Relating Numbers of Foodborne Pathogens to Human Illness: Abstracts
Food Safety Initiative Technical Workshop:
Relating Numbers of Foodborne Pathogens to Human Illness
Introduction | Program | Talks | Abstracts | Qs & As | Slides | Biographies
Tuesday, August
4, 1998
Atrium, Stamp
Student Union
University
of Maryland
College Park,
MD
Abstracts
Introductory
Overview
Wesley R. Long, PhD
Food and Drug Administration
Washington, DC
The President’s Food Safety
Initiative provided for increased funding to support research in risk assessment.
One critical focus of this effort is the development of data, methods and
models to accurately assess dose-response relationships for foodborne pathogenic
microorganisms. A Consortium of federal agencies with risk management responsibilities
for food safety has been formed as a result of The Initiative. This workshop
is a result of the collaboration of representatives of these federal agencies,
and seeks to identify methods and approaches that may prove useful in addressing
microbial dose-response modeling issues.
Uses
of Dose-Response Information in Regulating Food Additives and Chemical
and Microbial Contaminants
James D. Wilson, PhD
Resources for the Future
Washington, DC
Human health risk assessment for chemicals can include a "dose-response assessment" as one of its parts. Commonly, however, chemical risk assessors only identify an intake rate that poses a negligible risk. They do this using some variant of a general procedure for evaluating data from animal studies. This negligible risk intake rate is named differently by different regulatory agencies: FDA uses the term "Acceptable Daily Intake." To judge acceptability of food additives, for example, this ADI is compared with a predicted or estimated intake derived using "reasonable worst case" assumptions. This represents standard practice for food additives and pesticide residues -- substances intentionally used in food production, and regulated only on safety as used, and for some chemical contaminants as well. It has been used very successfully for more than fifty years.
On the other hand, for many "contaminants" -- both chemical and microbial -- more complex risk assessment practices are required. The regulatory agencies, FDA and FSIS, must weigh effects of their regulations on the quality and abundance of the food supply. In theory, full knowledge of the relation between exposure and consequence is needed. In practice, a simple safety test can often supply enough information for decision making; these "reasonable worst case" assessments often show that a practical contaminant level is safe. When this result is not obtained, as in the case of aflatoxin, a more complicated analysis must be done.
Microbial contaminants offer
a yet more complex regulatory regime. The analysis must include the estimation
of microbial growth rates under plausible conditions of food processing
and storage. Because this task at present is difficult due to lack of data,
the agencies are using a risk-management approach that seeks to identify
places in the processing chain that either lead to microbial contamination
or allow microbial growth. Empirical ways to inhibit this contamination
or growth are sought. At its present state of development this "HACCP"
process does not use microbial dose-response information. It uses performance
standards -- numbers of microbes per unit area of foodstuff -- which are
practical standards rather than scientific. Eventually, however, these
standards will have to be justified; if nothing else, world trade regulations
will require their doing so. And THEN regulators will need precise scientific
knowledge of how pathogenic microorganisms produce adverse effects in people.
Limitations
of Current Dose-Response Data and Models: Information Needs of Microbiological
Risk Assessors
Margaret Coleman
Food Safety Inspection
Service, USDA
Washington, DC
Some serious limitations
exist regarding the use of the available human data from clinical trials
for dose-response modeling to support microbial risk assessment. Very few
strains have been administered in human clinical trials. Dose groups are
generally small, commonly 6 volunteers but including as few as 4 and as
many as 193. However, these data do provide an advantage that chemical
dose-response modelers rarely possess: human endpoints that don't require
inter-species conversions. Microbial dose-response modelers may choose
to model infection/colonization (infectivity), illness (pathogenicity),
and severity (virulence). Three major issues are introduced: 1) selection
criteria for surrogates in the absence of specific data; 2) low dose extrapolation,
also an issue in chemical risk assessment; and 3) variability in all phases
of the disease triangle (host, pathogen, and environment, with interactions).
The strongest criteria for selection of an appropriate surrogate pathogen
may be similarity of mechanisms of pathogenesis, based on genetic studies
of pathogenicity islands and virulence gene expression. Particular emphasis
is given to shigellosis, non-typhoid salmonellosis, and illness from E.
coli O157:H7. Some implications of selection of model form (Beta-Poisson,
Logistic, and Gompertz) are explored. The possibility of thresholds for
the healthy adult volunteer populations modeled in the human clinical studies
may be reality for some pathogens. Biological plausibility of the assumption
of linearity in low dose extrapolation for all pathogens will be examined
in this presentation. The importance of accounting for variability in dose-response
modeling is crucial. Examples of host, pathogen strain, and environmental
variability are presented. Differential susceptibility of the human population
is addressed, along with procedures to apply mechanistic information from
animal studies to shift the dose-response curve for healthy adults to reflect
susceptibility due to previous antibiotic administration. Epidemiological
studies will be important as verification of dose-response models. Formal
dose-reconstruction studies of outbreaks of foodborne illness may be very
useful to dose-response modelers.
