|
Sigma Xi
|
CATEGORY A: BIOLOGICAL ENDPOINTS, BIOMARKERS, SURROGATE MARKERS, AND IMAGING TECHNOLOGIES
Menu
Display contrast changes with viewing-angle in medical AMLCDs: measurement methods and effect on observer detection performance
A. Badano, D. H. Fifadara, CDRH, FDA, Rockville, MD
The diagnostic value of medical imaging modalities that rely on display devices for interpretation, review, consultation or guidance for localization is affected by the display image quality. As a consequence, the assessment of image quality in medical- and consumer-grade display devices has become a key element in the characterization and quality control of local and teleradiology applications of diagnostic modalities. Increasingly for most consumer and high-end imaging applications, the active-matrix liquid-crystal display (AMLCD) is the device technology of choice due to its good image quality, small footprint, and low cost. However, and in spite of recent progress, LCD monitors are not Lambertian, and therefore display contrast varies significantly with viewing-angle. This has generated the need for manufacturers and users to develop methods for the characterization of the angular emission, and to study its implication on diagnostic performance. Currently, the two most widely used methods to measure the angular luminance response of AMLCD are the goniometric and the Fourier-optics methods. We show that while both methods have their advantages, there can be differences between them as large as 38%. The effect of these angular variations on visual detection tasks is then studied by measuring detection performance of human observers. The results prove that changes in display contrast translate into changes in subtle signal detectability.
Metabolomics analysis to find biomarkers of disease state
C. Beecher, Metabolon, Durham (RTP), NC 27713
Background: Metabolomics is an omics-style science that examines biochemical content of a biological sample to ascertain physiological status. When compared to "normal" tissues, disease tissues frequently demonstrate strong, statistically significant perturbations in their biochemical profiles that provide insight into the underlying pathology. We will highlight this approach with work done in collaboration with the NICHD (and for which we received the 2005 March of Dimes Research Award) on premature birth.
Methods: The total biochemical content of a biological sample is determined in order to understand the physiological status of the sample at the time the sample was collected. The detailed chemical measurements are derived through parallel mass spectral analyses of extracts of the biological sample. In this project, the samples consisted of amniotic fluid from women presenting with preterm labor. A statistical analysis revealed several classes of compounds that were found to be predictive of eventual outcome.
Results: The analysis of amniotic fluid of 113 women in pre-term labor revealed several classes of compounds that could serve as biomarkers for the medical outcome. The possible outcomes included premature birth, full term birth, or infection/inflammation. Since each of these outcomes has different medical requirements the ability to predict outcome was quite significant. The final results included many statistical approaches and was determined to have over 88% predictivity and excellent selectivity and sensitivity.
Conclusions: Metabolomic analyses show great promise in biomarker determination. In the case of pre-term labor, the markers discovered were highly predictive and medically significant.
NMR Metabonomics of Urine from Wistar Rats Dosed Separately with Kidney and Liver Toxicants
R. D. Beger1 , L. K. Schnackenberg1 , Y. P. Dragan1 , M. D. Reily2 , D. G. Robertson2 , 1NCTR, FDA, Jefferson, AR, 2Pfizer Global Research & Development, Ann Arbor Laboratories, Ann Arbor, MI
A collaboration between Pfizer and the Center of Metabolomics Research at NCTR was initiated to study the ability of NMR in combination with pattern recognition techniques to rapidly detect toxicity through analysis of urine samples. Single low and high doses of 4-aminophenol (PAP; 15 mg/kg, 150 mg/kg), galactosamine (Gal; 60 mg/kg, 600 mg/kg), and thiacetamide (THI; 20 mg/kg, 200mg/kg) were administered to male Wistar rats in groups of 4 for each toxin and dose level. Further, saline was administered to four rats as the drug vehicle control, for three toxins. Predose urine samples (Day 0) and samples from post-dosing (Day 2 and 3) were collected for each rat in metabolism cages at Pfizer and monitored by 1D 1H NMR at NCTR. Chenomx Eclipse software (Chenomx, Edmonton, Canada) was used to spectrally identify 198 metabolites in the 1D NMR spectra of urine. Partial least squares-discriminant function (PLS-DF) models of liver and kidney toxicity were developed from 8 metabolites. The metabolites used to build both the high and low dose PLS-DF models of toxicity were 3-hydroxyphenylacetate, homocysteine, 2-phosphoglycerate, 5-hydroxylysine, creatine, indole-3-acetate, sarcosine, and myo-inositol. The average of four training models of liver and kidney toxicity built from the high dose data was 98.2%. The average of four training models of liver and kidney toxicity built using the low dose data was 86.2 %.
