Common Issues and Challenges in Interpreting Biomonitoring Data

1
Common Issues and Challenges in Interpreting
Biomonitoring Data
Elaine Faustman, PhD University of Washington
2
Issues
3
Example of multiple-exposure pathways to
environmental pollutants
Sexton et al., 1992
4
Desire/challenge to use Biomonitoring data to
address important environmental health questions

• Is agent a hazard?
• Are we exposed? (yes, no)
• Who, what, where, when, why are we exposed?
• Are we affected now?
• Did the exposure cause the health effect?
• Is there a risk?
• What are the relative effectiveness and efficacy
  of viable control strategies?
• What are important risk factors and behaviors to
  address in public health programs?
• Is the situation changing over time?

Adapted from Wagener et al., 1995
5
Estimating Human Exposures to Environmental
Pollutants Availability and Utility of Existing
Databases

• Ken Sexton, Sherry G. Selevan, Diane K. Wagener,
  Jeffrey A. Lybarger
• Archives of Environmental Health, 1992, 47(6)398.

6
Common human exposure pathways for environmental
pollutants
Sexton et al., 1992
7
Relationship between exposure estimators and
their costs and ability to predict/classify human
exposures to environmental pollutants
Costs and Ability to Predict/Classify Exposures
Exposure Estimators

• Production Volumes
• Emission Inventories
• Environmental Concentrations
• Models Measurements
• Microenvironmental Concentrations
• Models Measurements
• Human Contact
• Models Measurements
• Internal Dose
• Models Measurements

Generally Lower Costs and Limited Accuracy
Generally Higher Costs and Better Accuracy
Sexton et al., 1992
8
Four major uses of human exposure information and
their interrelationships
Risk Assessment Estimation of the Likelihood and
Magnitude of Human Health Risks Associated
with Environmental Exposures
Risk Management Determination of Which Health
Risks are Unacceptable and What to Do About Those
That Are
Epidemiology Examination of the Link Between
Human Exposures and Health Outcomes
Status Trends Evaluation of Historical Patterns,
Current Status, and Possible Future Changes
in Human Exposures and/or Health Outcomes
Sexton et al., 1992
9
The Importance of Human Exposure Information A
Need for Exposure-Related Data Bases to Protect
and Promote Public Health

• Diane K. Wagener, Sherry G. Selevan, Ken Sexton
• Annu. Rev. Public Health, 1995, 16105.

10
Relationship between public health goals,
information sources and decision-making
PUBLIC HEALTH GOALS
Reduce Adverse Effects (Secondary Prevention)
2
Treat Adverse Effects (Tertiary Prevention)
3
Prevent Adverse Effects (Primary Prevention)
1
SAFEGUARDING ENVIRONMENTAL HEALTH
ASSESSING, MANAGING, AND COMMUNICATING RISKS
Risk Assessment
Risk Management
Knowledge and Understanding
Research and Surveillance
Estimating the magnitude, likelihood, and
uncertainty of risks
Deciding which risks are unacceptable
Deciding what actions to take
Exposure Effects Link between exposure
and effects Dose response
Epidemiology Toxicology Clinical
studies Cell/tissue experiments
Computational methods Monitoring Exposure
analysis
Risk Communication
Establishing risks to important
stakeholders Responding to stakeholders
concerns and questions
Adapted from Wegener et al., 1995
11
Data to address important environmental health
questions
Adapted from Wegener et al., 1995
12
Environmental health paradigm and mechanisms
affecting sequence of events leading to
environmental-related illness
EFFECTS ASSESSMENT
EXPOSURE ASSESSMENT
MECHANISTIC BASIS FOR THE SEQUENCE OF EVENTS
LEADING TO ENVIRONMENTAL-RELATED ILLNESS OR INJURY
Adapted from Wegener et al., 1995
13
CDC Environmental Public Health Tracking
Framework (EPHT)
Ongoing Evaluation
Risk Assessment Paradigm Hazard Exposure Health
Effect
TrackingNetwork Collection Integration Analysis I
nterpretation
Stakeholders Assessment Research Intervention Poli
cy
IMPROVED PUBLIC HEALTH
Ongoing Evaluation
Note the importance of the risk assessment
paradigm including hazard identification,
exposure and identification of health effects in
providing the foundation for the EPHT framework.
www.cdc.gov/nceh/tracking/default.htm
14
EPA Environmental Information Exchange Network
Grant Program
New approach for exchanging environmental data
- between EPA and state, tribal, and
territorial partners Electronically collects
and stores - accurate information, -
integrates information from across many
sources - provides secure access to information
http//www.epa.gov/neengprg/
15
Environmental Public Health Paradigm
ALTERED STRUCTURE FUCTION

