Applied
and Environmental Microbiology 77(2): 485-490.
Inglesby, T. V., D. T. Dennis, D. A. Henderson, J. G. Bartlett, M. S. Ascher, E. Eitzen, A. D.
Fine, A. M. Friedlander, J. Hauer, J. F. Koerner, M. Layton, J. E. McDade, M. T.
Osterholm, T. O'Toole, G. Parker, T. M. Perl, P. K. Russell, M. Schoch-Spana, K. Tonat
for the Working Group on Civilian Biodefense (2000). Plague as a Biological Weapon:
Medical and Public Health Management. Journal of the American Medical Association
283(17): 2281-2290.
Inglesby, T. V., T. O'Toole, D. A. Henderson, J. G. Bartlett, M. S. Ascher, E. Eitzen, A. M.
Friedlander, J. Gerberding, J. Hauer, J. M. Hughes, J. McDade, M. T. Osterholm, G.
Parker, T. M. Perl, P. K. Russell, K. Tonat for the Working Group on Civilian Biodefense
(2002). Anthrax as a Biological Weapon, 2002. Updated Recommendations for
Management. Journal of the American Medical Association 287(17): 2236-2252.
A-21
-------
Interagency Microbiological Risk Assessment Guideline Workgroup (2011). Microbial Risk
Assessment Guidance: Pathogenic Microorganisms with Focus on Food and Water
(Draft, Version 5.3).
International Life Sciences Institute (ILSI) (2000). Revised Framework for Microbial Risk
Assessment. Washington, D.C: ILSI Press.
Ketavarapu, J. M., A. R. Rodriguez, J.-J. Yu, Y. Cong, A. K. Murthy, T. G. Forsthuber, M. N.
Guentzel, K. E. Klose, M. T. Berton and B. P. Arulanandam (2008). Mast Cells Inhibit
Intramacrophage Francisella tularensis Replication via Contact and Secreted Products
Including H-4. Proceedings of the National Academy of Sciences of the United States of
America 105(27): 9313-9318.
Kuhn, J. H., L. E. Dodd, V. Wahl-Jensen, S. Radoshitzky, S. Bavari and P. B. Jahrling (2011).
Evaluation of Perceived Threat Differences by Filovirus Variants. Biosecurity and
Bioterrorism: Biodefense Strategy, Practice, and Science 9(4): 361-371.
Larcombe, L. A., P. H. Orr, A. M. Lodge, J. S. Brown, I. J. Dembinski, L. C. Milligan, E. A.
Larcombe, B. D. Martin and P. W. Nickerson (2008). Functional Gene Polymorphisms in
Canadian Aboriginal Populations with High Rates of Tuberculosis. The Journal of
Infectious Diseases 198(8): 1175-1179.
Martchenko, M., S. I. Candille, H. Tang and S. N. Cohen (2012). Human Genetic Variation
Altering Anthrax Toxin Sensitivity. Proceedings of the National Academy of Sciences of
the United States of America 109(8): 2972-2977.
Mayer, B. T., J. S. Koopman, E. L. Ionides, J. M. Pujol and J. N. S. Eisenberg (2011). A
Dynamic Dose-Response Model to Account for Exposure Patterns in Risk Assessment: A
Case Study in Inhalation Anthrax. Journal of the Royal Society Interface 8(57): 506-517.
Mitchell-Blackwood, J., P. L. Gurian, R. Lee and B. Thran (2012). Variance in Bacillus
anthracis Virulence Assessed through Bayesian Hierarchical Dose-Response Modelling.
Journal of Applied Microbiology 113(2): 265-275.
Parkin, R. T. (2008). Foundations and Frameworks for Human Microbial Risk Assessment,
Center for Risk Science and Public Health, School of Public Health and Health Services,
The George Washington University Medical Center. Washington, D.C.
A-22
-------
Pujol, J. M., J. E. Eisenberg, C. N. Haas and J. S. Koopman (2009). The Effect of Ongoing
Exposure Dynamics in Dose Response Relationships. PLoS Computational Biology 5(6):
el000399.
Rotz, L. D., A. S. Khan, S. R. Lillibridge, S. M. Ostroff and J. M. Hughes (2002). Public Health
Assessment of Potential Biological Terrorism Agents. Emerging Infectious Diseases 8(2):
225-230.
Roy, C. J., D. S. Reed and J. A. Hutt (2010). Aerobiology and Inhalation Exposure to Biological
Select Agents and Toxins. Veterinary Pathology 47(5): 779-789.
Swanson, M. S. and B. K. Hammer (2000). Legionella pneumophila Pathogenesis: A Fateful
Journey from Amoebae to Macrophages. Annual Reviews in Microbiology 54: 567-613.
Taft, S. C. and S. A. Hines (2012). Benchmark Dose Analysis for Bacillus anthracis Inhalation
Exposures in the Nonhuman Primate. Risk Analysis 32(10): 1750-1768.
Tamrakar, S. B., A. Haluska, C. N. Haas and T. A. Bartrand (2011). Dose-Response Model of
Coxiella burnetii (Q Fever). Risk Analysis 31(1): 120-128.
Teske, S. S., Y. Huang, S. B. Tamrakar, T. A. Bartrand, M. H. Weir and C. N. Haas (2011).
Animal and Human Dose-Response Models for Brucella Species. Risk Analysis 31(10):
1576-1596.
Teunis, P. F. M., C. L. Chappell and P. C. Okhuysen (2002). Cryptosporidium Dose-Response
Studies: Variation between Hosts. Risk Analysis 22(3): 475-485.
Teunis, P. F. M., C. L. Moe, P. Liu, S. E. Miller, L. Lindesmith, R. S. Baric, J. Le Pendu and R.
L. Caleron (2008). Norwalk Virus: How Infectious Is It? Journal of Medical Virology
80(8): 1468-1476.
Torok, T. J., R. V. Tauxe, R. P. Wise, J. R. Livengood, R. Sokolow, S. Mauvais, K. A. Birkness,
M. R. Skeels, J. M. Horan and L. R. Foster (1997). A Large Community Outbreak of
Salmonellosis Caused by Intentional Contamination of Restaurant Salad Bars. Journal of
the American Medical Association 278(5): 389-395.
U.S. Department of Homeland Security and U.S. Environmental Protection Agency (2009).
Planning Guidance for Recovery Following Biological Incidents (Draft). Washington,
A-23
-------
DC: Biological Decontamination Standards Working Group, National Science and
Technology Council. U.S. Department of Homeland Security.
U.S. Environmental Protection Agency (2004). An Examination of EPA Risk Assessment
Principles and Practices. Washington, DC: Office of the Science Adviser, Risk
Assessment Forum. EPA/100/B-04/001.
U.S. Environmental Protection Agency. (2007). "Thesaurus of Terms Used in Microbial Risk
Assessment." from http://www.epa.gov/waterscience/criteria/humanhealth/microbial/.
U.S. Environmental Protection Agency (2008). Summary Report of the U.S. EPA Workshop on
Challenges to Integrating Immunotoxicological and Microbial Risk Assessment for
Susceptible Populations and Life Stages: Washington, DC
U.S. Environmental Protection Agency (2010). Workshop Report: State-of-the-Science for the
Determination and Application of Dose-Response Relationships in Microbial Risk
Assessment. Washington, DC: Office of Research and Development, National Homeland
Security Research Center. EPA/600/R-10/022.
U.S. Environmental Protection Agency. (2011). "IRIS Glossary." from
http://www.epa.gov/ncea/iris/index.html.
U.S. Environmental Protection Agency (2014). Microbiological Risk Assessment (MRA) Tools,
Methods, and Approaches for Water Media. Washington DC: Office of Science and
Technology, Office of Water. EPA-820-R-14-009.
Wein, L. M., D. L. Craft and E. H. Kaplan (2003). Emergency Response to an Anthrax Attack.
Proceedings of the National Academy of Sciences of the United States of America
100(7): 4346-4351.
Weir, M. H. and C. N. Haas (2009). Quantification of the Effects of Age on the Dose Response
of Variola major in Suckling Mice. Human and Ecological Risk Assessment 15(6): 1245-
1256.
Weir, M. H. and C. N. Haas (2011). A Model for in-Vivo Delivered Dose Estimation for Inhaled
Bacillus anthracis Spores in Humans with Interspecies Extrapolation. Environmental
Science and Technology 45(13): 5828-5833.
A-24
-------
Whitlock, G. C., D. M. Estes and A. G. Torres (2007). Glanders: Off to the Races with
Burkholderia mallei. FEMS Microbiology Letters 277(2): 115-122.
Wilkening, D. A. (2006). Sverdlovsk Revisited: Modeling Human Inhalation Anthrax.
Proceedings of the National Academy of Sciences of the United States of America
103(20): 7589-7594.
A-25
-------
Appendix B - Historical Approaches to Microbial Dose-Response
Relationship Development for Bacillus anthracis
Introduction
A long history of publications describing anthrax infection in man and livestock dates back to the
original publications by Koch and Pasteur first describing the Bacillus anthracis organism and
disease transmission in the late 1800s (Hilmas et al., 2009). Early descriptions focused on
disease pathogenesis and livestock vaccination strategies, with little research effort spent
describing relationships between dose and effect in humans. However, there was a significant
change in research focus when B. anthracis was evaluated as a potential bioweapon after World
Wars I and II. Since that time, a body of literature has developed to model and report dose-
response data of relevance to the human from intentional or accidental release of spores of B.
anthracis.
This review will consider historical approaches to model microbial dose-response relationships
for B. anthracis in the United States and United Kingdom starting at the end of World War II.
During the 1940s, open source publications first began to identify animal model data to define
lethality values for animal models or relative estimates of human susceptibility. Since that time,
there has been an evolution in the development of microbial dose-response data for B. anthracis.
Early research on B. anthracis focused on military applications to evaluate general potency or
support preliminary development of medical countermeasures that lead to early biological
models of disease pathogenesis. An apparent slowdown in research progress occurred as interest
in B. anthracis waned after the Biological Weapons Convention in the 1970s, as there were no
B-l
-------
published refinements in the biological models in the open literature. However, the discovery of
biological weapons in Iraq during the Gulf War in the 1990s followed by the 2001 anthrax letter
event again accelerated basic research and medical countermeasure efforts.
The approaches used to model microbial dose-response relationships for B. anthracis can be
described by the themes in research: (1) Determination of median lethality values, (2) Early
attempts to generate biologically-based models, (3) Modeling the initiation of infection, (4)
Consideration of the independent event hypothesis, and (5) Current approaches to modeling B.
anthracis pathogenesis and dose-response relationships. The themes do not reflect a strict
historical timeline as some of the themes reflect current questions in the field (e.g., modeling the
initiation of infection or consideration of the independent event hypothesis). This appendix will
provide a brief review of each theme while considering the overall state of progress in modeling
dose-response data for B. anthracis.
Determination of Median Lethality Values
Microbial pathogenesis or dose-response data for B. anthracis from the 1940s through the 1960s
was typically associated with state-sponsored laboratories, principally the U.S. Army Chemical
Corps laboratories (e.g., Fernelius et al. [1960], Lincoln et al. [1965], Lincoln et al. [1967a],
Lincoln et al. [1962], Klein et al. [1966], Jones et al. [1967]) or the United Kingdom's Porton
Down facility (e.g., Barnes [1947], Henderson [1952], Druett et al. [1953], Widdicombe et al.
[1956], Ross [1957]). Published dose-response relationship data for the rabbit and nonhuman
primate during this time primarily focused on reporting of median lethality values (e.g., Young et
al. [1946], Druett et al. [1953], Barnes [1947]; Henderson et al. [1956]). In the case of Young et
B-2
-------
al. (1946) and Barnes (1947), these values were published absent the initial data set or the
calculation of the value, with a primary focus of the articles associated with studies describing
pathogenesis or treatment of disease. The measurement of the median lethality value was the
primary output for most studies, with little consideration for the evaluation of other values or
describing the relationship between dose and response overall.
The Druett et al. (1953) study and associated dose-response data set was unique relative to its
contemporaries for a number of reasons. The stated purpose of the paper was to elucidate lung
regions associated with infection by testing various particle sizes. However, the study design
yielded an excellent data set to evaluate dose-response relationships (i.e., sufficient numbers of
animals, detailed study design description, reported all raw data). The study design also
evaluated the dose-response data using probit analysis allowing for identification of different
response levels than the median lethality values.
Most studies reporting median lethality values after the 1960s were typically designed for
purposes other than dose-response (e.g., pathology, medical countermeasures). In these studies,
high dose challenges (e.g., 100 to 200 times current estimates of median lethality values) were
conducted to ensure a high likelihood of systemic anthrax infection in the challenge animals. In
addition to reporting the strain, the reporting of the lethality value can provide an assessment of
the general "potency" or "virulence" of the test material. Depending on the application of the
data, current users of median lethality values include modelers for population hazard prediction,
planners, and human health risk assessors (Gutting et al., 2015).
B-3
-------
Early Attempts to Generate Biologically-Based Models
As an advance from the direct calculation of median lethality (LD50) values or probit-based
empirical modeling to generate dose-response relationships, a biologically-based model of
anthrax illness was first developed in the 1960s. These early mechanistic models modeled
bacteremia or even lethality, but they cannot be termed dose-response models because they did
not predict the probability of response.
A biologically-based mathematical model was developed to describe the kinetics of bacteremia
after intravenous administration of B. anthracis spores through death (Lincoln et al., 1962). The
first mechanistic model describing anthrax infection evaluated kinetic data to model biological
events but stopped short of developing a predictive dose-response relationship because there was
no mathematical association determined between the dose, either administered or internal, with a
probability of response endpoint (e.g., lethality). Bacteremia concentration was modeled over
time with boundary assumptions for identified parameter values and a mathematical expression
evaluating dose, net bacterial growth rate, and host resistance (i.e., passive and active resistance).
Active resistance, defined as phagocytosis and other immune reactions (e.g., fever), was modeled
using a negative exponential function with resistance assumed to go toward zero for later values
of time after infection.