Serological Attraction of Non-toxin Egg Protein to Staphylococcal Antienterotoxins
R.W. Bennett, CFSAN, FDA, CP, MD
Raw whole liquid and dried eggs which were purported to contain staphylococcal enterotoxin were evaluated by a number of ELISA-based methods. The initial evaluation was to establish whether the purported positive ELISA reactions were as a result of toxin antienterotoxin serological activity. Secondary considerations were to study some basic characteristics of the protein, determine whether the protein occurred in the yolk and/or white portions of the eggs, make differential observations of this protein in fertilized and non-fertilized eggs and to follow the fate of the protein during extended egg incubation at 37°C. The controls used in this study were eggs offered in retail outlets for human consumption. Three ELISA-based methods (manual polyvalent detection system, manual monovalent ELISA and an automated polyvalent enzyme-linked fluorescent immunoassay) were used to investigate this protein in eggs. To determine the part of the egg containing the protein, egg white and yolks were separated and processed separately in the same manner prior to analysis of the extracts for the protein. Generally, the ELISA results were instantaneous and strongly positive reactions with the manual and automated polyvalent systems suggesting non-specific binding of the egg protein to one or more antibodies to the staphylococcal enterotoxins. Further analysis with the monovalent ELISA showed positive reactions when egg extracts were tested separately for enterotoxins A, B, C, D, and E. In that the egg protein reacted with antibody serotypes A - E, separately, it was conclusive evidence that this protein from fertilized eggs had caused false-positive reactions and that the eggs did not, indeed, contain preformed staphylococcal enterotoxin; thereby the eggs could not be deemed to be adulterated.
Graphical User Interface (GUI) for mammography review and CAD development
S. Bhat, S. Paquerault, N. Petrick, CDRH, FDA, Rockville, MD
Introduction: Radiologists have long known that some breast cancers remain undetected on screening mammograms due to technical and human factors. To help reduce the number of missed cancers, researchers and commercial firms have developed computer-aided diagnosis (CAD) systems to aid the radiologists in detecting breast cancer. The FDA has a keen interest in understanding how these devices perform and under what situations they may systematically fail to detect lesions.
Method: In order to assist in our CAD research on these devices, we are developing a GUI for displaying mammograms and CAD results using the cross-platform scripting language Tcl/Tk. The main advantage of this programming language is that it can be ported to just about any UNIX, Windows and Macintosh operating system.
Results: Our proposed GUI has been designed to: 1) allow the user to review mammograms using a configurable display environment that allows the number of displayed images and mammographic views to be customized; 2) display the CAD detected lesions using either simple markers or the computer defined boundaries; and 3) interface directly with our CAD and evaluation software allowing researchers to quickly modify and evaluate new CAD algorithms and algorithm parameters (e.g., the classifiers, filters, and thresholds used by a CAD algorithm)
Conclusions: We have designed a GUI, which can be used both as an image review tool as well as an interface for quickly modifying and evaluating new CAD algorithms. We plan to further extend this CAD tool for 3D CT imaging as well.