HAZARD CHARACTERIZATION
Modifications to population or toxicology study
design
Factors/Criteria
Factors/Criteria
EXPOSURE
DOSE
Individual Community Population
Factors/Criteria Examples Temporal
context Constant or intermittent Sources
Factors/Criteria Examples Comparative
toxicokinetics Uncertainty Short half-life
ILSI/HESI
16
ILSI/HESI Workshop Focus Area
ENVIRONMENTALCHARACTERIZATION
ALTERED STRUCTURE FUNCTION
HAZARD IDENTIFICATION
Modifications to population(s) or toxicology
study design
EARLY BIOLOGICALEFFECT
EXPOSURE
DOSE
POSSIBLE FACTORS/CRITERIA Temporal
context Constant or intermittent
use(s) Source(s) of exposure Bioaccumulative
Rapidly metabolized
POSSIBLE FACTORS/CRITERIA Toxicokinetics Unce
rtainty intraspecies extrapolation
(interindividual) interspecies extrapolation
(animal to human)
Individual Community Population(s)
ILSI/HESI
17
The Light at the End of the Tunnel for
Chemical-Specific Biomarkers Daylight or
Headlight?

• John D. Groopman and Thomas W. Kensler
• Carcinogenesis, 1999, 20(1)1.

18
Model for validating chemical-specific biomarkers
Groppman and Kensler, 1999
19
Large Within Person Variability for
Organophosphate Pesticide Urinary Biomarkers
Limits Our Ability to Identify High Exposure
Groups

• William C Griffith, Eric Vigoren,
• Richard A Fenske, Elaine M Faustman
• Department of Occupational Environmental Health
  Sciences
• University of Washington, Seattle, WA

This work was made possible by grant number PO2
ES09601 from the National Institutes of
Environmental Health Sciences (NIEHS), NIH and
EPA R826886 from the U.S. Environmental
Protection Agency, and Center for the Study and
Improvement of Regulation, Carnegie Mellon
University. Its contents are solely the
responsibility of the author and do not
necessarily represent the official views of the
NIEHS, NIH, EPA, or Carnegie Mellon University.
20
What can we learn from longitudinal studies?
21
Center for Child Environmental Health Risks
Research
Exposure Assessment Project
Three longitudinal studies of potential OP
metabolites Estimating within and between
person variabililty
22
Interpretation of Urinary Metabolites of OP
Pesticides

• Analysis of metabolites of Organophosphate
  Pesticides (OPs) in urine of adults and children
  are a common method for estimating potential
  exposure to OPs
• Interpretation of urinary metabolite levels
  present a number of problems with interpretation
• This study examined whether these measurements
  can be used to identify the more highly exposed
  individuals in a population and focused on the
  evaluation of within person variability compared
  to between person variability

23
Uncertainty Analysis of Pesticide Exposure Studies
Key findings and relevance for longitudinal
studies

• Within individual variability in exposure is
  greater than 10x the between individual
  variability in exposure
• Single snapshots of exposure are problematic
• Spray season varies dramatically from one year
  to the next due to weather patterns
• Types of pesticides used one year to the next can
  change (not only due to regulatory changes) but
  especially by meteorological/crop conditions
• Large number of samples of urinary metabolites
  (8-10) were needed to correctly identify persons
  in a population who are more highly exposed to
  Ops (with 60 prediction)

