The Toxicologist - Society of Toxicology

1815 MEASURING COMPOUND SELECTIVITY AND ITS
LINK TO IN VIVO TOXICITY STUDY OUTCOME.
X. Wang and N. Greene. Compound Safety Prediction, Pfizer, Groton, CT.
In order to reduce drug attrition, it’s important to assess the safety liabilities of early
leads. A common practice to assess compound toxicity is using pharmacologic profiling
panels. Several methods for measuring compound selectivity were developed
over the years, from simple hit rate (defined as the fraction of assays with inhibition
values greater than certain threshold for each compound) to more sophisticated
methods, such as Gini Coefficient or Thermodynamics-Based Partition Index. Gini
Coefficient measures statistical dispersion and is scale-free with no units. It is originally
used to measure inequality of income or wealth, and had been successfully applied
to measure Kinase inhibitors selectivity. The partition indexed can be defined
as a competition assay in which the compound is placed in a test tube, and then
compete for binding to a pool of targets. At thermodynamic equilibrium, the fraction
of bound compound for a particular target is the “partition” index.
In this study, we compared these three methods: hit rate, Gini Coefficient, and partition
index, based on a panel of 180 Bioprint (CEREP) inhibition assays. First, the
impact of different assay selection on selectivity score was tested by randomized
studies, and a panel of 15 assays was selected that best represent the overall assay selectivity
profiles. Second, ~200 Pfizer compounds had gone through in vivo toxicity
studies, and can be divided into two groups: clean@10uM and findings@10uM.
The capability of each method to separate two groups of compounds based on
those 15 assays was measured. The results show that each methods capture different
nuances of the data. Hit rate is most intuitive; Gini Coefficient needs to be modified
to deal with small panel of assay; while partition index is the most predictive.
It’s possible that the three methods could be combined to generate the optimal selectivity
index.
1816 IDENTIFICATION AND CLASSIFICATION OF
GENOTOXIC AND NON-GENOTOXIC SCAFFOLDS IN
RODENT CARCINOGENICITY QSAR MODELS.
K. P. Cross 1 and E. J. Matthews 2 . 1 Leadscope, Inc., Columbus, OH and 2 U.S.
FDA/CFSAN/OFAS, College Park, MD. Sponsor: R. Tice.
Quantitative structure-activity relationship (QSAR) models have been developed
for predicting mutagenicity in Salmonella typhimurium and carcinogenicity in rodents
using Leadscope software. These models identify structure-activity relationships
(SARs) between molecular features and chemical toxicity, and the SARs can
support plausible mechanism of action (MOA) hypotheses responsible for the toxicity.
The molecular features can be represented as molecular substructures that discriminate
for, or against, toxicity, and the features may be extracted as scaffolds. The
predictive performance of these scaffolds may be assessed for accuracy and precision
independently of the model they were derived from. This study revealed that a relatively
small number of primary scaffolds, alerts, and MOAs were associated with
mutagenicity; conversely, a relatively large number of scaffolds were associated with
carcinogenicity. We also extracted and compared the predominant scaffolds contributing
to carcinogenicity and mutagenicity through hierarchical classification of
discriminating model features. Preliminary analysis of the salmonella QSAR model
indicated that 58% of salmonella model scaffolds matched compounds predominantly
positive for rodent carcinogenicity, while only 2% of the scaffolds matched
compounds predominantly negative for rodent carcinogenicity (40% remained unclassified).
Preliminary analysis of the rodent carcinogenicity QSAR model indicated
that 43% of carcinogenicity model scaffolds matched compounds predominantly
positive for salmonella (and classified as genotoxic), while 30% of the
scaffolds matched compounds predominantly negative for salmonella (and classified
as non-genotoxic). The remaining 27% of the scaffolds were classified as only
partially responsible for genotoxicity. The analysis further classified scaffolds for
each of the rodent carcinogenicity models (mouse, rat, rodent, and sex) and identified
genotoxic and non-genotoxic scaffolds common across several different models.
1817 EFFECTS OF ORAL DOSAGE REGIMEN ON FIRST-PASS
ELIMINATION AND TOXICOKINETICS (TK) OF 1, 1, 1-
TRICHLOROETHANE (TRI) AND
TRICHLOROETHYLENE (TCE).
V. Srivatsan, S. Muralidhara, J. V. Bruckner and C. A. White. Pharmacology and
Biomedical Sciences, University of Georgia, Athens, GA.
The objective of this investigation was to compare the TK of TRI and TCE under
conditions representative of drinking water exposure, and to contrast these TK profiles
with those when the halocarbons are given by gavage. Fasted male Sprague-
Dawley rats were given 6 – 48 mg TRI/kg or 8 – 50 mg TCE/kg, as an aqueous
emulsion ia, iv or by constant gastric infusion (gi) over 2 h. Serial micro-blood sam-
ples were taken and analyzed by GC, while other rats were sacrificed serially for
blood and tissue collection. Bioavailability of TRI exceeded 95% and was independent
of dose and dosage regimen. This was not the case for TCE. First-pass
elimination of the 8 mg TCE/kg gi dose was virtually complete. Rapid absorption
of TRI ad TCE into blood and tissues was followed by relatively rapid elimination
of TCE. In vivo TRI blood:tissue partition coefficients were consistently higher
than corresponding values for TCE. First-pass pulmonary elimination of TRI was
independent of dose and dosage regimen and exceeded that of TCE, while the converse
was true for hepatic first-pass elimination. These TK data may prove useful in
physiological modeling and in assessment of health risks of these common drinking
water contaminants. Supported by U.S. DOE DE-FC09-02CH11109.