Predicting clinical resistance to Gleevec treatment by in vitro applied stable isotope-based dynamic metabolic profiling (SIDMAP)
L. G. Boros, P. W-L. Lee, UCLA, Los Angeles, CA
Purpose: Gleevec has been marketed for the treatment of BCR-ABL+ leukemias after a fast-track approval by the FDA in 2001. In the meantime significant resistance to Gleevec developed partially due to an adaptation of the metabolic network to non-oxidative nucleic acid precursor ribose synthesis in the pentose cycle (Drug Discov. Today Therapeutic Strategies 1:435-443, 2004). Herein we describe the foreseen limitations of Gleevec using stable isotope-based dynamic metabolic profiling as an in vitro tool to indicate drug resistance and treatment failure.
Methods: Lama84 myeloid leukemia cells obtained from Gleevec resistant patients were incubated with the [1, 2-13C2]-D-glucose tracer for 72 hours in the presence of 1000 nM Gleevec and metabolic network fluxes were compared with SIDMAP profiles obtained 5 years ago (J. Biol. Chem. 276:37747-53, 2001).
Results: Lama84 cells obtained 87% of their nucleic acid precursor ribose via the non-oxidative branch of the pentose cycle, while previous SIDMAP studies indicated a dose dependent increase in non-oxidative ribose synthesis by 5.2%, 12.1% and 19.3% in the presence of 100 nM, 500 nM and 2000 nM Gleevec treatments, respectively.
Conclusions: Clinical resistance to Gleevec in leukemic cells highly expressing the intact BCR-ABL tyrosine kinase construct is attributable to the up-regulation of non-oxidative ribose synthesis thereby escaping the drug's strong inhibitory effect on oxidative ribose-phosphate synthesis. SIDMAP studies, which accurately predicted the use of alternative nucleic acid precursor synthesis pathways five years ago, would assist the FDA in the approval of targeted drugs by addressing metabolic adaptation and limitations before human trials begin.
Comparative Genomic Hybridization Using Oligonucleotide Microarrays and Total Genomic DNA
L. Bruhn1 , N. Sampas1 , M. Barrett1 , A. Ben-Dor1 , P. Anderson1 , A. Scheffer1 , P. Tsang1 , C. Gooden2 , R. Walker3 , B. Curry1 , R. Kincaid1 , D. Lipson1 , M. Bittner2 , P. Meltzer3 , Z. Yakhini1 , S. Laderman1 , 1Agilent Technologies, Palo Alto, CA, 2Translational Genomics Research Institute, Phoenix, AZ, 3National Human Genome Research Institute, Bethesda, MD
Array-based comparative genomic hybridization (aCGH) measures copy number variations at multiple loci simultaneously, providing an important tool for studying genomic alterations associated with cancer and developmental disorders as well as germ-line copy number polymorphisms. The broadest utility of aCGH, including both a simplified preparation of targets and hybridization of samples to any array design of interest, requires preserving the greatest possible complexity of targets derived from whole genome samples. Therefore, we developed probe design criteria, assay conditions, and analysis methods that enable 60-mer oligonucleotide microarrays to be used for CGH measurements using total genomic DNA (Barrett, MT, et al., PNAS 101, 17765-17770 (2004). We designed a 60mer oligonucleotide array with 40K probes specifically designed for CGH representing sequences throughout the human genome with a bias for known and predicted gene loci. We tested the performance of this array for reproducibly measuring and mapping losses, and amplification events of varying levels and sizes using both unamplified and phi29 amplified total genomic DNA from a series of model systems. The mean slope of experimental versus theoretical log-ratios for chromosome X probes on this array in XY versus XX hybridizations typically exceeds 0.9, with probe by probe error rates of less than 10% in the separation of their log-ratio distributions. To further validate the performance of this platform we used well characterized cell lines, including diploid cells with partial deletions in chromosome 18q, and diploid and aneuploid tumor cell lines with known amplification and deletion events. We show that the highly processive DNA polymerase phi29 can be used to prepare aCGH templates from as little as 5-10ng of starting material that yield high quality aCGH measurements throughout the genome. Phi29 provides a simplified isothermal method for amplifying limiting material. However, nonspecific DNA fragments of high MW are generated in the absence of sufficient input template. To better ensure reproducible and robust aCGH assay quality, we have developed methods and protocols using the Agilent BioAnalyzer to enable quality control and accurate quantitation for key pre-hybridization steps including: phi29 amplification of genomic samples, restriction digestion of templates and target labeling. We have also developed visualization tools and statistically robust computational tools that take into account the estimated errors on the measured log ratios in mapping aberration boundaries, and for identifying common aberrations across multiple samples. We tested the reproducibility of our platform by using tumor cell line samples including the colon adenocarcinoma cell line HT29 in hybridizations performed in different laboratories (Agilent Labs, NHGRI, TGen). We present results, using these methods, demonstrating that in situ synthesized 60-mer oligonucleotide arrays can reproducibly detect genomic lesions including single copy and homozygous deletions, and variable amplicons throughout the genome using full complexity genomic DNA samples.