Using these results and other study data developed at Fort Detrick's U.S. Army Biological
Laboratories group, Klein et al. (1963) conceptually described the resistance to establishment of
anthrax infection as being the collective outcome of two distinct and competing host-pathogen
B-4
-------
interactions: (1) the ability to establish bacterial growth and infection versus (2) the susceptibility
of the host to toxins produced during bacterial growth.
Using data from various animal models, an inverse relationship was identified for the resistance
to infection and susceptibility to toxin (Lincoln et al., 1967b). Resistance to infection was
measured by spore germination in phagocytes and/or parenteral dose to establish anthrax;
susceptibility to toxin was defined by lethality after intravenous administration.
As reported in Lincoln et al. (1967b), Kashiba et al. (1959) assessed inhibition of phagocytes by
terminal guinea pig serum, but American researchers could not replicate the results. As a result,
American researchers then focused their efforts on other spore-phagocyte interactions including
intracellular germination relative to spore numbers per phagocyte. Continued research efforts on
the inhibition of phagocytes by toxin was possibly delayed by decades in the United States as a
result.
Within the same U.S. laboratories, modeling of B. anthracis pathogenesis focused on elucidation
of a primarily systemic mode of action for the toxins, as evidenced by a number of studies in the
1960s that evaluated LD50 values for toxins administered intravenously or intraperitoneally (e.g.,
Klein et al. [1963], Lincoln et al. [1967b]). Evidence for toxemia as the cause of anthrax
mortality was based on the elicitation of anthrax symptoms and lethality reported after toxin
challenge studies. Decades later, data linking immunity to a component of the toxin (specifically,
the protective antigen [PA] component of both lethal toxin [LT] and edema toxin [ET]) with
conferred protection from anthrax infection also strengthened the association of toxemia with
lethality (Moayeri and Leppla, 2009; Coggeshall et al., 2013).
B-5
-------
Modeling the Initiation of Infection
The Trojan horse model is the first and most currently cited model for initiation of inhalation
anthrax since its publication in 2002 (Weiner and Glomski, 2012). The Trojan horse model is
principally based on the Ross (1957) description of spore engulfment and germination in the
alveolar macrophage combined with the Lincoln et al. (1965) reporting of transport of vegetative
bacteria to the lymphatic system. The continued availability of in-vitro and in-vivo cellular
techniques generated increasingly detailed mechanistic data on a potential role for the
macrophage in anthrax infection (Shafa et al., 1966; Hanna et al., 1993; Guidi-Rontani et al.,
1999b; Dixon et al., 2000). Most of the early in-vitro mechanistic work cited in the initial
proposal of the Trojan horse model utilized the mouse animal model or murine-derived cell lines
(Hanna et al., 1993; Guidi-Rontani et al., 1999a; Dixon et al., 2000), though Shafa et al. (1966)
evaluated macrophages from the rabbit. Using these mechanistic data, the Trojan horse model
hypothesizes the establishment of inhalation anthrax infection as an intracellular competition
between the B. anthracis spore, host macrophage, and toxins expressed by vegetative B.
anthracis (Guidi-Rontani, 2002). In the Trojan horse model, infection is initiated through
engulfment of the spore by alveolar macrophages and subsequent spore germination either
during transport to or upon arrival in the lymph node (Guidi-Rontani, 2002).
Using the Trojan horse model as a conceptual approach to model the initiation of infection, the
first dose-response models incorporating host-pathogen interaction were not published until the
2000s, nearly 40 years after the Fort Detrick group developed their kinetic model. This
interaction was conceptualized differently from the interaction presented by Klein et al. (1963)
with the two competing outcomes defined at a more basic fundamental level: (1) successful spore
B-6
-------
germination allowing proliferation of vegetative bacteria (i.e., germination) versus (2) removal
and/or destruction of the spore and associated vegetative bacteria (i.e., spore clearance).
Accordingly, a competing risk model to biologically model host-pathogen dynamics for
inhalation anthrax at the level of an individual spore was first described in Brookmeyer et al.
(2005) and Brookmeyer et al. (2003). Though the purpose of the Brookmeyer et al. (2005) and
Brookmeyer et al. (2003) models was to mechanistically model the incubation period for human
inhalation anthrax, a dose-response function was embedded within the overall model that could
be parameterized with human and/or animal model data. Using the competing risk mathematical
concept described in Brookmeyer et al. (2003) and Brookmeyer et al. (2005), a biologically-
based dose-response (BBDR) model was then published for the rabbit (Gutting et al., 2013) and
the nonhuman primate (Toth et al., 2013). For the Gutting et al. (2013) and Toth et al. (2013)
BBDR models, a comparison of the BBDR model outputs with empirical models or study data
was provided. However, statistical measures of model fit for each model type to allow
comparison with empirical modeling approaches were not included.
After the Trojan horse model was published, additional phagocytic cell types capable of
transporting B. anthracis spores to lymph nodes were identified through in-vitro studies of
human dendritic cells3 (Brittingham et al., 2005) and murine B cells (Rayamajhi et al., 2012).
Spore germination outside phagocytic cells in a murine animal model after inhalation and oral
exposure was reported in the lymphoid tissue of the respiratory tract and Peyer's patch tissues of
the intestine, respectively (Glomski et al., 2007; Lowe et al., 2013). Spore translocation into lung
3 Dendritic cells were identified in the original article describing the Trojan horse model as possibly providing a
vehicle for transport to the lymphatic system and subsequent germination location (Guidi-Rontani, 2002).
B-7
-------
epithelial cells was also reported from an in-vivo murine study, providing a route whereby the
spores could have a direct intracellular route to the lymphatic system (Russell et al., 2008).
To accommodate these new data, the jailbreak model expanded the Trojan horse model in three
important ways: (1) increased emphasis on the host-pathogen interactions in lymphoid and
epithelial tissues, (2) broadened the role of alveolar macrophages to include important elements
of host defense, and (3) expanded the number of potential cellular carriers to initiate infection
(Weiner and Glomski, 2012). The model is unique because it provides a conceptually consistent
approach to model the early stages of infection across the three natural routes of exposure:
inhalation, gastrointestinal, and cutaneous anthrax (Weiner and Glomski, 2012). Multiple
pathways by which inhalation anthrax may be initiated from the same route of exposure were
identified (Weiner and Glomski, 2012). Weiner and Glomski (2012) note that multiple distinct
pathways for initiation of infection have been identified for other microbial pathogens (e.g.,
salmonellae, shigellae, Listeria monocytogenes).
New concepts introduced in the jailbreak model include the potential for extracellular
germination of spores that do not require an intracellular phagocytic location for germination
while still allowing for subsequent transport to the lymph system (Weiner and Glomski, 2012).
The differing role for toxins in early infection is also notable. In the jailbreak model, spores
germinate in an extracellular environment and toxins are necessary to damage the integrity of
cellular barriers to facilitate access to the lymph system (Weiner and Glomski, 2012). In contrast,
toxins in the Trojan horse model facilitate successful intracellular germination through
modulation of the oxidative burst process within the phagocytic cells (Weiner and Glomski,
B-8
-------
2012). A subsequent paper notes that the identification of these multiple pathways does not
imply that mediastinal lymph node-initiated infections are not occurring in the murine or other
animal models, but that alternative or additional pathways may not be recognized absent
sensitive test methods and study approaches designed to capture these other pathways (Lowe et
al., 2013).
There are important differences between the Trojan horse and jailbreak models with regard to the
action of toxins. The Trojan horse model (Guidi-Rontani, 2002) described a localized action for
toxins as facilitating successful intracellular germination in the phagocyte and then allowing for
proliferation of vegetative bacteria. Alternately, the jailbreak model of Weiner and Glomski
(2012) identified toxin damage to endothelial or epithelial tissues as important to breaking key
barriers necessary for establishment of infection.
The identification of the new pathways for infection associated with the jailbreak model were
identified using bioluminescent techniques with the mouse small animal model and B. anthracis
spores of attenuated virulence. Of most relevance for this assessment, data are unavailable to
support or contraindicate the functional presence of these pathways in large animal models. A
key challenge for the development of these data is a technology comparable to the
bioluminescent techniques previously used in small animals (Glomski et al., 2007; Sanz et al.,
2008; Dumetz et al., 2011) that can precisely delineate the locations involved in the earliest
stages of infection in large animal models, such as the rabbit or nonhuman primate.
A key modeling determination for mechanistic models is the definition of infection. Differences
have arisen over time in the definition of anthrax infection, definitions ranging from conceptual
B-9
-------
to analytical. When developing conceptual models for microbial dose-response analysis,
Buchanan et al. (2009) characterized infection as the state where a pathogen can "actively
multiply" inside the host. More analytically-oriented definitions include seroconversion as
measured by a humoral response to protective antigen (PA) (U.S. Environmental Protection
Agency, 2011, 2012), confirmation of B. anthracis bacteremia via culture, or a combination of
these measurements. Henning et al. (2012) defined infection as the presence of a positive B.
anthracis blood culture combined with an electrochemiluminescent measurement of circulating
PA, with diagnostic measures noted to be observed earlier in the disease process than nonspecific
clinical signs. Boyer et al. (2009) confirmed the presence of infection using a combination of
bacteremia, blood differentials, and detection of the PA gene via polymerase chain reaction
(PCR) analysis.
The definition has evolved based on basic knowledge of the disease process, available
technology (e.g., analytical targets, detection limit), and desired end-use of the data (e.g.,
modeling, confirming presence/absence of anthrax infection, assessment of kinetics of disease).
Any definition will continue to be subject to modification as more sensitive measurement
technologies of potential biomarkers or new insights related to the infection process are
developed.
Consideration of the Independent Action Hypothesis
Druett (1952) provides the first articulation of the independent action hypothesis. Parts of the
mathematical derivation of the independent action hypothesis were previously presented in Bald
(1937) and were built upon by Druett (1952). However, the model was not termed independent
B-10
-------
action until Meynell and Stocker (1957) (U.S. Environmental Protection Agency, 2014). The
model is also referred to as the independent event hypothesis. Independent action among
pathogens was described by Druett (1952) as a constant relationship between response and the
product of administered dose (e.g., environmental concentration) and exposure time.4 Druett
(1952) reported general consistency between the probit slope value derived from a mathematical
model of the independent action hypothesis and the calculated probit slope values from single
dose challenge studies reporting B. anthracis5 and Brucella suis inhalation exposure and
mortality. The following assumptions were made in the mathematical derivation: a constant
probability for each organism to cause the identified response (i.e., mortality or infection) in the
host, independent action of each organism (e.g., no immune system activation), an LD50 value
that can be determined, and a large homogenous experimental population (Druett, 1952).
A literature review conducted by the U.S. Environmental Protection Agency (2014) found a
number of studies that described their data as consistent with the independent action hypothesis.
However, rigorous experimental evidence to distinguish between independent and inter-
dependent action hypotheses was limited for most host-pathogen systems (U.S. Environmental
Protection Agency, 2014).
4 Druett (1952) independently described the microbial equivalent of Haber's Law. Haber's Law, reported in the
early 1900s, also described a constant concentration-time relationship between exposure and mortality response for
exposure to inhalation exposure to volatile chemicals. Since that time, Haber's Law has been updated to include a
fitted exponent on the concentration term to better fit tested chemicals (ten Berge et al., 1986). Likewise, a fitted
exponent may also be found appropriate for the mathematical description of independent action.
5 The B. anthracis dose-response data were subsequently published in Druett et al. (1953).
B-ll
-------
The independent action hypothesis may be relevant for dose-response modeling in two primary
ways: the selection of appropriate dose-response models (Haas et al., 1999; Food and Agriculture
Organization and World Health Organization (FAO and WHO), 2003) and the determination of
dose metrics for multiple dose exposures (U.S. Environmental Protection Agency, 2014). When
defined as mechanistic models, the exponential and beta-Poisson models are consistent with the
independent action hypothesis and therefore, some researchers have identified them as preferable
for microbial dose-response modeling (Haas et al., 1999; Food and Agriculture Organization and
World Health Organization (FAO and WHO), 2003). However, the use of empirical models does
not require a mechanistic interpretation of the model parameters and therefore a broader
consideration of available mathematical models for microbial dose-response analysis has also
been identified as appropriate (Holcomb et al., 1999; Coleman and Marks, 2000; Taft and Hines,
2012).
Independent action may not be a trait universally expressed among microbial pathogens at all
times, but may present some dependencies based on microbial pathogen, route of exposure,
magnitude of dose, or timing of doses. If the independent action hypothesis were correct, the
total dose would be an appropriate dose metric for a B. anthracis, and there would be no
biological rationale for consideration of a daily average dose. However, a limitation to the
exposure duration over which independent action could be assumed (e.g., short enough to
preclude immune system activation) was noted by Druett (1952) in the original formulation of
the hypothesis. Though Druett (1952) developed the hypothesis with single dose data, the
concept should be equally relevant to multiple dose assessments. The independent action
hypothesis should allow for the use of an aggregate dose metric only if the exposure time over
B-12
-------
which the daily doses were aggregated did not exceed the time duration associated with dose
independence. Mayer et al. (2011) also noted that dose-response models lacking consistency with
independent action assumptions may be warranted under conditions of time-dependency of doses
where independent action may be less likely to occur (e.g., exposures with multiple closely
spaced doses in B. anthracis).
The magnitude of exposure or exposure duration (Mayer et al., 2011; U.S. Environmental
Protection Agency, 2014) where independent doses can be delineated from dependent doses has
not been evaluated explicitly to date. Dose-dependencies may be present in the expression of
independent action whereby larger doses could affect response to subsequent doses if
overloading of clearance or other innate immune functions were affected (Mayer et al., 2011). If
overloading can occur, this implies that the presence of independent action could vary by route
of exposure if varying innate response levels are present (e.g., differential innate response for
dermal versus inhalation routes of exposure). The timing of the exposures relative to the dose
and clearance capabilities is also a critical exposure consideration relative to the selection of dose
metrics (Mayer et al., 2011).