24
New Directions in Biomonitoring
25
Flow of Information from New Omic Data for
Biomonitoring
Methods for Assessment
metabolomics metabonomics
genome sequencing
proteomics
microarray
SNP
adapted from Corton, et al., 1999
26
Flow of Information from New Omic Data for
Biomonitoring
Methods for Assessment
genome sequencing
proteomics
metabolomics metabonomics
microarray
SNP
Implications for Biomonitoring
inherent susceptibility
early biological response
altered function
adapted from Corton, et al., 1999
27
Development of genomic/proteomic technology
Tools have been developed to map entire genomes,
measure smaller amounts of mRNA and protein, and
simultaneously measure many more mRNAs and
proteins.
Patterson and Aebersold, 2003
28
Environmental Public Health Paradigm
ALTERED STRUCTURE FUCTION

HAZARD CHARACTERIZATION
Modifications to population or toxicology study
design
Factors/Criteria
Factors/Criteria
EXPOSURE
DOSE
Individual Community Population
Factors/Criteria Examples Temporal
context Constant or intermittent Sources
Factors/Criteria Examples Comparative
toxicokinetics Uncertainty Short half-life
ILSI/HESI
29
Identification of toxicologically predictive gene
sets using cDNA microarrays

• Russell S. Thomas, David R. Rank, Sharron G.
  Penn, Gina M. Zastrow, Kevin R. Hayes, Kalyan
  Pande, Edward Glover, Tomi Silander, Mark W.
  Craven, Janardan K. Reddy, Stevan B. Jovanovich
  and Christopher A. Bradfield
• Molecular Pharmacology, 601189, 2001.

30
Variation in expression of approx. 500
transcripts in 24 experimental treatments

• Rows depict specific transcripts
• Columns depict treatment
• Red - up regulated
• Green - down regulated
• Black - no change
• 2-D clustering allows for organizing of
  transcripts and treatments on the basis of
  similarity

Thomas et al., 2001
31
Conclusions

• Classification of toxic chemicals according to
  their transcript expression profiles is possible.
• Transcript expression may allow for
  prioritization of untested chemicals based on
  classification.
• Only a small number of transcripts allowed for
  predictive accuracy - once the diagnostic gene
  set is identified

32
Microarray analysis of hepatotoxins in vitro
reveals a correlation between gene expression
profiles and mechanisms of toxicity

• Jeffrey F. Waring, Rita Ciurlionis, Robert A.
  Jolly, Matthew Heindel and Roger G. Ulrich
• Toxicol Lett. 2001 Mar 31120(1-3)359-68

33
Cluster analysis of 15 different known
hepatotoxins. A total of 179 genes were shown to
be changed at least 2-fold by at least one
compound.
Waring et al., 2001
34
Study Summary

• Gene expression profiles formed clusters of
  compounds with similar toxic mechanisms.
• There was not complete identity, however,
  indicating that each compound produced a unique
  signature.
• Large scale analysis of gene expression using
  microarray technology can be an important
  diagnostic tool for toxicology.

35
The transcriptional program in the response of
human fibroblasts to serum

• Iyer VR, Eisen MB, Ross DT, Schuler G, Moore T,
  Lee JC, Trent JM, Staudt LM, Hudson J Jr, Boguski
  MS, Lashkari D, Shalon D, Botstein D, Brown PO
• Science. 1999 Jan 1283(5398)83-7