Mechanisms
of Pathogenesis of Salmonellae: Linking in vitro, Animal, and Human
Studies
James M. Slauch, PhD
Department of Microbiology,
University of Illinois
Urbana, IL
Salmonellae are enteroinvasive pathogens that are a significant cause of food-borne disease in the US. Non-typhoid Salmonella usually cause gastroenteritis. Although this is a self-limiting disease marked by diarrhea and abdominal cramps, the infection can be more severe, resulting in bacteremia, fever, or even death. Much of our understanding about the molecular mechanisms of Salmonella pathogenesis comes from the study of animal and in vitro models.
These studies are facilitated
by the well-defined genetic system in Salmonella that allows the
simple manipulation and characterization of mutations that affect virulence.
In vitro models using tissue culture cell lines have simplified the study
of particular virulence functions, such as invasion of epithelial cells
and survival in macrophages. A more complete analysis of virulence is obtained
using the well-established animal model of Salmonella typhimurium infection of mice. Although this model must closely mimic a systemic infection,
it is useful for establishing minimum infective doses under controlled
conditions. Bacterial strains containing defined mutations that affect
virulence can be tested in the animal to determine the role of the affected
gene product and the contribution to pathogenesis. In these experiments,
the relative virulence of a strain is defined by the number of bacteria
present in a certain tissue or the number of organisms required to kill
the animal. In a limited number of cases, strains with defined mutations
have been tested in human trials. Specific examples will be discussed to
exemplify both the usefulness and limitations of these model systems.
Correlating
Host Resistance and Susceptibility with Biomarkers from in vitro, ex
vivo and Animal Models
H. Kirk Ziegler, PhD
Dept. of Microbiology
and Immunology, Emory University School of Medicine
Atlanta, GA
Listeria monocytogenes is an intracellular bacterial pathogen that can cause infection via
contaminated foods. Listeria is considered an opportunistic pathogen
because severe infectious disease usually occurs in persons that are immunosuppressed.
Infectious disease occurs in newborn infants, pregnant women, elderly persons,
AIDS patients and individuals immunosuppressed for therapeutic purposes
such as organ transplantation. Important issues regarding the susceptibility
to Listeria infection include the identification of factors that
account for the opportunity for infection in immunocompromised persons.
In this regard, it is important to understand both the bacterial virulence
factors and host defense events associated with infection. Because of the
intense study of a mouse model for Listeria infection and immunity,
several microbial and host factors associated with susceptibility to infection
have been defined. The mechanisms of innate resistance and specific acquired
immunity to Listeria infection will be discussed with an emphasis
on the roles of macrophages, T lymphocytes and cytokines. The immunological
markers for susceptibility and resistance to infection will also be presented.
Information gained with modern mouse models of infection including genetically
engineered gene knockout mice will be presented in the context of risk
assessment in diverse human populations.
Use
of Epidemiologic Data in Dose-Response Modeling
Kirk Smith, DVM, PhD
Minnesota Department
of Health
Minneapolis, MN
Epidemiologic outbreak investigations have provided unique settings to increase our knowledge of dose-response relationships in human Salmonella infections. In two outbreaks of salmonellosis investigated by the Minnesota Department of Health, the highest levels of Salmonella contamination detected in implicated vehicles were 4.3 organisms per 100 g of cheese and 6.0 organisms per 65 g (one half cup) serving of ice cream. Previous outbreak investigations elsewhere in the United States and Canada have documented similar low level Salmonella contamination of implicated products such as cheese and chocolate. Estimates of infective doses of Salmonella provided by these outbreak investigations are several logarithms lower than estimates of minimum infective doses provided by clinical trials with a limited number of volunteers. A likely reason for this discrepancy is the dose-dependent attack rate that would limit the power of volunteer studies to detect an effect with low levels of contamination.