The determination of a theoretical time point separating independent and dependent doses may
be considerably more complicated for inhaled pathogens that have the potential to persist in the
lungs (U.S. Environmental Protection Agency, 2014). For example, spore persistence in the lung
with subsequent inhalation anthrax has been reported in one nonhuman primate that died 58 days
after exposure after initially receiving 30 days of antibiotic treatment starting on the exposure
B-13
-------
day (Friedlander et al., 1993). In this context, a total accumulated dose could be an appropriate
dose metric.
Current Approaches to Modeling B. anthracis Pathogenesis and Dose-
Response Relationships
Empirical dose-response relationships continue to be used for the modeling of dose-response
relationships in the nonhuman primate (Haas, 2002; Bartrand et al., 2008; Weir and Haas, 2011;
Taft and Hines, 2012) and rabbit (U.S. Environmental Protection Agency, 2011, 2012). The
availability of statistical software capable of fitting dose-response data to mathematical models
has considerably broadened the models available for evaluation. The U.S. Environmental
Protection Agency (2011, 2012) studies were designed to include representation of low-dose
exposure ranges. The purpose of the EPA studies was to design studies and derive dose-response
relationships relevant to the assessment of residual biological contamination present after
application of decontamination technologies. Data gaps identified during remediation after the
2001 anthrax letter event provide an impetus for new dose-response studies and identified the
need for reliable means to assess risk in the low-dose range (Gutting et al., 2008).
Hybrid models of empirically fit parameters combined with expert elicited dose-response values
have been included as elements of population-based anthrax models for the human (Webb and
Blaser, 2002; Wein et al., 2003; Wein and Craft, 2005). Likewise, empirically fit models have
been developed using a survival analysis framework to incorporate time dependencies in dosing
and/or response (Mayer et al., 2011; U.S. Environmental Protection Agency, 2014).
B-14
-------
Recently published biologically-based models for anthrax infection and illness evaluate the
timing, type, and likely success of medical countermeasures (Kumar et al., 2008), develop a
better understanding of early infection dynamics (Day et al., 2011), evaluate the incubation
period (Brookmeyer and Blades, 2003; Brookmeyer et al., 2003; Brookmeyer et al., 2005;
Wilkening, 2008), assess the spatial and temporal concordance of anthrax cases from the
Sverdlovsk outbreak (Wilkening, 2006), and evaluate time-dependence in dose-response analysis
of multiple doses (Mayer et al., 2011). Clearance of inhaled B. anthracis spores currently plays a
key role in mechanistic modeling approaches for infection and response to exposure. However,
the relationship between external exposure and clearance has been identified as a major
uncertainty in B. anthracis dose-response prediction (Coleman et al., 2008). These biologically
based models may provide important components of a comprehensive biologically-based dose-
response model if linkages are made between dose, model components, and response endpoint(s)
of potential interest. However, the mechanisms associated with dose-dependence in outcomes
exhibit significant uncertainty (U.S. Environmental Protection Agency, 2014).
Conclusion
As the primary end users for B. anthracis microbial dose-response outputs have broadened after
the 2001 anthrax letter event, an additional focus for modeling B. anthracis dose-response
relationships has included the prediction of the hazard posed by low-dose exposure. This
additional focus has led to a renewed interest in biologically-based dose-response models that
can incorporate dose-dependent mechanisms associated with low response levels and assist in
predicting response differences between the animal model and the human. Much progress has
B-15
-------
been made from the early emphasis on LD50 values to a more comprehensive understanding of
the disease pathogenesis and its translation to mathematical models.
Bibliography
Bald, J. G. (1937). The Use of Numbers of Infections for Comparing the Concentration of Plant
Virus Suspensions. I. Dilution Experiments with Purified Suspensions. Annals of Applied
Biology 24: 33-55.
Barnes, J. M. (1947). The Development of Anthrax Following the Administration of Spores by
Inhalation. British Journal of Experimental Pathology 28(6): 385-394.
Bartrand, T. A., M. H. Weir and C. N. Haas (2008). Dose-Response Models for Inhalation of
Bacillus anthracis Spores: Interspecies Comparisons. Risk Analysis 28(4): 1115-1124.
Boyer, A. E., C. P. Quinn, A. R. Hoffmaster, T. R. Kozel, E. Saile, C. K. Marston, A. Percival,
B. D. Plikaytis, A. R. Woolfitt, M. Gallegos, P. Sabourin, L. G. McWilliams, J. L. Pirkle
and J. R. Barr (2009). Kinetics of Lethal Factor and Poly-D-Glutamic Acid Antigenemia
During Inhalation Anthrax in Rhesus Macaques. Infection and Immunity 77(8): 3432-
3441.
Brittingham, K. C., G. Ruthel, R. G. Panchal, C. L. Fuller, W. J. Ribot, T. A. Hoover, H. A.
Young, A. O. Anderson and S. Bavari (2005). Dendritic Cells Endocytose Bacillus
anthracis Spores: Implications for Anthrax Pathogenesis. The Journal of Immunology
174: 5545-5552.
Brookmeyer, R. and N. Blades (2003). Statistical Models and Bioterrorism: Application to the
US Anthrax Outbreak. Journal of the American Statistical Association 98(464): 781-788.
Brookmeyer, R., E. Johnson and S. Barry (2005). Modelling the Incubation Period of Anthrax.
Statistics in Medicine 24(4): 531-542.
Brookmeyer, R., E. Johnson and R. Bollinger (2003). Modeling the Optimum Duration of
Antibiotic Prophylaxis in an Anthrax Outbreak. Proceedings of the National Academy of
Sciences of the United States of America 100(17): 10129-10132.
Buchanan, R. L., A. H. Havelaar, M. A. Smith, R. C. Whiting and E. Julien (2009). The Key
Events Dose-Response Framework: Its Potential for Application to Foodborne
Pathogenic Microorganisms. Critical Reviews in Food Science and Nutrition 49(8): 718-
728.
B-16
-------
Coggeshall, K. M., F. Lupu, J. Ballard, J. P. Metcalf, J. A. James, D. Farris and S. Kurosawa
(2013). The Sepsis Model: An Emerging Hypothesis for the Lethality of Inhalation
Anthrax. Journal of Cellular and Molecular Medicine 17(7): 914-920.
Coleman, M. E. and H. M. Marks (2000). Mechanistic Modeling of Salmonellosis. Quantitative
Microbiology 2(3): 227-247.
Coleman, M. E., B. Thran, S. S. Morse, M. Hugh-Jones and S. Massulik (2008). Inhalation
Anthrax: Dose Response and Risk Analysis. Biosecurity and Bioterrorism: Biodefense
Strategy, Practice, and Science 6(2): 147-160.
Day, J., A. Friedman and L. S. Schlesinger (2011). Modeling the Host Response to Inhalation
Anthrax. Journal of Theoretical Biology 276(1): 199-208.
Dixon, T. C., A. A. Fadl, T. M. Koehler, J. A. Swanson and P. C. Hanna (2000). Early Bacillus
anthracis-Macrophage Interactions: Intracellular Survival and Escape. Cellular
Microbiology 2(6): 453-463.
Druett, H. A. (1952). Bacterial Invasion. Nature 170(4320): 288.
Druett, H. A., D. W. Henderson, L. Packman and S. Peacock (1953). Studies on Respiratory
Infection: I. The Influence of Particle Size on Respiratory Infection with Anthrax Spores.
The Journal of Hygiene (London) 51(3): 359-371.
Dumetz, F., G. Jouvion, H. Khun, I. J. Glomski, J.-P. Corre, C. Rougeaux, W.-J. Tang, M. Mock,
M. Huerre and P. L. Goossens (2011). Noninvasive Imaging Technologies Reveal Edema
Toxin as a Key Virulence Factor in Anthrax. The American Journal of Pathology 178(6):
2523-2535.
Fernelius, A. L., I. A. Dearmon, F. Klein and R. E. Lincoln (1960). Comparison of Graded and
Quantal Virulence Tests for Bacillus anthracis Species. Journal of Bacteriology 79(4):
594-600.
Food and Agriculture Organization and World Health Organization (FAO and WHO) (2003).
Hazard Characterization for Pathogens in Food and Water Guidelines. World Health
Organization: Rome, Italy and Geneva, Switzerland. Microbiological Risk Assessment
Series No. 3.
Friedlander, A. M., S. L. Welkos, M. L. M. Pitt, J. W. Ezzell, P. L. Worsham, K. J. Rose, B. E.
Ivins, J. R. Lowe, G. B. Howe, P. Mikesell and W. B. Lawrence (1993). Postexposure
Prophylaxis against Experimental Inhalation Anthrax. Journal of Infectious Diseases
167(5): 1239-1243.
B-17
-------
Glomski, I. J., A. Piris-Gimenez, M. Huerre, M. Mock and P. L. Goossens (2007). Primary
Involvement of Pharynx and Peyer's Patch in Inhalational and Intestinal Anthrax. PloS
Pathogens 3(6): e76.
Guidi-Rontani, C. (2002). The Alveolar Macrophage: The Trojan Horse of Bacillus anthracis.
Trends in Microbiology 10(9): 405-409.
Guidi-Rontani, C., Y. Pereira, S. Ruffle, J.-C. Sirard, M. Weber-Levy and M. Mock (1999a).
Identification and Characterization of a Germination Operon on the Virulence Plasmid
pXOl of Bacillus anthracis. Molecular Microbiology 33(2): 407-414.
Guidi-Rontani, C., M. Weber-Levy, E. Labruyere and M. Mock (1999b). Germination of
Bacillus anthracis Spores within Alveolar Macrophages. Molecular Microbiology 31(1):
9-17.
Gutting, B. W., S. R. Channel, A. E. Berger, J. M. Gearhart, G. A. Andrews, R. L. Sherwood and
T. L. Nichols (2008). Mathematically Modeling Inhalational Anthrax. Microbe 3(2): 78-
85.
Gutting, B. W., D. Marchette, R. Sherwood, G. A. Andrews, A. Director-Myska, S. R. Channel,
D. Wolfe, A. E. Berger, R. S. Mackie, B. J. Watson and A. Rukhin (2013). Modeling
Low-Dose Mortality and Disease Incubation Period of Inhalational Anthrax in the Rabbit.
Journal of Theoretical Biology 329: 20-31.
Gutting, B. W., A. Rukhin, R. S. Mackie, D. Marchette and B. Thran (2015). Evaluation of
Inhaled Versus Deposited Dose Using the Exponential Dose-Response Model for
Inhalational Anthrax in Nonhuman Primate, Rabbit, and Guinea Pig. Risk Analysis.
35(5): 811-827.
Haas, C. N. (2002). On the Risk of Mortality to Primates Exposed to Anthrax Spores. Risk
Analysis 22(2): 189-193.
Haas, C. N., J. B. Rose and C. P. Gerba (1999). Chapter 7. Conducting the Dose-Response
Assessment. Quantitative Microbial Risk Assessment. New York, John Wiley & Sons,
Inc.: 260-319.
Hanna, P. C., D. Acosta and R. J. Collier (1993). On the Role of Macrophages in Anthrax.
Proceedings of the National Academy of Sciences of the United States of America
90(21): 10198-10201.
Henderson, D. W. (1952). An Apparatus for the Study of Airborne Infection. Journal of Hygiene
(London) 50(1): 53-68.
B-18
-------
Henderson, D. W., S. Peacock and F. C. Belton (1956). Observations on the Prophylaxis of
Experimental Pulmonary Anthrax in the Monkey. Journal of Hygiene (London) 54(1):
28-36.
Henning, L. N., J. E. Comer, G. V. Stark, B. D. Ray, K. P. Tordoff, K. A. B. Knostman and G. T.
Meister (2012). Development of an Inhalational Bacillus anthracis Exposure Therapeutic
Model in Cynomolgus Macaques. Clinical and Vaccine Immunology 19(11): 1765-1775.
Hilmas, C. J., A. M. Katos, P. T. Williams and J. Anderson (2009). Chapter 31. Anthrax.
Handbook of Toxicology of Chemical Warfare Agents. R. C. Gupta. Boston, Academic
Press: 433-462.
Holcomb, D. L., M. A. Smith, G. O. Ware, Y. C. Hung, R. E. Brackett and M. P. Doyle (1999).
Comparison of Six Dose-Response Models for Use with Food-Borne Pathogens. Risk
Analysis 19(6): 1091-1100.
Jones, W. I., Jr.,, F. Klein, J. S. Walker, B. G. Mahlandt, J. P. Dobbs and R. E. Lincoln (1967). In
Vivo Growth and Distribution of Anthrax Bacilli in Resistant, Susceptible, and
Immunized Hosts. Journal of Bacteriology 94(3): 600-608.
Klein, F., B. W. Haines, B. G. Mahlandt, I. A. DeArmon, Jr. and R. E. Lincoln (1963). Dual
Nature of Resistance Mechanisms as Revealed by Studies of Anthrax Septicemia. Journal
of Bacteriology 85(5): 1032-1038.
Klein, F., J. S. Walker, D. F. Fitzpatrick, R. E. Lincoln, B. G. Mahlandt, W. I. Jones, Jr., J. P.
Dobbs and K. J. Hendrix (1966). Pathophysiology of Anthrax. Journal of Infectious
Diseases 116(2): 123-138.
Kumar, R., C. C. Chow, J. D. Bartels, G. Clermont and Y. Vodovotz (2008). A Mathematical
Simulation of the Inflammatory Response to Anthrax Infection. Shock 29(1): 104-111.
Lincoln, R. E., D. R. Hodges, F. Klein, B. G. Mahlandt, W. I. Jones, Jr., B. W. Haines, M. A.
Rhian and J. S. Walker (1965). Role of the Lymphatics in the Pathogenesis of Anthrax.
Journal of Infectious Diseases 115(5): 481-494.
