The Toxicologist - Society of Toxicology

889 APPLICATION OF TISSUE-TIME COURSE DATA TO
ELUCIDATE MECHANISTIC DETAILS OF CARBON
TETRACHLORIDE (CCL4) TRANSPORT USING AN
UPDATED PHYSIOLOGICALLY BASED
PHARMACOKINETIC (PBPK) MODEL IN RATS.
M. V. Evans 1 , C. R. Eklund 1 , U. Y. Sanzgiri 2 , J. V. Bruckner 2 and J. E.
Simmons 1 . 1 Pharmacokinetics, U.S. EPA, Research Triangle Park, NC and
2
Department of Pharmaceutical and Biomedical Sciences, University of Georgia,
Athens, GA.
CCl4 is a common environmental contaminant in water and superfund sites, and a
model liver toxicant. One application of PBPK models used in risk assessment is
simulation of internal dose for the metric involved with toxicity, particularly for different
routes of exposure. Time-course pharmacokinetic data for different tissues
(Sanzgiri et al., 1997) were used to evaluate a rat PBPK model employing previous
metabolic estimates. The flow-limited PBPK model contained: fat, liver, brain, rapidly
and slowly perfused compartments. Arterial concentration data measured for
100 or 1000 ppm constant inhalation exposure lasting 2 hours were used to evaluate
the initial PBPK predictions. These simulations were able to describe the timecourse
data well at both inhaled concentrations. However, a closer examination of
tissue uptake data of 1000 ppm CCl4 revealed an inconsistency with the preliminary
simulations. Upon further evaluation of tissue time-course results, the PBPK
model underpredicted brain and fat tissue concentrations. Also, the peak arterial
concentration was about twice that predicted by the initial calibration simulation.
Further modeling is needed to incorporate physiological and structural details (i.e.
diffusion) that may be taking place at higher exposure concentrations. Ongoing
model refinement strategies will include: (1) addition of a blood brain barrier component
(for brain tissue), and (2) a diffusion-limited compartment for fat, and (3)
diffusion in the respiratory tract during the breathing cycle. Diffusion may help explain
the difference between observed and predicted tissue concentrations. In summary,
tissue time-course data are essential for the determination of mechanistic detail
for different organs, and the accurate prediction of internal liver dose. (This
abstract does not reflect EPA policy.)
890 ALTERNATIVE APPROACH TO MAXIMUM FLUX FOR
TTC APPLIED TO SAFETY EVALUATION OF
COSMETIC INGREDIENTS.
A. Garrigues-Mazert, S. Grégoire and J. Meunier. Safety Research Department,
L’Oréal, Aulnay sous Bois, France. Sponsor: G. Nohynek.
Relevant and accurate prediction of dermal uptake/exposure of topically applied
chemicals is essential for risk assessment. It could be obtained without recourse to
an experimental measurement and avoids any problems with ethical issues, recruiting
volunteers or housing animals.
Typically, QSAR model predicting permeability coefficients (i.e. kp) are used.
Many models were developed; all of them lead to the same conclusion: small
lipophilic chemicals have greatest skin permeability. This analysis often rises to confusion.
Dataset used to build up this relation concerns percutaneous transport from
aqueous solution. Whereas, kp increases with log P, aqueous solubility decreases
with lipophily. Resulting flux, and effective absorbed amount of chemical, is then
balanced between two competitive factors (permeability and solubility).
Concept of maximum flux means that a chemical cannot cross the skin higher than
flux measured at steady state with a chemical applied on the surface in saturated solution
(or in neat chemical form). It allows assessing the maximum absorbed dose.
This concept was recently used in TTC approach for cosmetic ingredient. A classification
of potential of cutaneous chemical absorption was proposed on the basis of
their physico-chemical properties. Unfortunately, the proposed classification does
not cover all range of molecular weight and log P. Moreover, the default proposed
values greatly overestimate the absorption experimentally obtained.
To overcome these limitations, a QSAR model recently developed by L’Oréal can be
used. It estimates the cumulative mass of a chemical absorbed into and through the
skin in typical ‘in-use’ cosmetic conditions. Applicability domain is clearly defined
and covers a wide range of physico-chemical properties. Moreover, the model was
build up with 101 data. More than 90% were well predicted (i.e. difference between
predicted and experimental values less than a factor 5). At least, improvement
was recently done to take into account property of volatile chemicals.