Metabolite profiling for biomarker discovery and applications
K. Busch1 , T. Walk1 , B. Bethan1 , V. Haake1 , R. N. Trethewey1 , A. J. Krotzky2 , 1metanomics GmbH & Co. KGaA, Berlin, Germany, 2metanomics GmbH & Co. KGaA & metanomics Health GmbH, Berlin, Germany
Background
Metabolite profiling, the parallel analysis and bioinformatics interpretation of hundreds of metabolites and thousands of metabolome signals is applied to discover novel biomarkers to support all phases of drug development and nutrition.
Methods
metanomics leading metabolite profiling technology platform combines >50 mass spectrometers (GC-MS and LC-MS/MS) with fully integrated powerful data mining systems. This technology is being applied to a variety of sample types like blood, urine and tissues. Each sample analysis reliably covers several hundreds of known and unknown metabolites and yields additional several thousand metabolite signatures. The bioinformatics platform includes a wide range of univariate and multivariate data mining techniques and innovative data bases to discover single and complex biomarkers. The proven technology is being utilized in pre-clinical and clinical studies, accelerates necessary decisions, improves risk management and disease monitoring.
Results
Results are presented from metabolite profiling in blood matrices for mode of action and biomarker discovery in the context of toxicology studies and diabetes research. It is shown that metabolite profiling conclusively differentiates between different drugs, gender responses and between dose groups as well as identifies biomarkers for diseases such as diabetes type 2. The clear identification of key marker metabolites enables interpretation of biological contexts, underlying pathways and mechanisms related to drug effects.
Conclusions
Mass spectrometry based metabolite profiling is a powerful and reliable technology that supports all phases of drug development and toxicology. Very broad and in parallel highly sensitive and precise metabolite analysis enables biomarker discovery and allows direct insights into mode of action, underlying pathways.
Molecular Network Analysis of Metastatic Colorectal Carcinoma Using Reverse Phase Protein Microarray Based Phosphoproteomic Portraits
V. S. Calvert1 , K. M. Sheehan2 , E. Mammano3 , C. Belluco3 , V. E. Espina2 , J. D. Wulfkuhle2 , D. Nitti3 , M. Lise3 , L. Young4 , P. J. Munson4 , L. A. Liotta2 , E. F. Petricoin1 , 1CBER, FDA, Bethesda, MD, 2CCR, NCI, Bethesda, MD, 3University of Padova, Padova, Italy, 4CIT, NIH, Bethesda, MD
Protein kinases represent some of the most important drug targets in medicine today since aberrant kinase activity underpins many disease states. In the past it has been difficult to interpret kinase activity in tissue because extraction of the kinase separates it from surrounding regulatory context. Past attempts at cell network analysis have relied on gene microarray analysis, which does not recapitulate what is occurring at the proteomic level. On the other hand, the record of ongoing kinase activity states and cell signaling processes are imprinted on their phosphorylated protein substrates. An understanding of the pathways responsible for tumor survival are critical for developing novel targeted therapies tailored for the metastatic signature. To begin understanding the critical clinical question of whether and to what extent signaling network changes upon metastasis, we exploited two colorectal cancer related study sets in which patient-matched colorectal cancer specimens along with hepatic and lung metastatic lesions were analyzed by reverse phase protein microarrays. Unsupervised clustering analysis were developed using 40 different phospho-specific signaling endpoints. Phosphoproteomic network analysis was performed to find correlates with the metastatic phenotype compared to the primary portrait. Our results indicate that, unlike analysis of gene microarray data, we observe a significant difference between the metastatic lesion and the patient matched primary tumor at the phosphoproteomic level. This study represents the first reported attempt at the systematic functional mapping of cellular signal networks in the metastatic setting, which will be critical to realize the promise of patient-tailored therapy.