Lincoln, R. E., F. Klein, D. R. Hodges, B. W. Haines and M. A. Rhian (1962). Pathogenesis of
Anthrax - A Progress Report: No. TM-19, United States Army Biological Laboratories:
Fort Detrick, MD. CBRNIAC No. ADA388172.
Lincoln, R. E., J. S. Walker and F. Klein (1967a). Value of Field Data for Extrapolation in
Anthrax, Department of the Army Experimental Aerobiology Division: Fort Detrick,
MD. CBRNIAC No. AD810305.
B-19
-------
Lincoln, R. E., J. S. Walker, F. Klein, A. J. Rosenwald and W. I. Jones, Jr. (1967b). Value of
Field Data for Extrapolation in Anthrax. Federation Proceedings 26(5): 1558-1562.
Lowe, D. E., S. M. C. Ernst, C. Zito, J. Ya and I. J. Glomski (2013). Bacillus anthracis Has Two
Independent Bottlenecks That Are Dependent on the Portal of Entry in an Intranasal
Model of Inhalational Infection. Infection and Immunity 81(12): 4408-4420.
Mayer, B. T., J. S. Koopman, E. L. Ionides, J. M. Pujol and J. N. S. Eisenberg (2011). A
Dynamic Dose-Response Model to Account for Exposure Patterns in Risk Assessment: A
Case Study in Inhalation Anthrax. Journal of the Royal Society Interface 8(57): 506-517.
Meynell, G. G. and B. A. D. Stocker (1957). Some Hypotheses on the Aetiology of Fatal
Infections in Partially Resistant Hosts and Their Application to Mice Challenged with
Salmonella paratyphi-B or Salmonella typhimurium by Intraperitoneal Injection. Journal
of General Microbiology 16(1): 38-58.
Moayeri, M. and S. H. Leppla (2009). Cellular and Systemic Effects of Anthrax Lethal Toxin
and Edema Toxin. Molecular Aspects of Medicine 30(6): 439-455.
Rayamajhi, M., C. Delgado, T. V. Condon, D. W. Riches and L. L. Lenz (2012). Lung B Cells
Promote Early Pathogen Dissemination and Hasten Death from Inhalation Anthrax.
Mucosal Immunology 5(4): 444-454.
Ross, J. M. (1957). The Pathogenesis of Anthrax Following the Administration of Spores by the
Respiratory Route. The Journal of Pathology and Bacteriology 73(2): 485-494.
Russell, B. H., Q. Liu, S. A. Jenkins, M. J. Tuvim, B. F. Dickey and Y. Xu (2008). In Vivo
Demonstration and Quantification of Intracellular Bacillus anthracis in Lung Epithelial
Cells. Infection and Immunity 76(9): 3975-3983.
Sanz, P., L. D. Teel, F. Alem, H. M. Carvalho, S. C. Darnell and A. D. O'Brien (2008). Detection
of Bacillus anthracis Spore Germination in Vivo by Bioluminescence Imaging. Infection
and Immunity 76(3): 1036-1047.
Shafa, F., B. J. Moberly and P. Gerhardt (1966). Cytological Features of Anthrax Spores
Phagocytized in Vitro by Rabbit Alveolar Macrophages. The Journal of Infectious
Diseases 116(4): 401-413.
Taft, S. C. and S. A. Hines (2012). Benchmark Dose Analysis for Bacillus anthracis Inhalation
Exposures in the Nonhuman Primate. Risk Analysis 32(10): 1750-1768.
ten Berge, W. F., A. Zwart and L. M. Appelman (1986). Concentration-Time Mortality Response
Relationship of Irritant and Systemically Acting Vapours and Gases. Journal of
Hazardous Materials 13(3): 301-309.
B-20
-------
Toth, D. J. A., A. V. Gundlapalli, W. A. Schell, K. Bulmahn, T. E. Walton, C. W. Woods, C.
Coghill, F. Gallegos, M. H. Samore and F. R. Adler (2013). Quantitative Models of the
Dose-Response and Time Course of Inhalational Anthrax in Humans. PloS Pathogens
9(8): el003555.
U.S. Environmental Protection Agency (2011). Acute Low Dose Bacillus anthracis Ames
Inhalation Exposures in the Rabbit. Cincinnati, OH: National Homeland Security
Research Center. U.S. Environmental Protection Agency. EPA/600/R-11/075.
U.S. Environmental Protection Agency (2012). Multiple Daily Low-Dose Bacillus anthracis
Ames Inhalation Exposures in the Rabbit. Washington, DC: Office of Research and
Development, National Homeland Security Research Center. U.S. Environmental
Protection Agency. EPA/600/R-11/145.
U.S. Environmental Protection Agency (2014). Modeling Rabbit Responses to Acute and
Multiple Inhalation Exposures of Bacillus anthracis Spores. Cincinnati, Ohio: National
Homeland Security Research Center. U.S. Environmental Protection Agency.
EPA/600/R-14/058.
Webb, G. F. and M. J. Blaser (2002). Mailborne Transmission of Anthrax: Modeling and
Implications. Proceedings of the National Academy of Sciences of the United States of
America 99(10): 7027-7032.
Wein, L. M. and D. L. Craft (2005). Evaluation of Public Health Interventions for Anthrax: A
Report to the Secretary's Council on Public Health Preparedness. Biosecurity and
Bioterrorism: Biodefense Strategy, Practice, and Science 3(4): 348-356.
Wein, L. M., D. L. Craft and E. H. Kaplan (2003). Emergency Response to an Anthrax Attack.
Proceedings of the National Academy of Sciences of the United States of America
100(7): 4346-4351.
Weiner, Z. P. and I. J. Glomski (2012). Updating Perspectives on the Initiation of Bacillus
anthracis Growth and Dissemination through its Host. Infection and Immunity 80(5):
1626-1633.
Weir, M. H. and C. N. Haas (2011). A Model for in-Vivo Delivered Dose Estimation for Inhaled
Bacillus anthracis Spores in Humans with Interspecies Extrapolation. Environmental
Science and Technology 45(13): 5828-5833.
Widdicombe, J. G., R. Hughes and A. J. May (1956). The Role of the Lymphatic System in the
Pathogenesis of Anthrax. British Journal of Experimental Pathology 37(4): 343-349.
B-21
-------
Wilkening, D. A. (2006). Sverdlovsk Revisited: Modeling Human Inhalation Anthrax.
Proceedings of the National Academy of Sciences of the United States of America
103(20): 7589-7594.
Wilkening, D. A. (2008). Modeling the Incubation Period of Inhalational Anthrax. Medical
Decision Making 28(4): 593-605.
Young, Jr., G. A., M. R. Zelle and R. E. Lincoln (1946). Respiratory Pathogenicity of Bacillus
anthracis Spores I. Methods of Study and Observations on Pathogenesis. Journal of
Infectious Diseases 79(3): 233-246.
B-22
-------
Appendix C - Data Summary Table for End-stage Inhalation Anthrax
Pathology of the Human, Nonhuman Primate, and Rabbit
C-l
-------
Table C-l. Data Summary Table for End-stage Inhalation Anthrax Pathology of the Human, Nonhuman Primate, and Rabbit
System
Rabbit
Nonhuman Primate
Human
Immune System
Including Lymph
Nodes (LNs),
Spleen, Thymus,
and Gut-
associated
Lymphoid Tissue
Hemorrhagic lymphadenitis, most often
mediastinal and submandibular LN, with
lymphoid necrosis in draining LN (Zaucha et al.,
1998; U.S. Environmental Protection Agency,
2011; Lovchik et al., 2012); lymphoid depletion
(Zaucha et al., 1998); presence of fibrin,(U.S.
Environmental Protection Agency, 2011;
Lovchik et al., 2012; U.S. Environmental
Protection Agency, 2012); edema (Lovchik et al.,
2012; U.S. Environmental Protection Agency,
2012)
Mediastinal lesions, less severe than noted in
human (Zaucha et al., 1998); connective tissue
and fat displaying edema and hemorrhage
(Lovchik et al, 2012)
Lesions in gut-associated lymphoid tissues of
sacculus rotimdus (Zaucha et al., 1998); cecal
appendix (Zaucha et al., 1998) and ileum (Zaucha
et al., 1998); lymphocyte necrosis and depletion
in lymphoid tissue of sacculus rotundus and cecal
appendix (Lovchik et al., 2012)
Hemorrhage and necrosis in appendix (U.S.
Environmental Protection Agency, 2012)
Lymphoid atrophy and edema in thymus (U.S.
Environmental Protection Agency, 2011) or
lymphocyte necrosis and depletion in thymus
(Lovchik et al., 2012)
Splenomegaly, with acute fibrinous splenitis
(Zaucha et al., 1998; Yee et al., 2010; Lovchik et
al., 2012); necrosis (Zaucha et al., 1998; Yee et
al., 2010; Lovchik et al., 2012); hemorrhage
(Zaucha et al., 1998; Lovchik et al., 2012);
Hemorrhagic, enlarged and/or edema in
mediastinal LN (Albrink and Goodlow, 1959;
Twenhafel et al., 2007; Lever et al., 2008;
Henning et al., 2012); Necrosis in mediastinal
LN (23/23) (Dalldorf et al., 1971);
tracheobronchial LN (Albrink and Goodlow,
1959; Fritz et al., 1995; Twenhafel et al., 2007;
Lever et al., 2008); intrathoracic LN with some
necrosis (Gleiser et al., 1963); axillar and
inguinal LN (Fritz et al., 1995; Twenhafel et al.,
2007); mesenteric LN (Twenhafel et al., 2007);
cervical LN engorged with neutrophils (16/23);
with some necrosis (4/21) (Dalldorf et al., 1971)
Secondary follicular development including focal
fibrin deposition (Lever et al., 2008); edema
(Middleton and Standen, 1961; Fritz et al., 1995;
Twenhafel et al., 2007); depletion and necrosis
of lymphocytes (Middleton and Standen, 1961;
Fritz et al., 1995; Henning et al., 2012); sinus
histiocytosis (Middleton and Standen, 1961; Fritz
et al., 1995); infiltration by neutrophils (Albrink
and Goodlow, 1959)
Mediastinal tissues with edema and/or
hemorrhage (Gleiser et al., 1963; Vasconcelos et
al., 2003); massive hemorrhagic mediastinitis not
observed (Gleiser et al., 1963); acute suppurative
inflammation (4/14) (Vasconcelos et al., 2003)
Mesenteric LN with hemorrhage and/or edema
(Fritz et al., 1995)
Splenomegaly (Albrink and Goodlow, 1959;
Middleton and Standen, 1961; Gleiser et al.,
1963; Lever et al., 2008); though with low
incidence identified from one study (3/13) (Fritz
Mediastinal LN with hemorrhage (Barakat et al.,
2002; Gill and Melinek, 2002; Guarner and del
Rio, 2011); necrosis (Barakat et al., 2002; Gill
and Melinek, 2002; Guarner and del Rio, 2011)
lymphocytosis (Guarner and del Rio, 2011)
infiltration by neutrophils and immunoblasts
(Guarner and del Rio, 2011) and hemorrhagic
necrosis of thoracic LN (Abramova et al., 1993)
Hilar and peribronchial LNs enlarged, necrotic,
with hemorrhage (Mina et al., 2002)
Mediastinitis with hemorrhage (Albrink et al.,
1960; Suffin et al., 1978; Inglesby et al., 2002;
Mina et al., 2002); necrosis (Suffin et al., 1978;
Inglesby et al., 2002) and acute inflammation
(Suffin et al., 1978) or edema (Albrink et al.,
1960)
Mesenteric lymphadenitis in limited number of
cases (9/42); with less severe involvement than
thoracic LN (Abramova et al., 1993)
Splenomegaly with hemorrhage (Albrink et al.,
1960); congestion (Suffin et al., 1978); necrosis
(Barakat et al., 2002; Guarner et al., 2003);
moderate to marked lymphocytolysis, minimal
atrophy of follicles, thickening of Bilroth cords
(Grinberg et al., 2001)
C-2
-------
System Rabbit Nonhuman Primate Human
lesions, lymphocyte necrosis and depletion
et al., 1995) or described as mild (Twenhafel et
(Lovchik et al., 2012)
al., 2007); with diffuse hepatic congestion, fibrin
deposition, and expanded germinal center (Lever
et al., 2008); lymphocytic depletion (Fritz et al.,
1995); histiocytosis (Fritz et al., 1995) with
hemorrhage in splenic marginal zone (Fritz et al.,
1995); necrosis of lymph follicles and/or necrosis
of red and white pulp with hemorrhage (21/23)
(Dalldorf etal., 1971)
Respiratory
System
Cardiovascular
System
Including Heart
and Blood
Vessels
Necrotizing hemorrhagic pulmonary lesions, with
lower incidence of pneumonia than human
(Zaucha et al., 1998)
Congestion of alveolar capillaries with large
numbers of bacteria, interstitial edema, and
minimal to mild perivascular infiltration of
heterophils (Zaucha et al., 1998); or occasional
edema, presence of fibrin, and hemorrhage
(Lovchik et al., 2012)
Congestion, edema, fibrin, and bacteria in lamina
propria and submucosa of trachea (Yee et al.,
2010)
Suppurative inflammation in lung (U.S.