Epidemiologic investigations
have the potential to yield much more useful data on infective doses of
foodborne pathogens, especially as the ability increases to detect interstate
outbreaks due to low level or sporadic contamination of widely distributed
products. Characteristics of outbreaks that make them amenable to yielding
useful dose-response data include large magnitude, clear implication of
a single vehicle, successful trace back of implicated product to the source,
prompt action to allow recovery of product while still in distribution,
and a product type that allows accurate determination of contamination
level. Coordination among state and federal regulatory agencies during
the early phases of outbreak investigations is critical to product identification
and collection. More detailed attention to the quantity of implicated product
consumed by laboratory-confirmed case patients would increase the utility
of epidemiologic investigations in evaluating dose-response models.
Enumeration
of Salmonella enteritidis in an Outbreak Associated with Ice Cream
S.R. Tatini1,
F. Harrell3, K. Vought4. A. Varbanov2 and D. M. Hawkins2
1Department
of Food Science and Nutrition, University of Minnesota, Minneapolis, MN
2School of
Statistics, University of Minnesota, Minneapolis, MN
3Midwest
Laboratory for Microbiological Investigations, FDA, Minneapolis, MN
4Minnesota
Department of Agriculture, St. Paul, MN
Several containers of ice
cream (from consumers and distributors) from the 1994 multi-state outbreak
were tested by the 3 tube, 3 dilution most probable number (M.P.N.) Procedure
(100g, 10 g and 1 g) in conjunction with the USFDA procedure (7th ed. Bacteriological Analytical Manual, 1992) using lactose broth for pre-enrichment.
Several sub-samples from these containers were also tested by (1) a 5 tube
MPN procedure using 2.5g, 0.25 and 0.025 g and (2) individual 25 g or 10
g samples. These data provided MPN estimates of S. enteritidisfor individual containers; the average MPN/g varied from 0.004 to 0.462.
Additional tests were performed with ice cream from two containers (one
from consumer and one from distributor) over a period of nearly 2.5 years.
There was no storage time effect on S. enteritis (weighed regression
analysis). Ice cream from these containers 10, 20, or 25 g aliquots were
directly incubated at 35oC (without the addition of lactose
broth). Sample aliquots of 0.1 or 0.2 ml were plated onto Xylose lysine
desoxycholate agar (XLD) and in some instances onto Trypticase soy agar
(TSAYE) at 0, 2, 3, 4, 6 and 24 hours. H2S positive black colonies
were first detected in some samples as early as 3 hours. The cell numbers
(colony forming units) seen at 3 hours on samples from two different containers
were used to calculate the number of clusters and cells within each cluster
using a statistical model we developed. The model assumed that the number
of clusters in a sample volume of V grams follows Poisson distribution
with parameter 2 x V and that the numbers of cells in the cluster follows
a log-series distribution with parameter P. Inference about these parameters
was based n the maximum likelihood theory. There was evidence for clustering
and there were 0.064 to 0.12 clusters per g with an average of 3.62 cells
per cluster. A parameters for all the containers. These were combined to
estimate a representative 2 for the population of contaminated containers
which was used together with estimate P to calculate the number of cells
per serving (65 g). The average number of S. enteritidis cells per
serving was 22 with a confidence interval of 17-27.
Suitability
of Small Human Clinical Studies to Measure Pathogenesis of Foodborne Pathogens
David R. Tribble, MD,
CDR
U.S. Naval Research
Institute
Bethesda, MD
Experimental infection of volunteers with foodborne pathogens has been used for many years to study pathogenesis and virulence, infection-induced immunology and subsequent immune correlates of protection at repeat challenge. In recent years, with greater numbers of candidate vaccines, this model has increasingly been applied to study preliminary vaccine efficacy. The strengths from the vaccine developmental standpoint are potential limitations for evaluating pathogenesis as it relates to dose-response. The artificial nature of inoculum delivery, limitation to enrolling healthy young adults, and use of higher inocula to yield consistent illness attack rates in "naive" volunteers are all factors restricting application to real world quantitative risk assessment. Methodologic requirements that would broaden applicability of these studies for risk assessors include the following: 1) well-defined outcomes covering the illness spectrum using quantitative measures, 2) range of inocula based on prior knowledge of infective capability and illness attack rates using clinical studies, outbreak investigations and quantitative food microbiology, 3) accurate quantitative microbiology of dose delivery and subsequent pathogen recovery in stool specimens, 4) adequate observation period post-inoculation to survey for clinical and immunological outcomes, and 5) reproducible methodology to allow inter-study comparisons including inoculum standardization, volunteer selection criteria considerations (i.e., pre-screening pathogen-specific immunology), similar outcome definitions and correlation with biomarkers and animal models.
Last modified: May 19, 1999