Development of a mammalian retinal ganglion cell model of electrical stimulation.
E. D. Cohen, D. D. Chan, CDRH, FDA, Rockville, MD
Background: Many retinal implants for the blind rely on electrical current stimulation of the retina for eliciting the sensation of vision. These designs use retinal ganglion cells, stimulated either directly or indirectly by electrical current, to relay action potentials to the brain. To evaluate their safety and efficacy, we must understand the effects of electrical current stimulation on different ganglion cell types involved in central vision. A well-studied model is cat X and Y ganglion cells. The light-evoked firing of X-cells is sustained, while firing of Y-cells is transient.
Methods: We are developing a model system for simulating electrical current stimulation of ganglion cells using real physiological and anatomical data. Neuronal models of ganglion cells were generated from dye-filled recorded neurons using a Neurolucida anatomical reconstruction system (Courtesy McBain Lab, NINCDS/NIH), and imported into the modeling program NEURON. A five-channel, voltage-gated conductance model was used. We compared real firing patterns from electrical current-stimulation to the model's firing patterns.
Results: The firing patterns of Y-cells to injected current showed more spike adaptation than X ganglion cells. The total membrane area of dendrites of the two ganglion cell types also differed by approximately 2-fold, with Y-cells having larger areas. The axon diameters of Y-cells were twice that of X-cells, as were cell body diameters.
Conclusions: X and Y ganglion cell types differ in their axon diameters, firing patterns and physical membrane area, which may generate different responses to electrical stimulation in retinal implants.
Automation: An Approach to Standardizing Multi-Functional Cellular Profiling Assays in Clinical Trials
C. Smith, J. Wilkinson, S. D'Costa, E. Rabellino, Custom Biopharma Solutions, Biomedical Research Division, Beckman Coulter Inc.
Background
Functional immune cell-profiling assays have been evaluated as surrogate end points to determine the efficacy of vaccine or therapeutic strategies for clinical trials. However, their utility has not been fully realized due to the variable nature of the assays which is particularly pronounced in T cell functional assays. With a view to reducing variability and standardizing targeted steps of T cell functional assays, an automated methodology for simultaneous staining and analysis of multiple intracellular cytokines and cytotoxicity markers via flow cytometry was developed and validated.
Method
To verify the concept of improved standardization with automation, a modification of available sample preparation instruments that enabled automated pipetting, incubation, staining and analysis of intracellular and surface molecules via flow cytometry was devised. A 5-color flow cytometry assay (2-3 surface markers; 2-3 intracellular cytokines/cytotoxic markers) was developed to characterize the T-helper and T-cytotoxic responses in human peripheral blood mononuclear cells (PBMCs) stimulated in a restricted polyclonal manner by SEB/CD28, and by an antigen-specific peptide pool.
Results and Conclusions
The automated sample staining and analysis greatly improved both inter and intra assay precision while reducing the "hands-on" sample preparation and analysis time. The complex nature of the immune response in PBMC specimens was revealed on evaluation of the cytokine and cytotoxic marker combinations in the context of multiparametric T cell response evaluations. Comparable approaches can be applied to other cellular participants such as Natural Killer T, Natural Killer and Dendritic cells. The use of automation thus provides a greater degree of standardization in complex cell functional assays enabling a correlation of the complex immune response profiles to vaccine efficacy or disease progression without interference due to the variable nature of the manual assays.