Environmental Protection Agency, 2011,2012)
Potential indirect exposure effect reported as
infiltration of multi-nucleated giant cells in
response to foreign body (U.S. Environmental
Protection Agency, 2011)
Necrotizing hemorrhagic lesions in myocardium
(Zaucha et al., 1998)
Mild myodegeneration, necrosis, and subacute
inflammation, with histiocytes, mononuclear
cells, and heterophils (Note: Reported from study
administering lethal toxin only) (Lawrence et al.,
2011 )
Hemorrhagic pneumonia (Albrink and Goodlow,
1959; Lever et al., 2008); low incidence of
pneumonia (2/13) but presence of hemorrhages
(Fritz et al., 1995)
Pleural effusions (Albrink and Goodlow, 1959;
Dalldorf et al., 1971; Vasconcelos et al., 2003;
Twenhafel et al., 2007); though not reported in
rhesus macaque (Twenhafel et al., 2007)
Edema of the trachea and bronchial mucosa
(Albrink and Goodlow, 1959)
Hemorrhage of varying severity in the lung
(Gleiser et al., 1963; Vasconcelos et al., 2003;
Twenhafel et al., 2007), alveoli filled with edema
often mixed with fibrin, hemorrhage,
macrophages, and neutrophils (Twenhafel et al.,
2007); acute suppurative inflammation (4/14)
(Vasconcelos et al., 2003)
Hemorrhage in myocardium (2/13) (Fritz et al.,
1995) and (4/14) (Vasconcelos et al., 2003), with
acute myocarditis (1/13) (Fritz et al., 1995) and
acute suppurative inflammation (4/14)
(Vasconcelos et al., 2003)
Pericardial effusions (Twenhafel et al., 2007)
Necrotizing, hemorrhagic pneumonia with
primary foci present (Abramova et al., 1993)
Pleural effusions (at autopsy or drained prior to
death) (LaForce et al., 1969; Jernigan et al.,
2001; Barakat et al., 2002; Mina et al., 2002;
Guarner et al., 2003)
Perihilar interstitial pneumonia (Grinberg et al.,
2001); acute bronchial pneumonia (Grinberg et
al., 2001)
Pulmonary edema (Abramova et al., 1993; Mina
et al., 2002), including intra-alveolar and
interstitial edema with focal hemorrhage and
fibrin deposition (Barakat et al., 2002)
Hemorrhage and edema in laminae propriae of
the major bronchi and trachea, with lymph nodes
and connective tissue adjacent to bifurcation of
the trachea hemorrhagic and edematous (Albrink
et al., 1960)
Evidence of hematogenous spread of disease
(Grinberg et al., 2001)
Vasculitis, with necrosis of arteries and veins
(Grinberg et al., 2001)
High and low pressure hemorrhages (Grinberg et
al., 2001); with high pressure hemorrhages more
C-3
-------
System Rabbit
Nonhuman Primate
Human
Gastrointestinal
System
Hemorrhage, necrosis, and lymphoid depletion in
appendix (U.S. Environmental Protection
Agency, 2012)
Edema, hemorrhage, and necrosis in cecum (U.S.
Environmental Protection Agency, 2012)
Central Nervous
System
Brain and/or meningeal lesions with no
leukocytic infiltrate (Zaucha et al., 1998)
Liver congestion (Albrink and Goodlow, 1959;
Lever et al., 2008)
Acute inflammation/leukocytosis (13/14) and
acute necrosis (5/14) in liver (Vasconcelos et al.,
2003); sinusoidal leukocytosis (9/10); necrosis
(6/10) and acute inflammation (4/10) (Henning et
al., 2012)
Foci of hemorrhage in pancreas (1/13) (Fritz et
al., 1995)
Elemorrhages of varying severity in the small and
large intestine serosa and esophagus mucosa
(Fritz et al., 1995) or stomach mucosa and/or
submucosal (Fritz et al., 1995; Vasconcelos et
al., 2003) with acute colitis with necrotizing
vasculitis (1/13) (Fritz et al., 1995); necrosis of
villus tips in ileum or jejunum (9/14)
(Vasconcelos et al., 2003); or stomach with
inflammation (2/14) or ulceration (1/14)
(Vasconcelos et al., 2003)
Edema, congestion, and hemorrhage in the
gastrointestinal tract (Twenhafel et al., 2007)
Meningeal hemorrhage (Gleiser et al., 1963;
Dalldorf et al., 1971; Fritz et al., 1995;
Vasconcelos et al., 2003; Twenhafel et al., 2007;
frequently identified in Sverdlovsk than
Amerithrax victims (Guarner et al., 2003)
No specific cardiac microscopic findings
(Grinberg et al., 2001)
Pericardial effusions (Jemigan et al., 2001; Mina
et al., 2002); wall of left ventricle increased in
thickness (Albrink et al., 1960) and moderate
subendocardial hemorrhage of left ventricle
(Albrink et al., 1960)
Gastrointestinal submucosal lesions (Abramova
et al., 1993; Inglesby et al., 2002)
Necrosis, hemorrhage, and edema of the ileum
(Albrink et al., 1960)
Meningitis (Inglesby et al., 2002) including
hemorrhagic meningitis (Plotkin et al., 2002);
"Cardinal's Cap" (Inglesby et al., 2002) from
C-4
-------
System Rabbit
Nonhuman Primate
Human
Bacilli in meninges (Peterson et al., 2007)
Meningitis with suppurative inflammation (U.S.
Environmental Protection Agency, 2011)
Lever et al., 2008); including relatively minor
levels of hemorrhage (Lever et al., 2008); higher
incidence in high versus low-dose groups
(Gleiser et al., 1963), low overall incidence
(1/10) (Henning et al., 2012);
hemorrhage over entire surface of cerebrum,
cerebellum, and brain stem (Twenhafel et al.,
2007); and necrotizing vasculitis (2/14)
(Vasconcelos et al., 2003)
Meningeal edema (Dalldorf et al., 1971;
Vasconcelos et al., 2003)
Parenchymal hemorrhage in the brain (3/13)
(Leveret al., 2008)
Meningitis (9/21) (Dalldorf et al., 1971);
suppurative meningitis (10/13) (Fritz et al., 1995)
Edema in brain without hemorrhage (Albrink and
Goodlow, 1959; Gleiser et al., 1963); or with
acute hemorrhage (1/10) (Henning et al., 2012)
Occasional neuronal necrosis, spongiosis, gliosis,
hemorrhage, neutrophils, and edema in cerebrum
and cerebellum (Twenhafel et al., 2007)
Localized necrosis with accompanying cellular
changes and overall decrease in number of glia
(Henning et al., 2012)
hemorrhage of leptomeninges, more frequently
identified from Sverdlovsk than 2001 anthrax
letter event victims (Guarner and del Rio, 2011)
Subarachnoid hemorrhage, extensive at times
including covering frontal, parietal, temporal,
and occipital lobes (Suffin et al., 1978) or fully
covering both cerebral hemispheres (Albrink et
al., 1960)
C-5
-------
System Rabbit
Nonhuman Primate
Human
Other Systems
(e.g., Urogenital,
Reproductive,
etc.)
Adrenal hemorrhage (Zaucha et al., 1998)
Ovarian hemorrhage (Zaucha et al., 1998)
Foci of hemorrhage in the kidney (1/13) (Fritz et
al., 1995)
Adrenal hemorrhages (Gleiser et al., 1963), with
extensive hemorrhage of cortex and medulla of
adrenal glands (1/4) (Albrink and Goodlow,
1959); cortical necrosis (2/14) (Vasconcelos et
al., 2003); and extravasation of blood in the
cortex with thrombi in veins (8/23) (Dalldorf et
al., 1971)
Minimal cortical atrophy, occasionally minimal
cortical hemorrhage in adrenal glands (Grinberg
et al., 2001)
Hemorrhagic thyroiditis (Albrink et al., 1960)
Periovarian or peritesticular congestion and/or
hemorrhages (Twenhafel et al., 2007)
Ovarian hemorrhage and necrosis (1/14)
(Vasconcelos et al., 2003)
Retroperitoneal hemorrhages (Gleiser et al.,
1963)
Laryngeal inflammation and edema (1/14)
(Vasconcelos et al., 2003)
C-6
-------
Bibliography
Abramova, F. A., L. M. Grinberg, O. V. Yampolskaya and D. H. Walker (1993). Pathology of
Inhalational Anthrax in 42 Cases from the Sverdlovsk Outbreak of 1979. Proceedings of
the National Academy of Sciences of the United States of America 90(6): 2291-2294.
Albrink, W. S., S. M. Brooks, R. E. Biron and M. Kopel (1960). Human Inhalation Anthrax. A
Report of Three Fatal Cases. The American Journal of Pathology 36(4): 457-471.
Albrink, W. S. and R. J. Goodlow (1959). Experimental Inhalation Anthrax in the Chimpanzee.
The American Journal of Pathology 35(5): 1055-1065.
Barakat, L. A., H. L. Quentzel, J. A. Jernigan, D. L. Kirschke, K. Griffith, S. M. Spear, K.
Kelley, D. Barden, D. Mayo, D. S. Stephens, T. Popovic, C. Marston, S. R. Zaki, J.
Guarner, W. J. Shieh, H. W. Carver, 2nd, R. F. Meyer, D. L. Swerdlow, E. E. Mast, J. L.
Hadler and Anthrax Bioterrorism Investigation Team (2002). Fatal Inhalational Anthrax
in a 94-Year-Old Connecticut Woman. Journal of the American Medical Association
287(7): 863-868.
Dalldorf, F. G., A. F. Kaufman and P. S. Brachman (1971). Woolsorters' Disease. An
Experimental Model. Archives of Pathology 92(6): 418-426.
Fritz, D. L., N. K. Jaax, W. B. Lawrence, K. J. Davis, M. L. Pitt, J. W. Ezzell and A. M.
Friedlander (1995). Pathology of Experimental Inhalation Anthrax in the Rhesus
Monkey. Laboratory Investigation 73(5): 691-702.
Gill, J. R. and J. Melinek (2002). Inhalation of Anthrax: Gross Autopsy Findings. Archives of
Pathology and Laboratory Medicine 126(8): 993-994.
Gleiser, C. A., C. C. Berdjis, H. A. Hartman and W. S. Gochenour, Jr. (1963). Pathology of
Experimental Respiratory Anthrax in Macaca mulatta. British Journal of Experimental
Pathology 44: 416-426.
Grinberg, L. M., F. A. Abramova, O. V. Yampolskaya, D. H. Walker and J. H. Smith (2001).
Quantitative Pathology of Inhalational Anthrax I: Quantitative Microscopic Findings.
Modern Pathology 14(5): 482-495.
Guarner, J. and C. del Rio (2011). Pathology, Diagnosis, and Treatment of Anthrax in Humans.
Bacillus anthracis and Anthrax. N. H. Bergman, (pp. 251-268). Hoboken, New Jersey:
John Wiley & Sons, Inc.
Guarner, J., J. A. Jernigan, W.-J. Shieh, K. Tatti, L. M. Flannagan, D. S. Stephens, T. Popovic,
D. A. Ashford, B. A. Perkins, S. R. Zaki and the Inhalational Anthrax Pathology Working
Group (2003). Pathology and Pathogenesis of Bioterrorism-Related Inhalational Anthrax.
The American Journal of Pathology 163(2): 701-709.
D-l
-------
Henning, L. N., J. E. Comer, G. V. Stark, B. D. Ray, K. P. Tordoff, K. A. B. Knostman and G. T.
Meister (2012). Development of an Inhalational Bacillus anthracis Exposure Therapeutic
Model in Cynomolgus Macaques. Clinical and Vaccine Immunology 19(11): 1765-1775.
Inglesby, T. V., T. O'Toole, D. A. Henderson, J. G. Bartlett, M. S. Ascher, E. Eitzen, A. M.
Friedlander, J. Gerberding, J. Hauer, J. M. Hughes, J. McDade, M. T. Osterholm, G.
Parker, T. M. Perl, P. K. Russell, and K. Tonat for the Working Group on Civilian
Biodefense (2002). Anthrax as a Biological Weapon, 2002. Updated Recommendations
for Management. Journal of the American Medical Association 287(17): 2236-2252.
Jernigan, J. A., D. S. Stephens, D. A. Ashford, C. Omenaca, M. S. Topiel, M. Galbraith, M.
Tapper, T. L. Fisk, S. Zaki, T. Popovic, R. F. Meyer, C. P. Quinn, S. A. Harper, S. K.
Fridkin, J. J. Sejvar, C. W. Shepard, M. McConnell, J. Guarner, W.-J. Shieh, J. M.
Malecki, J. L. Gerberding, J. M. Hughes, B. A. Perkins and Members of the Anthrax
Bioterrorism Investigation Team (2001). Bioterrorism-Related Inhalational Anthrax: The
First 10 Cases Reported in the United States. Emerging Infectious Diseases 7(6): 933-
944.
LaForce, F. M., F. H. Bumford, J. C. Feeley, S. L. Stokes and D. B. Snow (1969). Epidemiologic
Study of a Fatal Case of Inhalation Anthrax. Archives of Environmental Health 18(5):
798-805.
Lawrence, W. S., J. R. Marshall, D. L. Zavala, L. E. Weaver, W. B. Baze, S. T. Moen, E. B.
Whorton, R. L. Gourley and J. W. Peterson (2011). Hemodynamic Effects of Anthrax
Toxins in the Rabbit Model and the Cardiac Pathology Induced by Lethal Toxin. Toxins
3(6): 721-736.
Lever, M. S., A. J. Stagg, M. Nelson, P. Pearce, D. J. Stevens, E. A. M. Scott, A. J. H. Simpson
and M. J. Fulop (2008). Experimental Respiratory Anthrax Infection in the Common
Marmoset (Callithrix jacchus). International Journal of Experimental Pathology 89(3):
171-179.
Lovchik, J. A., M. Drysdale, T. M. Koehler, J. A. Hutt and C. R. Lyons (2012). Expression of
Either Lethal Toxin or Edema Toxin by Bacillus anthracis Is Sufficient for Virulence in a
Rabbit Model of Inhalational Anthrax. Infection and Immunity 80(7): 2414-2425.
Middleton, G. K. and A. C. Standen (1961). The Electrocardiogram in Fatal Anthrax Bacteremia.
The Journal of Infectious Diseases 108: 85-89.
Mina, B., J. P. Dym, F. Kuepper, R. Tso, C. Arrastia, I. Kaplounova, H. Faraj, A. Kwapniewski,
C. M. Krol, M. Grosser, J. Glick, S. Fochios, A. Remolina, L. Vasovic, J. Moses, T.
Robin, M. DeVita and M. L. Tapper (2002). Fatal Inhalational Anthrax with Unknown
Source of Exposure in a 61-Year-Old Woman in New York City. Journal of the
American Medical Association 287(7): 858-862.