36
Cluster Analysis
Rows Genes Columns Fibroblast samples Time
course after serum stimulation Clusters of
Genes AJ identified through their similarity in
expression profiles. The immediate response was
dominated by genes that encoded transcription
factors such as c-FOS, JUN B and MAPK. Approx.
half the total number of genes differentially
expressed were unnamed expressed sequence tags
(ESTs).
Iyer et al., 1999
37
Illustration of the complexity of dose and time
relationships between various levels of
biomonitoring assessment
Faustman and Omenn, 2001
38
Integrating dose-response relationships for
critical responses/ impacts across assessment
levels
Rodents
Populations
Humans
Lung Liver
Lung Liver
Polymorphisms
Multinomial Distribution
Levels of Biological Assessment
Epigenetic Factors
39
The Source-to-Outcome Continuum
EPA., 2003
40
Final Points/Considerations
Continued interest in using biomonitoring
information to inform public health. Continued
struggle to pull information into interpretable
risk assessment frameworks. Case studies have
illustrated that details and knowledge in PK, TD
and mechanisms can facilitate this
interpretation. New efforts are focusing on
interpretation of omics data. Recent
reports/activities should help move our efforts
forward.
41
Is it safe to eat vegetables from my garden?
42
(No Transcript)
43
(No Transcript)
44
Within and Between Child Distributions for DMTP
Between Child Distribution of Means
Geometric standard deviation 2.2 ( 1.5,2.8
/-1std.err.)
Geometric standard deviation 4.3 ( 3.9,4.7
/-1std.err.)
Distribution for Child at 80th Percentile
Distribution for Child at 20th Percentile
Urinary DMTP (µM/L)
45
Within and Between Child Distributions for DMTP
Study of 44 children (2 to 5 years old) of
Hispanic farmworkers in Yakima followed over 21
months (Koch, et al., Env Hlth Persp, 2002). 993
samples with 32 of DMTP analyses below limit of
detection (other DAP metabolites had much higher
percentages - 60 to 99). Covariates of gender
and season were treated as fixed effects.
46
Within and Between Person Distributions for TCP
(Maryland)
Population based random sample collected over 1
year Up to 6 samples per person, samples
separated by 2 months 79 individuals and 341
samples Only 14 samples, 4 below limit of
detection Covariates of gender, age, season were
treated as fixed effects NHEXAS data
47
Within and Between Person Distributions for TCP
(Minnesota)
Population based random sample collected at 3
times, separated by 2 days 90 individuals and 263
samples Only 20 samples at 8, below limit of
detection Covariates of gender, age, residence
were treated as fixed effects NHEXAS data shared
by John Quackenboss
48
Predictive Value
Predictive value positive shown at right is the
percent whose exposure is correctly classified as
being in the upper 10 of the population for the
number of measurements per person shown above.
49
Steps of Analysis
1) Use three longitudinal studies of OP pesticide
urinary metabolites to estimate within and
between person variances to describe the
variability of exposure to OPs. Nonspecific
Metabolites Diakyl Phosphate (DAP) metabolites in
children of farmworkers in Yakima
Valley Specific metabolite for chlorpyrifos and
chlorpyrifos-methyl TCP in NHEXAS MN TCP in
NHEXAS MD 2) Use the estimates of within and
between person variances as estimates of the
variability of exposures to OPs of a pregnant
woman and her fetus. 3) Based upon the within and
between person variances calculate the predictive
value positive of correctly classifying a
pregnant woman as being more highly exposed than
the remainder of the population In these
calculations we used as an example the upper 10
of the population The predictive value
positive is the percent of the population
assigned to a group that are correctly classified
50
Implications of Using Urinary OP Metabolites to
Identify Highly Exposed People

• TCP as a specific metabolite for two OPs had much
  smaller between and within person variability
  than DMTP a non-specific OP metabolite
• However for both cases the within person
  variability was larger than the between person
  variability
• Large degree of misclassification for identifying
  highly exposed people would occur using these
  biomarkers
• Epidemiology studies that need to classify
  exposure would need 8-10 measurements per person
  for the predictive value positive to be above 60
• Simulations show that lower limits of detections
  would provide only small reductions in
  misclassication in a high exposure group