A Systems Biology Approach to Biomarker Discovery Utilizing Integrated Cytomic and Proteomic Techniques
S. D'Costa1 , C. Snow1 , E. Betgovargez2 , E. Rabellino1 , M. Simonian2 , 1Custom Biopharma Solutions, Biomedical Research Division, Beckman Coulter Inc., 2Proteome Lab Development, Biomedical Research Division, Beckman Coulter Inc.
Purpose.
A multi-pronged approach needs to be pursued to identify and evaluate relevant biomarkers. This requires that assays that have so far been carried out in isolation, for e.g. gene expression, protein synthesis, phenotype and function, etc. need to be integrated to better understand the "complex" responses that are associated with biomarker discovery. This "systems biology" approach can to some extent be achieved by unified genomic, proteomic and cytomic analyses. In the current study, the authors have attempted such an evaluation by integrating well-known techniques of flow cytometry-based phenotypic analysis and cell sorting with protein fractionation and analysis to identify biomarkers of cellular activation.
Methods.
Peripheral blood mononuclear cells were subjected to "restricted"polyclonal stimulation with staphylococcal enterotoxin B (SEB), fluorescently labeled with antibodies to distinguish activated and non-activated cells and subsequently isolated using flow-based sorting techniques. Lysates of activated and non-activated cells as well as cell-free supernatants were fractionated by two-dimensional gel-free liquid chromatography. Intact proteins were separated by their isoelectric points in the first-dimension and these fractions were further separated by hydrophobicity on a second-dimension. Using powerful and easy to use differential display software a high resolution protein profile of the complex mixture was obtained and analyzed.
Results and Conclusions
Qualitative and quantitative differences in protein profiles in activated and non-activated cells and supernatants could be identified thus allowing for easy discovery of putative protein/peptide biomarkers. The liquid-phase fractions of proteins in the native state, permits direct characterization of the differentially-expressed proteins, and thus accomplishes a thorough identification and analysis of biomarkers for monitoring and measuring cellular functional responses. Further combinations of genomic, proteomic and cytomic profiling will accomplish a unified evaluation of biomarker discovery and validation relevant to this response and other research models
Cloning and Analysis of Mono-Specific Single Chain Fragment Variable (scFv) Fragments that Recognize German Cockroach Allergens Bla g 1, Bla g 2, Bla g 4, and Bla g 5
N. C. DeVore, W. JJ. Finlay, E. N. Dobrovolskaia, J. E. Slater, CBER, FDA, Bethesda, MD
Rationale. Cockroach allergens may be an important cause of inner city asthma.Although many allergens have been standardized, cockroach extract has not. We have shown (Clin. Exp. Allergy 2002; 32:721-727) that there is a wide disparity in potency and allergen content among commercially available cockroach extracts. As part of our efforts to standardize cockroach extracts, we have produced scFvs that recognize Bla g 1, Bla g 2, Bla g 4 and Bla g 5.
Methods. Three 6-month-old female white leghorn chickens were immunized with rBla g 1, rBla g 2, rBla g 4, and rBla g 5 (Indoor Biotechnologies) using Freund's adjuvants. The animals were sacrificed, and RNA was isolated from spleens and bone marrow. Libraries were constructed from amplified cDNA, and scFv recombinant antibodies were produced from the antibody repertoire of these chickens. Phage bearing these scFvs were panned against rBla g 1, rBla g 2, rBla g 4, and rBla g 5. Four clones from each library were identified as recognizing the specific recombinant Bla g.
Results. Anti-Bla g 1, anti-Bla g 2, anti-Bla g 4, and anti-Bla g 5 scFvs specifically recognize rBla g 1, rBla g 2, rBla g 4, and rBla g 5, respectively, by ELISA and Western blot analysis. The antibodies also recognize specific proteins in crude cockroach extracts.