Peterson, J. W., J. E. Comer, W. B. Baze, D. M. Noffsinger, A. Wenglikowski, K. G. Walberg, J.
Hardcastle, J. Pawlik, K. Bush, J. Taormina, S. Moen, J. Thomas, B. M. Chatuev, L.
Sower, A. K. Chopra, L. R. Stanberry, R. Sawada, W. W. Scholz and J. Sircar (2007).
D-2
-------
Human Monoclonal Antibody AVP-21D9 to Protective Antigen Reduces Dissemination
of the Bacillus anthracis Ames Strain from the Lungs in a Rabbit Model. Infection and
Immunity 75(7): 3414-3424.
Plotkin, S. A., P. S. Brachman, M. Utell, F. H. Bumford and M. M. Atchison (2002). An
Epidemic of Inhalation Anthrax, the First in the Twentieth Century: I. Clinical Features.
(Reprint). The American Journal of Medicine 112(1): 4-12; Discussion 12-13.
Suffin, S. C., W. H. Carnes and A. F. Kaufmann (1978). Inhalation Anthrax in a Home
Craftsman. Human Pathology 9(5): 594-597.
Twenhafel, N. A., E. Leffel and M. L. M. Pitt (2007). Pathology of Inhalational Anthrax
Infection in the African Green Monkey. Veterinary Pathology 44: 716-721.
U.S. Environmental Protection Agency (2011). Acute Low Dose Bacillus anthracis Ames
Inhalation Exposures in the Rabbit. Cincinnati, OH: National Homeland Security
Research Center. U.S. Environmental Protection Agency. EPA/600/R-11/075.
U.S. Environmental Protection Agency (2012). Multiple Daily Low-Dose Bacillus anthracis
Ames Inhalation Exposures in the Rabbit. Washington, DC: Office of Research and
Development, National Homeland Security Research Center. U.S. Environmental
Protection Agency. EPA/600/R-11/145.
Vasconcelos, D., R. Barnewall, M. Babin, R. Hunt, J. Estep, C. Nielsen, R. Carnes and J. Carney
(2003). Pathology of Inhalation Anthrax in Cynomolgus Monkeys (Macaca fascicularis).
Laboratory Investigation 83(8): 1201-1209.
Yee, S. B., J. M. Hatkin, D. N. Dyer, S. A. Orr and M. L. M. Pitt (2010). Aerosolized Bacillus
anthracis Infection in New Zealand White Rabbits: Natural History and Intravenous
Levofloxacin Treatment. Comparative Medicine 60(6): 461-468.
Zaucha, G. M., M. L. M. Pitt, J. Estep, B. E. Ivins and A. M. Friedlander (1998). The Pathology
of Experimental Anthrax in Rabbits Exposed by Inhalation and Subcutaneous
Inoculation. Archives of Pathology and Laboratory Medicine 122(11): 982-992.
D-3
-------
Appendix D - Bacillus anthracis Dose-Response Data for the Rabbit
Characterized as Supportive Data or Additional Data
This appendix identifies and reviews the dose-response data sets for the rabbit categorized as
Supporting Data or Additional Data for development of an inhalation dose-response relationship
for B. anthracis spores. The literature search and the criteria used to categorize each data set are
provided in the main body of the report (Section 5.3.2). The categorization of the dose-response
data was based on a determination of the suitability of the data set for the development of B.
anthracis dose-response relationships.
Key Studies were defined to be representative of the highest quality dose-response studies that
met criteria for selection during the literature search. Key Studies identified for the rabbit are
provided in the main body of the report (Section 5.4.2.4). Supporting Studies had identifiable
limitations in assessment quality indicators relative to Key Studies, yet were found to have
potential in bounding the dose-response relationship(s) as described by Key Studies. As noted
previously, Additional Data were defined by missing data points critical to assessing dose-
response relationships (e.g., original dose and response data set) or study design elements that
limit utility for development of low-dose dose-response relationships. As a result, their utility in
dose-response analysis may be limited to providing corroborative support for higher quality data.
Supporting Studies
No single dose-response data for the rabbit were categorized as Supporting Studies.
D-4
-------
Additional Data
Table D-l identifies the single dose dose-response data categorized as Additional Data for the
rabbit. Studies are presented in alphabetical order by the first study author. The most cited rabbit
LD50 value of 1.05 x 105 originated from Zaucha et al. (1998) study, though the original dose-
response data set was not published until Gutting et al. (2013) (Table D-l). The Zaucha et al.
(1998) LD50 value is based on a challenge of 50 animals with mean group doses of 98 to 713,000
spores (Gutting et al., 2013). The Zaucha et al. (1998) value has been cited directly or others
have reported values that differ only by varying adjustments in the number of significant figures
(Table D-l). The Zaucha et al. (1998) study was categorized as Additional Data due to the lack
of response data in the range between 1% and 49%. Particle size data were not associated with
the study exposures for which the LD50 value was derived, and the inhalation rate was assumed
to be determined via plethysmography but prior to the actual aerosol challenge. The dose spacing
and the lack of responses between 0 and 50% lethality are problematic because there are
insufficient data to differentiate between possible mathematical dose-response models based on
the fit to the observable data. Given the interest in the low-dose region of the B. anthracis dose-
response relationship, it is important to select the mathematical model appropriately to maximize
the reliability of a low dose extrapolation.
One seemingly outlier value of 600,000 single spore particles (Barnes, 1947) was identified as an
inhaled dose. Additional LD50 values were identified that were derived from intranasal (Peterson
et al., 2006; Weiss et al., 2006; Peterson et al., 2007) or bronchoscopic (Lovchik et al., 2012)
administration. However, these values are not directly comparable to inhaled LD50 values absent
evaluation of potential modifications to ensure dosimetric equivalence to an inhaled dose metric.
D-5
-------
Table D-l. Single Dose Additional Data for the Rabbit
Study Citation,
Empirical Model Parameters and/or
LDso or Other Modeled Values,*
Outputs
Rabbit Breed, and
Strain(s)
Barnes (1947)
Gutting et al. (2013)
LD50 = 600,000 single spore particles
Unspecified rabbit
Note: Analysis combined New Zealand white
Unknown strain
rabbit dose-response data sets reported in
Lovchik et al. (2012)
Zaucha et al. (1998), U.S. Environmental
Bronchoscopic dose LD50 = 103 98 spores
Protection Agency (2011), and previously
SE (logio) =±0.19
unpublished data
New Zealand white rabbit
Ames strain
Exponential model
Peterson et al. (2006)
k = 7.223 x 10"6
Intranasal LD50 = 1 x 105 CFU
Exponential model predicted attack rate (i.e.,
Unspecified rabbit
probability of disease for given dose) for 10
Peterson et al. (2007)
spores = 7.22 x 10"5
Intranasal LD50 = 1.125 x 105 CFU
Dwarf Dutch-belted rabbit
Competing risks model
Ames strain
X
- 6 605 x 10"6
Weiss et al. (2006)
(A + 0)
ATCC 14578 (Vollum) strain intranasal dose LD50 = 3 x
105 spores
Competing risks predicted attack rate = 6.61
ATCC 6605 strain intranasal dose LD50 = 2 * 104 spores
X 10"5
New Zealand white rabbit
Zaucha et al. (1998)
LDso = 105,000 CFU
LD99 = 136,000 CFU
New Zealand white rabbit
Ames strain
Dose-response data set published in Gutting et al. (2013)
Note: This LD50 value is the most commonly cited value
after adjusting for differing significant figures.
Fellows et al. (2001) LD50 = 105 spores
Little et al. (2004) LD50 = 1.1 x 105 spores
Little et al. (2006) LD50 = 1.1 x 105 spores
Pitt et al. (2001) LD50 =1.1 x 105 spores
* Inhaled dose metric unless otherwise noted
X - hazard rate, risk per unit time that spore will germinate
0 - clearance rate, hazard rate, risk per unit time that an ungerminated spore will be cleared from lung 6
ATCC - American Type Culture Collection
CFU - colony forming unit(s)
k - fitted parameter, potency estimate in exponential dose-response model
LD50 - median lethality value
SE - standard error
D-6
-------
Bibliography
Barnes, J. M. (1947). The Development of Anthrax Following the Administration of Spores by
Inhalation. British Journal of Experimental Pathology 28(6): 385-394.
Fellows, P. F., M. K. Linscott, B. E. Ivins, M. L. M. Pitt, C. A. Rossi, P. H. Gibbs and A. M.
Friedlander (2001). Efficacy of a Human Anthrax Vaccine in Guinea Pigs, Rabbits, and
Rhesus Macaques against Challenge by Bacillus anthracis Isolates of Diverse
Geographical Origin. Vaccine 19(23-24): 3241-3247.
Gutting, B. W., D. Marchette, R. Sherwood, G. A. Andrews, A. Director-Myska, S. R. Channel,
D. Wolfe, A. E. Berger, R. S. Mackie, B. J. Watson and A. Rukhin (2013). Modeling
Low-Dose Mortality and Disease Incubation Period of Inhalational Anthrax in the Rabbit.
Journal of Theoretical Biology 329: 20-31.
Little, S. F., B. E. Ivins, P. F. Fellows, M. L. M. Pitt, S. L. W. Norris and G. P. Andrews (2004).
Defining a Serological Correlate of Protection in Rabbits for a Recombinant Anthrax
Vaccine. Vaccine 22(3-4): 422-430.
Little, S. F., B. E. Ivins, W. M. Webster, P. F. Fellows, M. L. M. Pitt, S. L. W. Norris and G. P.
Andrews (2006). Duration of Protection of Rabbits after Vaccination with Bacillus
anthracis Recombinant Protective Antigen Vaccine. Vaccine 24(14): 2530-2536.
Lovchik, J. A., M. Drysdale, T. M. Koehler, J. A. Hutt and C. R. Lyons (2012). Expression of
Either Lethal Toxin or Edema Toxin by Bacillus anthracis Is Sufficient for Virulence in a
Rabbit Model of Inhalational Anthrax. Infection and Immunity 80(7): 2414-2425.
Peterson, J. W., J. E. Comer, W. B. Baze, D. M. Noffsinger, A. Wenglikowski, K. G. Walberg, J.
Hardcastle, J. Pawlik, K. Bush, J. Taormina, S. Moen, J. Thomas, B. M. Chatuev, L.
Sower, A. K. Chopra, L. R. Stanberry, R. Sawada, W. W. Scholz and J. Sircar (2007).
Human Monoclonal Antibody AVP-21D9 to Protective Antigen Reduces Dissemination
of the Bacillus anthracis Ames Strain from the Lungs in a Rabbit Model. Infection and
Immunity 75(7): 3414-3424.
Peterson, J. W., J. E. Comer, D. M. Noffsinger, A. Wenglikowski, K. G. Walberg, B. M.
Chatuev, A. K. Chopra, L. R. Stanberry, A. S. Kang, W. W. Scholz and J. Sircar (2006).
Human Monoclonal Anti-Protective Antigen Antibody Completely Protects Rabbits and
Is Synergistic with Ciprofloxacin in Protecting Mice and Guinea Pigs against Inhalation
Anthrax. Infection and Immunity 74(2): 1016-1024.
Pitt, M. L. M., S. F. Little, B. E. Ivins, P. F. Fellows, J. Barth, J. Hewetson, P. Gibbs, M.
Dertzbaugh and A. M. Friedlander (2001). In Vitro Correlate of Immunity in a Rabbit
Model of Inhalational Anthrax. Vaccine 19(32): 4768-4773.
D-7
-------
U.S. Environmental Protection Agency (2011). Acute Low Dose Bacillus anthracis Ames
Inhalation Exposures in the Rabbit. Cincinnati, OH: National Homeland Security
Research Center. U.S. Environmental Protection Agency. EPA/600/R-11/075.
Weiss, S., D. Kobiler, H. Levy, H. Marcus, A. Pass, N. Rothschild and Z. Altboum (2006).
Immunological Correlates for Protection against Intranasal Challenge of Bacillus
anthracis Spores Conferred by a Protective Antigen-Based Vaccine in Rabbits. Infection
and Immunity 74(1): 394-398.
Zaucha, G. M., M. L. M. Pitt, J. Estep, B. E. Ivins and A. M. Friedlander (1998). The Pathology
of Experimental Anthrax in Rabbits Exposed by Inhalation and Subcutaneous
Inoculation. Archives of Pathology and Laboratory Medicine 122(11): 982-992.
D-8
-------
Appendix E - Bacillus anthracis Dose-Response Data for the Nonhuman
Primate Characterized as Supportive Data or Additional Data
The classic Druett et al. (1953) study presents the only data categorized as a Supporting Study
for the nonhuman primate (Table E-l). Druett et al. (1953) aerosol challenged rhesus macaque
monkeys with the M36 strain B. anthracis single spores and 12 |im particles in nine and eight
dose groups of eight monkeys, respectively. This study was identified as a Supporting Study due
to the presence of raw dose-response data, particle size data, presence of low-dose groups, and
sufficient animal numbers for modeling. However, the lack of real-time determination of
inhalation rates was the primary reason that this study was categorized as a Supporting Study.
The Druett et al. (1953) paper was unclear on the length of observation post-challenge but did
identify that the experiments for "each particle size were completed within a period of two to
three weeks." The infection endpoint was not reported.
The inhaled dose LD50 value reported for single spore particles was 53,000 spores (Druett et al.,
1953). Re-analyses of the Druett et al. (1953) data set reported LD50 or equivalent BMD50 values
ranging from 96,800 (Haas, 2002) to approximately 50,000 (Curling et al., 2010; U.S.
Environmental Protection Agency, 2010; Taft and Hines, 2012; Toth et al., 2013) (Table E-2).