891 SKIN ABSORPTION STRATEGY: FROM IN SILICO TO
EX VIVO.
A. Garrigues, S. Grégoire, W. Wargniez, C. Patouillet, I. Durand, J. Becquet
and J. Meunier. Safety Research, L’Oreal, Aulnay sous Bois, France. Sponsor: G.
Nohynek.
Skin is daily exposed to many chemicals that could represent a risk to human
health. Thus, determination of cutaneous absorption is a key parameter to evaluate
the exposure and assess the risk. Since animal experiments have to be avoided for
ethical considerations and for scientific relevance, the bioavailability is typically
achieved by conducting relevant in vitro permeation studies in accordance with accepted
guidelines. Particularly the OECD guidelines 428 clearly have stated that
excised Human skin, or alternatively pig skin, are the most suitable models to predict
in vivo dermal absorption. However, limitations of this method are the variability
between donors, the availability of skin explants and time consuming for
sample preparation. To overcome these limitations and to reduce the number of ex
vivo experiments, reliable and widely applicable in silico models and in vitro reconstructed
Human epidermis models (RHE) can be used. On one hand, a new QSAR
model was developed by L’Oréal Laboratories to estimate a cumulative dose absorbed
into the skin with a concordance of more than 90% (I.e. the difference between
predicted and experimental data was less than a factor 5). On the other hand,
it has been proved that the RHE models Episkin are appropriate alternatives to
human skin and useful for ranking chemicals according their skin permeation potential,
despite their lower barrier function. To conclude, sequential testing strategy
for cutaneous absorption provides information very quickly to prioritize the ex vivo
experimental investigations, to predict the permeation of a new chemical entity
(even before it synthesis) and to lead to an efficient screening of chemicals.
892 COMPARISON OF ESTIMATED PCB-153
CONCENTRATIONS IN HUMAN MILK USING
VARIOUS PHARMACOKINETIC MODELS.
D. G. Farrer 1 , M. Poulsen 2 , D. Davoli 3 , M. Bailey 3 , D. Moffett 4 , D. Fowler 4 ,
C. Welsh 4 , R. Yang 5 , P. Ayotte 6 , M. Verner 7 , G. Muckle 6 and S. Haddad 7 .
1
Oregon DHS, Portland, OR, 2 Oregon DEQ, Portland, OR, 3 U.S. EPA, Seattle,
WA, 4 ATSDR, Atlanta, GA, 5 Ray Yang Consulting LLC, Ft. Collins, Co., 6 Centre
Hospitalier Universitaire de Québec, Montréal, QC, Canada and 7 Department of
Biological Sciences, Université du Québec, Montréal, QC, Canada.
Risk to infants from consuming human milk contaminated with lipophilic chemicals
is difficult to assess. Typically, risk assessors use measured chemical concentrations
in media of concern. Human milk is often difficult to sample for chemical
concentrations. Therefore, models that predict chemical levels in human milk and
subsequent average daily dose to the nursing infant (ADDi) are desirable. We compared
adaptations of three published models in an effort to help risk assessors and
public health practitioners choose an appropriate method to estimate ADDi.
Models chosen for comparison included a classic single-compartment model, a 3-
compartment physiologically-based pharmacokinetic (PBPK) model, and an 8-
compartment PBPK model. The models were compared by running two sets of
simulations in each model using the polychlorinated biphenyl congener 153 (PCB-
153), a lipophilic environmental contaminant. The first set of simulations used
back-calculated average maternal daily oral dose (ADDm) values as starting points.
ADDm was calculated using the 8-compartment PBPK model based on PCB-153
blood concentrations measured in 8 human subjects. From derived ADDm values,
the models simulated both the milk concentration and ADDi for each subject.
Estimated milk concentrations were then compared to observed concentrations.
The second set of simulations used an ADDm derived for PCB-153 assuming consumption
of contaminated fish. All 3 model results were similar to within a factor
of 2. The classic single compartment model consistently produced the highest estimates
of PCB-153 concentration in human milk and ADDi. Our results indicate
that the simplest model studied may be appropriate for risk assessors and public
health practitioners to use for predicting the ADDi for lipophilic environmental
contaminants.
190 SOT 2010 ANNUAL MEETING