Conclusions. We have generated scFvs that specifically recognize each of the recombinant Bla g proteins. These recombinant antibodies will be useful in developing assays for the standardization of cockroach extracts.
A Low Molecular Weight, Orally Active TpoR Agonist, SB-497115, Does Not Prime Platelets for Activation or Agonist-Induced Aggregation In Vitro
J. Erhardt1 , P. Tapley2 , C. L. Erickson-Miller2 , 1GlaxoSmithKline, King of Prussia, PA, 2GlaxoSmithKline, Collegeville, PA
SB-497115 is a small molecular weight, orally active molecule that requires Tpo receptor (TpoR) expression for activity and activates JAK/STAT signalling in platelets. When administered to healthy male subjects at oral doses of 30 mg and above for 10 days a dose dependent increase in the platelet count was observed. The ability of recombinant Tpo to prime platelets for enhanced agonist-induced aggregation is well-documented, thus this study was undertaken to compare the ability of rhTpo (150 ng/ml) and SB-497115 (0.01-10 uM) to affect platelet activation. Platelets were obtained from healthy human volunteers, following written informed consent, by standard venipuncture technique. Platelet aggregation was induced by a submaximal concentration of adenosine diphosphate (ADP) (1.0 uM) in experiments examining synergy with rhTpo (150ng/mL) or SB-497115 (0.01-10 uM). For examination of potential inhibitory actions of SB-497115 on platelet function, aggregation was induced by ADP (3 uM), collagen (2 ug/mL), or the thrombin receptor activating peptide (TRAP; 20 uM). Pre-treatment of platelet samples with rhTpo, but not SB-497115, potentiated the effects of 1.0 or 1.5 uM ADP. No inhibitory effect on normal ADP, Collagen or TRAP-induced aggregation was seen with either SB-497115 or rhTpo. The expression of platelet P-selectin, an early marker of platelet activation, was analyzed by flow cytometry. Treatment with SB-497115 (0.1 - 10 uM) demonstrated no induction of platelet CD62P expression above background levels, while rhTpo produced a modest but significant and reproducible increase in the percent of CD62P+ platelets. This study demonstrates that the orally active small molecular weight agonist of the Tpo receptor, SB-497115, does not mimic the ability of rhTpo to enhance ADP-induced aggregation or to induce P-selection expression on human platelets. Finally, SB-497115 has no antagonistic effect on ADP-, TRAP-, or collagen-induced platelet aggregation. Thus, SB-497115 is differentiated from rhTpo with respect to priming platelets for activation.
AGE-RELATED DIFFERENCES IN SUSCEPTIBILITY TO TOXIC EFFECTS OF VALPROIC ACID IN SPRAGUE-DAWLEY RATS
P. Espandiari, T. J. Miller, A. Knapton, J. Zhang, J. Hanig, CDER, FDA, Silver Spring, MD
Prediction of adverse drug effects in children is a timely issue driven by recent legislation specifically mandating safe and effective drugs for children. Doses for children are often adult doses normalized to appropriate pediatric parameters in the clinic. They generally do not take cognizance of very special developmental differences that may produce significant vulnerabilities in vital target organs of children that can result in serious toxicities that are atypical in adults. Valproic acid (VPA) has been reported to cause several cases of fatal hepatotoxicity in infants and temporary increases in serum transaminases (~ 40% of patients). In this study, different age groups of male Sprague-Dawley (SD) rats (25, 40, 80 days old) corresponding to human toddlers, young and mature adults were treated with VPA at doses of 120, 360, 500, or 650 mg/kg i.p. injection for 4 days to determine pediatric vs. adult susceptibility. The order of sensitivity to the toxicity of VPA observed in this study was 25>80>40 days. The most dramatic toxic effects were observed in 25 day old rats at 650 mg/kg VPA: 57% survival rate, early elevations in ALT, serious G.I. disturbances, and white blood cell deficiencies. In addition, histopathology at this dose showed focal necrosis and apoptosis in the liver. Decreased serum platelet counts and diminished rate of growth were observed at all doses. In comparison, no lethality occurred in the 40 and 80 day old rat groups, and a decreased growth rate was detected in 40 day rats with all treatments. There was also an absolute weight loss in the highest dose groups at 40 and 80 days as well as the middle dose group at 80 days. This correlated with lowering of serum total protein and white blood cell counts in these age groups. Focal necrotic lesions and apoptotic cell death were observed in the liver of 80 day old rats treated with the two highest dose levels. These data indicate that the pattern of toxicity of VPA in different age groups of SD male rats is quite dissimilar. Further studies will explore these findings and its implications for pediatric drug safety.