The reason for the difference in published LD50 values has been attributed to the two-fold higher
inhalation rate used by Haas (2002) and Bartrand et al. (2008) in lieu of the inhalation value
identified by Druett et al. (1953) (Curling et al., 2010; U.S. Environmental Protection Agency,
2010; Taft and Hines, 2012; Toth et al., 2013).
E-l
-------
Table E-l. Single Dose Supporting Studies for the Nonhuman Primate
Study Citation,
Nonhuman
Primate Species,
and Strain
Supporting Study
Outputs
Reanalysis Studies
Additional Data Outputs
Single Dose
Druett et al. (1953)
Rhesus macaque
(Macaca mulatto)
M36 strain
Environmental
concentration associated
with 50% mortality:
Nt* = 0.045 x 10"6 Single
spores - minutes/L
(Inhaled dose x 10 5
single spores = 0.53
[53,000])
Logio probit slope = 3.19
with intercept of 2.91
using exposure
concentration x 104 as
dose
Environmental
concentration associated
with 50% mortality:
Nt = 0.64 x 10"6 12 nm
Spore particles -
minutes/L*
(Inhaled dose x 105 12
Hm spores = 7.6
[760,000])
Haas (2002)
Exponential model
LD5o = 96,800 single spores
(CI = 70,700 to 136,000)
k = 7.16 x 10"6
(CI = 5.1 x 10"6to 9.8 x 10"6)
Bartrand et al. (2008)
Exponential model
LD5o = 92,000 single spores
(CI = 29,440 to 70.932) [sic]
k = 7.16 xlO"6
Curling et al. (2010)
Exponential model
LD5o = approx. 51,000 spores
1= 1.36 x 10"5
U.S. Environmental
Protection Agency
(2010)
Taft and Hines (2012)
Exponential model
k= 1.44 x 10"5
(CI = 9.81E-6 to 1.9E-5)
BMD5o = 48,000 single spores
BMDL50 = 37,000 single spores
BMD10 = 7,300 single spores
BMDL10 = 5,600 single spores
BMDi = 700 single spores
BMDLi = 540 single spores
Toth et al. (2013)
Exponential model
r= 1.43 x 10-5
ID50 = 48,000 single spores
ID10 = 7,400 single spores
IDi = 700 single spores
*Druett et al. (1953) used the term "dosage" (Nt) to describe the product of environmental concentration and period
of exposure (e.g., Nt x 10 6 = 0.168); for ease in reading the table, this term has been recorded as Nt (e.g., 0.168 x
10"6), all exposures were of one minute duration
BMDx - benchmark dose for response in x% of individuals
BMDLx - the 95% lower statistical confidence limit of the BMD when the 95% lower confidence limit is applied to
the estimated slope parameter value
CI - 95% confidence interval
IDX - infectious dose for x percent exposed, Toth et al. (2013) assumed ID5o = LD5o
k, X, or r - fitted parameter, potency estimate in exponential dose-response model
LD5o - median lethality value
Nt - dosage
E-2
-------
A significant amount of nonhuman primate dose-response data was categorized as Additional
Data (Table E-2). The majority of these data were in the form of reported inhaled dose LD50
values or ranges with little or no accompanying data. One exception was the Young et al. (1946)
LD50 value of 200,000 that utilized an environmental concentration dose metric. The remaining
data for LD50 values or ranges in Table E-2 tended to group into two main ranges. The low end
of the range was between approximately 4,000 and 11,000 CFU or spores (Brachman et al.,
1960; Glassman, 1966; Peters and Hartley, 2002; Estep et al., 2003; Leffel and Pitt, 2006; Rossi
et al., 2008) and a high-end range was between approximately 50,000 to 62,000 CFU or spores
(Henderson et al., 1956; Ivins et al., 1996; Vasconcelos et al., 2003; Coleman et al., 2008). A
range of historical LD50 values for rhesus monkeys (30,000 to 172,000 CFU) was also identified
by Leffel and Pitt (2006).
However, the identified LD50 values should be evaluated carefully prior to use for informing risk
assessment. It is important to recognize that most values were derived from studies with the
primary purpose of evaluating pathology or medical countermeasures; the LD50 values were
generated with study designs that did not explicitly evaluate statistical considerations regarding
animal and dose range to generate a representative median value.
With the exception of the Vasconcelos et al. (2003) LD50 value, the remaining identified values
in the 50,000 to 62,000 CFU range were cited as a personal communication or unpublished data
from an author associated with the USAMRIID laboratories (e.g., Ivins et al. (1996),
Vasconcelos et al. (2003), Coleman et al. (2008)) or were directly cited by an author with
USAMRIID affiliation (e.g., Henderson et al. [1956] in Friedlander et al. [1993]). It is possible
that multiple published citations of approximately the same LD50 value may not represent
E-3
-------
multiple independent studies that corroborate the identified value, but may be the same study or a
limited number of studies repeatedly cited.
Table E-2. Single Dose Additional Data for the Nonhuman Primate
Study and LDso Value,* Nonhuman
Primate Species, and Strain
Study for Data Set, Nonhuman Primate
Species, Reanalysis Study, Model
Parameters or Outputs, and Strain
Other Data, Nonhuman
Primate Species, and
Strain
Single Dose
Brachman et al. (1960)
LD5o = 6,000 sporest
Unspecified NHP
Goat hair mill aerosol, unknown
strain(s)
Coleman et al. (2008)
59,000 unspecified unitst
Rhesus monkey
(Macaca mulatto)
Unknown strain
Estep et al. (2003)
Ames strain LD50 = 10,900 CFU
(Fieller's CI = 1,320 to 241,000)
Vollum strain LD50 = 6,750 CFU
(Fieller's CI = 21 to 116,000)
Rhesus monkey
(Macaca mulatto)
Glassman (1966)
Cynomolgus monkey
(Macaca fascicularis)
Reanalyzed by Peters and Hartley (2002)
using the reported probit slope = 0.67 per
logio dose spores and LD50 = 4,100 spores,
each value rounded to two significant figures
LD10 = 50 spores
LD2 = 4 spores
LDi = 1 spore
Unknown strain
Barnewall et al. (2001)
Rhesus monkey
(Macaca mulatto)
Reanalyzed by U.S. Environmental Protection
Agency (2010) and Taft and Hines (2012)
BMD50 = 10,000 CFU
BMDL50 = 4,900 CFU
BMD10 = 1,100 CFU
BMDL10 = 550 CFU
Unknown strain
Janssen (1955a), Janssen (1955b), and Janssen
(1955c)
Original studies did not identify nonhuman
primate species, assumed to be Macaca
mulatto by Taft and Hines (2012)
Reanalyzed by U.S. Environmental Protection
Agency (2010) and Taft and Hines (2012)
Albrink and Goodlow
(1959)
Chimpanzee
(Pan troglodytes
[,Schwarz] and Pan
troglodytes troglodytes)
Single dose administered
to 4 animals:
Melvin: 32,800 inhaled
viable spores - survived
John: 34,350 inhaled
viable spores - survived
Grove: 39,700 Inhaled
viable spores - died
Bill: 66,500 inhaled
viable spores - died
Vollum rB strain
E-4
-------
Study and LDso Value,* Nonhuman
Primate Species, and Strain
Study for Data Set, Nonhuman Primate
Species, Reanalysis Study, Model
Parameters or Outputs, and Strain
Other Data, Nonhuman
Primate Species, and
Strain
Henderson et al. (1956)
LD5o = approximately 50,000 spores
(Originally reported three individual
results as 4 LNt50 -2.14 x 105 spores,
8 LNt50 ~ 3.9 x 105 spores, and 4
LNt50 ~ 2 x 105 spores)
Rhesus monkey
(Macaca mulatto)
M36 strain
BMD50 = 660 CFU
BMDLso = 530 CFU
BMD10 = 180 CFU
BMDL10 = 150 CFU
Strain not identified in original study reports,
but Vollum identified in use by U.S.
Department of Defense researchers at that
time by U.S. Environmental Protection
Agency (2010) and Taft and Hines (2012)
Glassman (1966)
LD5o= 4,130 sporest
CI = 1,980 to 8,630 spores
Probit slope = 0.669 probits/log dose
CI = 0.520 to 0.818
Cynomolgus monkey
(Macaca fascicularis)
Unknown strain
Ivins et al. (1996)
Rhesus monkey
(Macaca mulatto)
LD50 = 55,000 CFUt
Ames strain
Leffel and Pitt (2006)
Historically reported range of LD50
values for unspecified strain: 30,000 to
172,000 CFU
Rhesus monkey
(Macaca mulatto)
LD50 values from head-to-head test of
same Ames spore lot:
Rhesus monkey = 7,200 CFUt
African green monkey = 8,300 CFU
Peters and Hartley (2002)
LD50 = approximately 8,000 CFUt
Cynomolgus monkey
(Macaca fascicularis)
Unknown strain
Rossi et al. (2008)
LD50= 11,000 CFUt
CI = 2.9 x 103 to 8.1 x 104
African green monkey
(Chlorocebus aethiops)
Ames strain
Sharp and Roberts (2006)
LD50 value = c. 5,000 to 8,000 CFUt
Cynomolgus monkey
(Macaca fascicularis)
Unknown strain
E-5
-------
Study and LDso Value,* Nonhuman
Primate Species, and Strain
Study for Data Set, Nonhuman Primate
Species, Reanalysis Study, Model
Parameters or Outputs, and Strain
Other Data, Nonhuman
Primate Species, and
Strain
Vasconcelos et al. (2003)
LD50 = 61,800 CFU
95% CI = 34,800 to 110,000 CFU
Probit slope = 4.21
Cynomolgus monkey
(Macaca fascicularis)
Ames strain
Young et al. (1946)
LD50 = 20 x 10"4 spores
(Note: Dose metric for LD50 value is an
environmental concentration for a 5-
minute exposure)
Unspecified NHP
Detrick 25 strain
Twenhafel et al. (2007)
African Green Monkey
(Chlorocebus aethiops)
Data describing low-dose
lethality at the lowest
tested dose of 204 CFU
Ames strain
* Inhaled dose unless otherwise noted
t LD5o value cited from unpublished data or personal communication
BMDx - benchmark dose for response in x% of individuals
BMDLx - the 95% lower statistical confidence limit of the BMD when the 95% lower confidence limit is applied to
the estimated slope parameter value
CFU - colony forming unit(s)
CI - 95% confidence interval
LDX - lethality value for x% of individuals
NHP - nonhuman primate
E-6
-------
Bibliography
Albrink, W. S. and R. J. Goodlow (1959). Experimental Inhalation Anthrax in the Chimpanzee.
The American Journal of Pathology 35(5): 1055-1065.
Barnewall, R., J. Estep and R. DeBell (2001). Inhalation Median Lethal Dose (LD50)
Determinations in Rhesus Monkeys Exposed to Bacillus anthracis (For Official Use
Only). Prepared for Defense Intelligence Agency. Battelle Memorial Institute, Study
Number CG463810D.
Bartrand, T. A., M. H. Weir and C. N. Haas (2008). Dose-Response Models for Inhalation of
Bacillus anthracis Spores: Interspecies Comparisons. Risk Analysis 28(4): 1115-1124.
Brachman, P. S., S. A. Plotkin, F. H. Bumford and M. M. Atchison (1960). An Epidemic of
Inhalation Anthrax: The First in the Twentieth Century. II. Epidemiology. American
Journal of Hygiene 72(1): 6-23.
Coleman, M. E., B. Thran, S. S. Morse, M. Hugh-Jones and S. Massulik (2008). Inhalation
Anthrax: Dose Response and Risk Analysis. Biosecurity and Bioterrorism: Biodefense
Strategy, Practice, and Science 6(2): 147-160.
Curling, C. A., J. K. Burr, L. Danakian, D. S. Disraelly, L. A. LaViolet, T. J. Walsh and R. A.
Zirkle (2010). Technical Reference Manual: NATO Planning Guide for the Estimation of
Chemical, Biological, Radiological, and Nuclear (CBRN) Casualities, Allied Medical
Publication-8(C). Alexandria, VA: Institute for Defense Analyses. IDA-D-4082.
Druett, H. A., D. W. Henderson, L. Packman and S. Peacock (1953). Studies on Respiratory
Infection: I. The Influence of Particle Size on Respiratory Infection with Anthrax Spores.
The Journal of Hygiene (London) 51(3): 359-371.
Estep, J. E., R. Barnewall, R. DeBell and N. Niemuth (2003). Inhalation Median Lethal Doses of
Bacillus anthracis, Ames, and Vollum Strains in the Rhesus Monkey. Toxiological
Sciences 72(S-1): 161-162.
Friedlander, A. M., S. L. Welkos, M. L. M. Pitt, J. W. Ezzell, P. L. Worsham, K. J. Rose, B. E.
Ivins, J. R. Lowe, G. B. Howe, P. Mikesell and W. B. Lawrence (1993). Postexposure
Prophylaxis against Experimental Inhalation Anthrax. Journal of Infectious Diseases
167(5): 1239-1243.
E-7
-------
Glassman, H. N. (1966). Industrial Inhalation Anthrax - Discussion. Bacteriological Reviews
30(3): 657-659.
Haas, C. N. (2002). On the Risk of Mortality to Primates Exposed to Anthrax Spores. Risk
Analysis 22(2): 189-193.
Henderson, D. W., S. Peacock and F. C. Belton (1956). Observations on the Prophylaxis of
Experimental Pulmonary Anthrax in the Monkey. Journal of Hygiene (London) 54(1):
28-36.
Ivins, B. E., P. F. Fellows, M. L. M. Pitt, J. E. Estep, S. L. Welkos, P. L. Worsham and A. M.
Friedlander (1996). Efficacy of a Standard Human Anthrax Vaccine against Bacillus
anthracis Aerosol Spore Challenge in Rhesus Monkeys. Salisbury Medical Bulletin
Special Supplement Number 87: 125-126.
Janssen, R. E., Jr. (1955a). Operation "Jungle Boy. " Trial Report. BWAL BW 6a-3-54: Dugway
Proving Ground, UT. AD 596073, CBRNIAC No. CB022712. Distribution Limited to
U.S. Government Agencies Only.