1818 IN VITRO METABOLISM OF BDE-99 IN HUMAN LIVER
MICROSOMES.
C. Erratico and S. Bandiera. Faculty of Pharmaceutical Sciences, The University of
British Columbia, Vancouver, BC, Canada.
Polybrominated diphenyl ethers (PBDEs) are persistent, bioaccumulative, and toxic
environmental pollutants frequently detected in human samples. Limited information
is available for PBDE metabolism in humans. In this study, metabolism of
2,2′,4,4′,5-pentabromodiphenyl ether (BDE-99), the major component of a widely
used commercial PBDE mixture (Penta-BDE), was investigated in vitro to determine
the hydroxy metabolites formed, rates of metabolite formation, and the cytochrome
P450 (CYP) enzymes involved. BDE-99 was incubated with human liver
microsomal preparations. Hydroxy metabolites were liquid-to-liquid extracted and
quantified using liquid chromatography/mass spectrometry. Incubation conditions
were optimized for substrate, NADPH, protein concentration, and incubation
time. BDE-99 was metabolized to two major (5′-OH-BDE-99 and 2,4,5-tribromophenol),
one intermediate (4′-OH-BDE-101), and two minor (4-OH-BDE-90
and 6′-OH-BDE-99) oxidative products by human liver microsomes. Rates of formation
of major and minor metabolites ranged between 18 and 20 and between 0.6
and 0.9 pmol/min/mg protein, respectively. No other hydroxy or di-hydroxy
metabolite of BDE-99 was detected under the experimental conditions used.
Structures of the major and minor BDE-99 metabolites produced by human liver
microsomes are different from structures of the major and minor BDE-99 metabolites
identified in a previous study using rat liver microsomes. Moreover, BDE-99
undergoes more extensive metabolism in human than in rat liver microsomes (50 vs
3.5 pmols/min/mg protein). Formation of BDE-99 hydroxy metabolites is of toxicological
concern as in vitro studies have recently shown that hydroxy PBDEs affect
steroidogenesis and are agonists of human estrogen and transthyretin receptors.
Experiments are in progress to identify the CYP enzymes responsible for the formation
of BDE-99 metabolites using a panel of human recombinant CYP enzymes.
The same experimental design will be used to characterize the metabolism of the
second major component of Penta-BDE mixture, BDE-47.
1819 BIOTRANSFORMATION OF ETHANOL TO ETHYL
GLUCURONIDE IN A RAT MODEL AFTER A SINGLE
ORAL DOSAGE.
T. Haupt and K. Ferslew. East Tennessee State University, Johnson City, TN.
Ethyl glucuronide (EtG) is a minor ethanol (Et) metabolite that confirms the absorption
and metabolism of Et after oral or dermal exposure. Human data suggest
maximum blood EtG concentrations are reached between 3.5 - 5.5 hours post Et
administration 1 . This study was undertaken to determine if the Sprague Dawley
(SD) rat biotransforms Et to EtG after a single oral dose of Et. SD rats (male, n=6)
were gavaged with a single Et dose (4g/kg) and urine was collected for 3 hours in
metabolic cages, followed by euthanization and collection of heart blood. Blood
and urine were analyzed for Et and EtG by gas chromatography and enzyme immunoassay.
Blood and urine Et concentrations (BEt, UEt) were 195 ± 23 and 218
± 19 mg/dL while blood and urine EtG concentrations (BEtG, UEtG) were 1363 ±
98 ng/mL and 2.1 ± 0.29 mg/mL (mean ±SE). Sixty-six, male, SD rats were gavaged
Et (4 g/kg) and placed in metabolic cages to determine the extent and duration
of Et to EtG biotransformation and urinary excretion. Blood and urine were
collected up to 24 hours post administration for Et and EtG analysis. Maximum
BEt, UEt, and UEtG were reached within 4 hours while maximum BEtG was
reached 6 hours post administration. Maximum concentrations were: BEt, 213 ±
20 mg/dL; UEt, 308 ± 34 mg/dL; BEtG, 2683 ± 145 ng/mL; UEtG, 1.2 ±
0.06mg/mL (mean ±SE). Areas under the concentration-time curve were: BEt,
1632 hr*mg/dL; UEt, 2936 hr*mg/dL; BEtG, 11764 hr*ng/mL; UEtG, 8.5
hr*mg/mL. BEt and BEtG were reduced to below limits of detection (LOD) within
12 and 18 hours post Et administration. UEt’s were below LOD at 18 hours, but
UEtG was still detectable at 24 hours post administration. Our data proves that the
SD rat biotransforms Et to EtG and excretes both in the urine and suggests it is
similar to that of the human.
SOT 2011 ANNUAL MEETING 389