51
(No Transcript)
52
(No Transcript)
53
Environmental Public Health Continuum
54
Exposure Assessment Project

• Within and between person variability treated as
  a random effect and other variables such as age,
  gender, residence, season treated as fixed
  effects- Design of the three studies differed in
  term of fixed effects available in the study
• Different urinary OP metabolites measured in the
  three studies- Nonspecific DAP metabolites
  measured in one study and TCP, a metabolite
  specific to CP and CP-methyl, measured in two
  studies- DAP metabolites had a high percentage
  below limits of detection (LOD) and TCP a low
  percentage
• Measurements below limit of detection (LOD) were
  treated as being left censored in statistical
  analyses- Maximum likelihood methods used to
  estimate random and fixed effect parameters-
  Maximum likelihood methods provide unbiased
  estimates compared to using LOD/2 for values
  below LOD

55
Geometric mean and 95 confidence interval for
dimethyl DAP concentrations by sampling months
Arrows indicate months when dimethyl OP compounds
were applied to tree fruit orchards.
Koch et al., EHP, 2002
56
Geometric mean and 95 confidence interval for
diethyl DAP concentrations by sampling months
Arrows indicate months when dimethyl OP compounds
were applied to tree fruit orchards.
Koch et al., EHP, 2002
57
Study Design

• Recruited 44 children, 25 years old at clinic in
  Lower Yakima Valley, Washington State
• At least one parent a Hispanic farm worker
• Diakyl metabolites of organophosphorus pesticides
  (OP) measured in urine
• Multiple pesticides produce the same metabolite
• Dimethyl OPs produce DMP (dimethylphosphate),
  DMTP (dimethylthiophosphate), DMDTP
  (dimethyldithiophosphate)
• Diethyl OPs produce DEP (diethylphosphate), DETP
  (diethylthiophosphate), DEDTP (dimethyldithiophosp
  hate)
• Many samples below level of detection
• Urine collected on biweekly basis over a 21 month
  period
• Dec 1997 to Aug 1999
• Two spray seasons
• May-July 1998 June-Aug 1999 for dimethyl OPs
• Mar-Apr 1998 April 1999 for diethyl OPs

58
Implications of Within and Between Child
Distributions for DMTP

• Within child variation is much larger than
  between child variance
• On the logarithmic scale within child variance is
  3.5 times larger than within child variance
• log(4.3)2 / log(2.2) 2 3.5
• Large within child variance makes it difficult to
  identify which children are highly exposed
• Example Identify the top 10 of exposed children

59
Organophosphate Pesticides (OPs)

• Toxicity of OPs
• Common mode of action inhibits breakdown of
  neurotransmitter acetylcholine
• Toxicity occurs in humans, other animals, and
  insects
• OPs among most widely used pesticides
• Agricultural and residential use about 40 OPs
  registered for use
• Inexpensive and potent toxicity to insects
• Account for half of all insecticides used
• Concerns about exposure of people to OPs
• Multiple routes of exposures
• Outdoors degrade rapidly but indoors OPs are more
  persistent
• First class of pesticides that were evaluated
  under the Food Quality Protection Act
• Residential use of chlorpyrifos and diazanon
  eliminated because of child safety concerns
• Metabolites of OPs used to estimate exposure
• Six nonspecific diakyl-phosphate (DAP)
  metabolites
• Specific metabolites

60
Statistical Analysis of Longitudinal Data
Statistical methods to estimate within and
between person variances
Maximum Likelihood Measurements below limit of
detection are treated as being left
censored. Utilizes correlations among
metabolites to improve estimates for metabolites
below limit of detection. Mixture of Random and
Fixed Effects Fixed effects are trends with
time, gender, age, or residence. Random effects
are between and within child variances. Fixed
and Random effects are log-normally distributed.
61
Metabolites of Organophosphate Pesticides