Janssen, R. E., Jr. (1955b). Operation "Jungle Boy. " Trial Report. BWAL BW 6a-4-54: Dugway
Proving Ground, UT. AD596081, CBRNIAC No. CB022713. Distribution Limited to
U.S. Governmental Agencies Only.
Janssen, R. E., Jr. (1955c). Operation "Jungle Boy. " Trial Report. BWAL BW 6a-5-54 Dugway
Proving Ground, UT. AD596083, CBRNIAC No. CB022714. Distribution Limited to
U.S. Governmental Agencies Only.
Leffel, E. and L. M. Pitt (2006). Chapter 6. Anthrax. Biodefense: Research Methodology and
Animal Models, (pp. 77-94) J. R. Swearengen. CRC Press.
Peters, C. J. and D. M. Hartley (2002). Anthrax Inhalation and Lethal Human Infection. The
Lancet 359(9307): 710-711.
Rossi, C. A., M. Ulrich, S. Norris, D. S. Reed, L. M. Pitt and E. K. Leffel (2008). Identification
of a Surrogate Marker for Infection in the African Green Monkey Model of Inhalation
Anthrax. Infection and Immunity 76(12): 5790-5801.
Sharp, R. J. and A. G. Roberts (2006). Anthrax: The Challenges for Decontamination. Journal of
Chemical Technology and Biotechnology 81(10): 1612-1625.
E-8
-------
Taft, S. C. and S. A. Hines (2012). Benchmark Dose Analysis for Bacillus anthracis Inhalation
Exposures in the Nonhuman Primate. Risk Analysis 32(10): 1750-1768.
Toth, D. J. A., A. V. Gundlapalli, W. A. Schell, K. Bulmahn, T. E. Walton, C. W. Woods, C.
Coghill, F. Gallegos, M. H. Samore and F. R. Adler (2013). Quantitative Models of the
Dose-Response and Time Course of Inhalational Anthrax in Humans. PloS Pathogens
9(8): el003555.
Twenhafel, N. A., E. Leffel and M. L. M. Pitt (2007). Pathology of Inhalational Anthrax
Infection in the African Green Monkey. Veterinary Pathology 44: 716-721.
U.S. Environmental Protection Agency (2010). Benchmark Dose Analysis for Bacillus anthracis
Inhalation Exposures in the Nonhuman Primate and Application to Risk-Based Decision
Making. Washington, DC: National Homeland Security Research Center. U.S.
Environmental Protection Agency. EPA 600/R-10/138.
Vasconcelos, D., R. Barnewall, M. Babin, R. Hunt, J. Estep, C. Nielsen, R. Carnes and J. Carney
(2003). Pathology of Inhalation Anthrax in Cynomolgus Monkeys (Macaca fascicularis).
Laboratory Investigation 83(8): 1201-1209.
Young, G. A., M. R. Zelle and R. E. Lincoln (1946). Respiratory Pathogenicity of Bacillus
anthracis Spores I. Methods of Study and Observations on Pathogenesis. Journal of
Infectious Diseases 79(3): 233-246.
E-9
-------
Appendix F - Conducting Benchmark Dose Analysis for Microbial
Pathogens
Introduction
Benchmark dose (BMD) analysis empirically fits models to dose-response data and identifies the
dose associated with a specific response level (U.S. Environmental Protection Agency, 2012).
The following section describes the process and special considerations for the use of BMD
modeling with microbial pathogens. While there is a focus on the use of EPA's Benchmark
Dose Software (BMDS) in some of the examples, the process description is applicable to other
software capable of conducting the empirical modeling and reporting the necessary outputs.
Conducting the BMD Analysis
BMD analysis is conducted using the following general steps:
• Evaluate the data set,
• Fit selected dose-response models,
• Identify the best fitting mathematical model(s), and
• Report the modeling results.
The following sections discuss each step in the process and identify potential considerations
when modeling dose-response relationships of microbial pathogens.
Evaluate the Data Set
Prior to use in BMD modeling, the dose-response data should be assessed for the sufficiency of
the data for BMD analysis. This step is distinct from a quality assessment that evaluates the
study design, documentation, and development of the data set. The minimum data set
F-l
-------
requirements for BMD analysis are: (1) a dose-related trend in the assessment endpoint (either
statistical and/or biological significance), (2) a data set with data points between the maximum
response levels in control or higher-level dose groups and no response levels, and (3) typically
more than one dose group (U.S. Environmental Protection Agency, 2012). However, two dose
groups may also be insufficient to evaluate some models based on parameter number and may
affect the ability to evaluate model uncertainty (U.S. Environmental Protection Agency, 2012).
There should be at least as many dose groups as model parameters to estimate mean response
and confidence levels (U.S. Environmental Protection Agency, 2012).
As with all analyses based on curve-fitting, there is a preference for studies that have more dose
groups as well as a graded monotonic response with regard to dose (U.S. Environmental
Protection Agency, 2015). However, many of the available dose-response data sets for
B. anthracis reported dose-response data, but their original purpose was not derivation of dose-
response data (e.g., pathology studies that also report median lethality [LD50] values). Current
dose-response data sets that are generated for inhalation challenge studies typically use
plethysmographic inhalation data that allow for reporting both individual-specific inhalation
doses and targeted dose group data. In these instances, individual dose-response data can be used
instead of dose group-level data. Additionally, many of these data sets may have limited
coverage below the LD50 value, which limits the lower end of the observable range and may
affect selection of statistically appropriate benchmark response (BMR) values. Accordingly, the
use of these data in empirical model curve-fitting approaches may be associated with higher
levels of uncertainty for lower dose levels than the levels typically found in analyses of chemical
dose-response data sets with better low-dose coverage. This is not to suggest that BMD may not
F-2
-------
be a useful modeling approach for the microbial data sets, but that the uncertainty associated
with the BMD outputs from these data sets should be acknowledged.
Fit Selected Dose-Response Models
The EPA does not advocate use of any specific BMD or curve-fitting software package (U.S.
Environmental Protection Agency, 2012), but recommends that selected software have a
sufficiently documented methodology to evaluate the statistical algorithms used for model fit and
the development of outputs. The BMDS (available from the product web page
(http://www.epa.gov/ncea/bmds/) is one option to conduct BMD. The BMDS can be an
important tool to evaluate commonly used empirical dose-response models for microbial
pathogens, including B. anthracis (Taft and Hines, 2012). The BMDS was originally developed
for chemical agents, but the empirical curve-fitting process employed in BMD has relevance for
microbial agents (Taft and Hines, 2012). U.S. Environmental Protection Agency (2012)
addresses considerations for benchmark dose analysis of chemical agents, but there is a gap in
technical guidance for the use of BMD for microbial dose-response analysis.
A second software with BMD modeling capabilities is the PROAST software package (National
Institute for Public Health and the Environment [RIVM], 2014). PROAST was developed by the
National Institute for Public Health and the Environment (RIVM, The Netherlands) for the
statistical analysis of dose-response and other similarly structured data sets. The software can be
used to fit mathematical models, report goodness of fit (GOF) measures, and generate graphics
(National Institute for Public Health and the Environment [RIVM], 2014). Potential advantages
of PROAST may include the possibility of statistically comparing dose-response relationships
F-3
-------
among subgroups in the data set and greater flexibility in plotting that was used to develop the
BMDS software.
Figure F-l (U.S. Environmental Protection Agency, 2015) shows a decision tree to assist in
conducting the BMD modeling and determining the best fitting model(s). The first two
determinations are selection of the BMR and the dose metric(s) for modeling. Considerations for
the selection of the dose metric were discussed previously in Section 5.3.4 of the main report.
START
BMD - benchmark dose
BMDL - benchmark dose limit
AIC - Akaike Information Criterion
Figure F-l. BMD decision tree from U.S. Environmental Protection Agency (2015).
The BMR is the level of change in the response rate (e.g., a BMR of 10% would be equivalent to
a 10% increase in the response rate of the endpoint of interest) that forms the basis for the
reported BMD value. A BMR value of 10% is identified for chemical hazards and dichotomous
data to standardize reporting of the benchmark dose limit (BMDL) values, but the value is not to
F-4
-------
be interpreted as a default value (U.S. Environmental Protection Agency, 2012). The
determination of a BMR should be based upon the intended use of the BMD outputs, the
statistical features of the data set, and the biological basis of the modeled disease process (U.S.
Environmental Protection Agency, 2012). An identified BMR value, or a range of BMR values,
specific for microbial data to support risk-informed decision-making from BMDS outputs or for
standardized reporting is not available. In chemical dose-response analysis, the reporting of
BMDS outputs for B. anthracis data sets has also included the 10% BMR value for the BMDL
value (e.g., Taft and Hines [2012]). However, the determination of the appropriate BMR values
may require a unique evaluation relative to the values for chemical agents due to the reliance on
lethality endpoints in B. anthracis dose-response data sets, high lethality levels associated with
exposure levels of concern, and limited statistical power of most dose-response data sets. The
identification of the BMR range of values or guidance for their selection is a science policy gap
for microbial dose-response analysis.
A prior analysis using the BMDS and B. anthracis dose-response data sets evaluated the fit of
the data to the following models: the Weibull model, the Weibull model run as exponential (with
the power coefficient fixed as one), probit, loge probit, logistic, loge logistic, Gamma model,
dichotomous Hill, probit-background response, and logistic-background response (Taft and
Hines, 2012). The rationale for evaluation of a diverse group of empirical models was to
minimize the model uncertainty associated with selection of one model and its associated
assumptions (e.g., threshold, nonthreshold) (Taft and Hines, 2012).
When using modeling software for dose-response analysis, care should be taken to identify all
assumptions or default restrictions placed on model parameters (Taft and Hines, 2012). There
F-5
-------
should be sufficient information to allow an individual to recreate the dose-response model
outputs from the input identified data set. For example, the BMDS places the default restriction
on the slope and power terms to ensure that they do not have values greater than or equal to one.
This prevents supralinear behavior in the low-dose region of the dose-response curve (U.S.
Environmental Protection Agency, 2012). Since historically used microbial dose-response
models (e.g., exponential, beta-Poisson) are linear in the low-dose region (Haas et al., 1999), the
identified restrictions on term values are appropriate for microbial pathogens. The BMDS also
includes a suite of models that allows for setting the background incidence to zero (e.g.,
alternative dichotomous models) if an individual lacks this fundamental assumption. This is
appropriate for B. anthracis since it should be assumed that there is no background incidence in
the challenge studies.
Identify the Best Fitting Mathematical Models
There are no differences in the assessment of goodness of fit (GOF) for microbial dose-response
analysis and chemical dose-response analysis. The Chi-square statistical test is used to evaluate
the overall GOF for an individual model (U.S. Environmental Protection Agency, 2012). An
insignificant p-value (p > 0.1) does not allow for the rejection of the null hypothesis (H0) and
indicates that the tested model fits the data. If the estimated BMDs and BMDLs are "sufficiently
close" (as determined by decision-making needs) for models that have acceptable statistical fits
to the data, the model with the lowest Akaike Information Criterion (AIC) value will be
considered to have the best fit (U.S. Environmental Protection Agency, 2012). However, it
should be noted that an AIC comparison should not be made across models with different
restrictions in the slope, power, or background parameters (U.S. Environmental Protection
F-6
-------
Agency, 2012). From this model with the lowest AIC value, the point of departure (POD) can be
determined from the BMDL associated with the selected BMR value. An evaluation of visual fit
as well as scaled residuals near the BMR(s) of interest should also be conducted (U.S.
Environmental Protection Agency, 2012).
The selection of the POD(s) for use in the interspecies extrapolation process can involve
additional steps to focus the model review to ensure that there is adequate statistical fit to the
data and visual fit, especially in the low-dose regions (U.S. Environmental Protection Agency,
2012). Detailed analysis for determination of the POD across multiple suitably fitting models
should be done in consultation with statistical experts (U.S. Environmental Protection Agency,
2012).
Report the Modeling Results
Guidance is available on preferred reporting for BMD outputs that is applicable regardless of the
platform used to conduct the analysis. If using the BMDS, it is recommended that summary
reporting capability provided by the BMDS Wizard be used to facilitate reporting of BMD model
fit and outputs. As with all dose-response modeling, the restriction of any model parameters
(e.g., slope, power) should be clearly identified. If varying dose metrics were generated, the base
assumptions and data used to calculate the dose metric should be clearly identified. For situations
where multiple models exhibit a statistically significant fit, the rationale for model selection
should be transparent and clearly describe the basis for selection.
When colony-based counting methods (e.g., bacterial plate counts) are used for the measurement
of challenge doses for B. anthracis, care must be taken in reporting dose-response outputs. It is
generally recognized that these analytical methods are only precise to two significant digits.
F-7
-------
Accordingly, dose-response model outputs for BMD and BMDL values are reported to an
equivalent number of significant figures.
F-8
-------
Bibliography
Haas, C. N., J. B. Rose and C. P. Gerba (1999). Quantitative Microbial Risk Assessment. New
York: John Wiley & Sons, Inc.
National Institute for Public Health and the Environment (RIVM). (2014). "PROAST (V. 38.9)."
from http://www.rivm.nl/en/Documents and publications/Scientific/Models/PROAST.
Taft, S. C. and S. A. Hines (2012). Benchmark Dose Analysis for Bacillus anthracis Inhalation
Exposures in the Nonhuman Primate. Risk Analysis 32(10): 1750-1768.
U.S. Environmental Protection Agency (2012). Benchmark Dose Technical Guidance.
Washington DC: Office of the Science Adviser, Risk Assessment Forum. U.S.
Environmental Protection Agency. EPA/100/R-12/001.
U.S. Environmental Protection Agency. (2015). "Benchmark Dose Software (BMDS): BMD
Methods." Retrieved May 4, 2015, from
http://www.epa.gov/ncea/bmds/methodology.html.
F-9
-------
vvEPA
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
------- |