• Biomarkers of exposure
• Nonspecific Diakyl Phosphate (DAP) metabolites
• Six DAP Metabolites
• Each metabolite can be produced by multiple OPs
• Divided into two groups
• Dimethyl metabolites
• DMP, DMTP, DMDTP
• Diethyl metabolites
• DEP, DETP, DEDTP

Selected OPs and DAP metabolites
Diethyl OPs chlorpyrifos DEP DETP diazinon
DEP DETP disulfoton DEDTP DEP
DETP ethion DEDTP DEP DETP parathion DEP
DETP Dimethyl OPs azinophos methyl DMDTP DMP
DMTP chlorpyrifos methyl DMP DMTP dichlorvos
(DDVP) DMP malathion DMDTP DMP DMTP methyl
parathion DMP DMTP naled DMP phosmet DMDTP
DMP DMTP trichlorfon DMP

• Nonspecific metabolites
• Chlorpyrifos metabolites
• TCP, DEP, DETP
• Chlorpyrifos-methyl metabolites
• TCP, DMP, DMTP

62
Longitudinal Study of DAP Metabolites in Children
of Farmworkers
Study of 44 children (2 to 5 years old) of
Hispanic farmworkers in Yakima followed over 21
months (Koch, et al, Env. Hlth.Persp.,
2002) 993 samples Covariates of gender and
season were treated as fixed effects
63
Diakyl Phosphate Metabolites Below Limit of
Detection
Washington State Children of Farmworkers
64
Seasonal Variation in Urinary DMTP Fixed Effects
Washington State Children of Farmworkers
Urinary DMTP (µM/L)
Error Bars are /- 1 Standard Error
Dec97-Apr98 May98-Jul98 Aug98-May99
Jun99-Aug99
Non-Spray Spray Non-Spray Spray
65
Large Within Person Variability

• Statistical Methods
• Estimate within and between person variability
• Large fraction of measurements below level of
  detection
• Implications for identifying highly exposed
  individuals
• Three case examples for OP pesticide urinary
  metabolites
• Nonspecific Metabolites
• Diakyl Phosphate (DAP) metabolites in children of
  farmworkers in Yakima Valley
• Specific metabolite for chlorpyrifos and
  chlorpyrifos-methyl
• TCP in NHEXAS MN
• TCP in NHEXAS MD

66
A Framework for a Computational Toxicology
Research Program

• US Environmental Protection Agency

EPA/600/R-03/065 November 2003
67
Computational Toxicology
Defined as The application of mathematical and
computer models and molecular biological
approaches to improve the Agencys prioritization
of data requirements and risk assessments.
EPA., 2003
68
Linking Observations Across Levels of Biological
Organization An example of a toxicity pathway
EPA., 2003
69
Objectives of the Computational Toxicology
Research Program
Improve linkages in source-to-outcome continuum
Provide predictive models for hazard
identification Enhance quantitative risk
assessment
EPA., 2003
70
Systems Biology
Defined as Computational models that
reconstruct a cell, organ or organisms function
from component parts, and Allows validation and
simulator experiments that build confidence in
predictive ability of adverse effects.
EPA., 2003
71
Identification of toxicologically predictive gene
sets using cDNA microarrays
Russell S. Thomas, David R. Rank, Sharron G.
Penn, Gina M. Zastrow, Kevin R. Hayes, Kalyan
Pande, Edward Glover, Tomi Silander, Mark W.
Craven, Janardan K. Reddy, Stevan B. Jovanovich
and Christopher A. Bradfield Molecular
Pharmacology, 601189, 2001.
72
Study Design

• Acute treatment of mice with various
  toxicologically relevant chemicals.
• RNA isolated from the liver.
• RNA (n3) was used in microarray analysis using a
  custom cDNA microarray.
• Hierarchical clustering performed to identify
  chemicals with similar expression patterns.

Thomas et al., 2001
73
Study Design
Treatment, dose, vehicle, and time of sacrifice
Thomas et al., 2001
74
Variation in expression of approximately 500
transcripts in 24 experimental treatments

• Rows depict specific transcripts
• Columns depict treatment
• Red - up regulated
• Green - down regulated
• Black - no change
• 2-D clustering allows for organizing of
  transcripts and treatments on the basis of
  similarity

Thomas et al., 2001
75
General toxicological classes and corresponding
treatments classified in this study using gene
expression profiles from cDNA microarrays
Thomas et al., 2001
76
Toxicological classes identified by microarray

• Each of the 12 chemicals at multiple time points
  fell into one of 5 different toxicological
  classes identified through cluster analysis.
• Noncoplanar-PCBs
• Peroxisome proliferators
• Inflammatory
• Hypoxia
• AHR Agonist

Thomas et al., 2001
77
Estimated predictive accuracy of the
classification model
The predictive accuracy is a function of the
number a transcripts added. In this model,12
transcripts is diagnostic for the
classification of treatments.
Predictive
Relative
Thomas et al., 2001
78
Transcriptional profile of the diagnostic
transcripts
This identified set of 12 diagnostic
transcripts provided 100 predictive accuracy for
the toxicological classes chosen.
Diagnostic Transcripts
Thomas et al., 2001
SEE COLOR SLIDE
79
Conclusions

• Classification of toxic chemicals according to
  their transcript expression profiles is possible.
• Transcript expression may allow for
  prioritization of untested chemicals based on
  classification.
• Only a small number of transcripts allowed for
  predictive accuracy - once the diagnostic gene
  set is identified

80
Microarray analysis of hepatotoxins in vitro
reveals a correlation between gene expression
profiles and mechanisms of toxicity

• Jeffrey F. Waring, Rita Ciurlionis, Robert A.
  Jolly, Matthew Heindel and Roger G. Ulrich

Toxicol Lett. 2001 Mar 31120(1-3)359-68
81
Study Design

• Treated rat with 15 known hepatotoxins
• Carbon tetrachloride, allyl alcohol, aroclor
  1254, methotrexate, diquat, carbamazepine,
  methapyrilene, arsenic, diethylnitrosamine,
  monocrotaline, dimehty-formamide, amiodarone,
  indomethacin, etoposide and 3-methylcholanthrene.
• RNA from livers isolated
• Microarray used to characterize compounds based
  on gene expression.

Waring et al, 2001
82
Cluster analysis of 15 different known
hepatotoxins. A total of 179 genes were shown to
be changed at least 2-fold by at least one
compound.
Waring et al, 2001
SEE COLOR SLIDE
83
Hepatotoxicant Clustering
Gene expression profiles for compounds with
similar toxic mechanisms formed clusters,
suggesting a similar effect on transcription.
Waring et al., 2001
84
Study Summary

• Gene expression profiles formed clusters of
  compounds with similar toxic mechanisms.
• There was not complete identity, however,
  indicating that each compound produced a unique
  signature.
• Large scale analysis of gene expression using
  microarray technology can be an important
  diagnostic tool for toxicology.

85
Flow of Information from New Omic Data for
Biomonitoring
adapted from Corton, et al., 1999
86
ILSI/HESI Workshop Focus Area
ENVIRONMENTALCHARACTERIZATION
ALTERED STRUCTURE FUNCTION
HAZARD IDENTIFICATION
Modifications to population(s) or toxicology
study design
EARLY BIOLOGICALEFFECT
EXPOSURE
DOSE
POSSIBLE FACTORS/CRITERIA Temporal
context Constant or intermittent
use(s) Source(s) of exposure Bioaccumulative
Rapidly metabolized
POSSIBLE FACTORS/CRITERIA Toxicokinetics Unce
rtainty intraspecies extrapolation
(interindividual) interspecies extrapolation
(animal to human)
Individual Community Population(s)
ILSI/HESI