Pregabalin: Pharmacokinetics, Pharmacodynamics and Interspecies Scaling
R. Feng, J. Koup, L. Radulovic, H. Bockbrader, P. Worboys, Pfizer
Background: Pregabalin (PGB, Lyrica), a derivative of the inhibitory neurotransmitter γ-aminobutyric acid, is clinically effective in the treatment of neuropathic pain, epilepsy, and anxiety. In current work, PGB pharmacokinetics and pharmacodynamics in preclinical species and interspecies scaling were discussed.
Methods. Plasma and/or tissue concentrations were determined in mice, rats, and monkeys following PGB administration. Pharmacokinetic/pharmacodynamic modeling was performed using NONMEM.
Results. PGB was rapidly absorbed in animals and mainly excreted in urine as unchanged. PGB renal clearance and volume of distribution in humans are well predicted by preclinical species. A minor metabolite in mouse and rat urine was identified as the N-methyl metabolite. In dogs, approximately 45% of the PGB dose was excreted in urine as N-methyl metabolite. No significant inhibition of human CYP1A2, 2A6, 2C9, 2C19, 2D6, 2E1, or 3A4 was observed in vitro at PGB concentrations up to 1000 µM, suggesting low probability for PGB to elicit drug-drug interactions through inhibition of CYP450 isozymes. PGB exhibited potent activities in rats in the maximum electroshock seizer test for epilepsy, in chemical allodynia models for neuropathic pain, and in Vogel conflict test for anxiety. The estimated EC50 values ranged from 1.9 - 3.7 µg/mL in the rat models, which are consistent with the steady state average PGB concentrations (1.6 - 5.6 µg/mL) in humans at efficacious doses of 150 - 600 mg/day.
Conclusion. The projected human pharmacokinetic parameters and the efficacious concentrations are consistent with clinical data.
JIFSAN STUDENT PROJECT: DEVELOPMENT OF IMMUNOHISTOCHEMICAL STAINING TECHNIQUES FOR PCNA AND APOPTOSIS TISSUE MARKERS ON RAT TISSUES IN THE CFSAN PATHOLOGY LABORATORY.
M. Ferri1 , S. Francke-Carroll2 , B. Magnuson3 , 1College of Live Sciences, UMD, College Park, MD, 2CFSAN, FDA, College Park, MD, 3Dept. of Nutrition and Food Science, UMD, College Park, MD
Purpose: The purpose of this one-year JIFSAN student project was to establish immunohistochemical (IH) staining capabilities at the CFSAN/Pathology Laboratory (PB).
Background: The markers that were chosen for method development were PCNA (proliferating cell nuclear antigen) and Apoptosis. At the start of this JIFSAN project, the IH capacity of the PB laboratory was limited, consisting of a time intensive, manual staining procedure. IH positive staining antigen could often not be demonstrated because often sub-optimally fixed tissues are submitted to Pathology for IH staining while advanced antigen retrieval procedures were not available. Another problem area was the degree of unspecific background staining common in laboratory rodent tissues due to endogenous biotin.
Methods: Various instruments and reagents were adopted into a standard protocol, including a commercial decloaker device and a semi-automated-stainer. Best results were obtained using a conjugated secondary antibody, which reduced background staining significantly.
Results: During the one-year JIFSAN project, standard operating protocols for IH with PCNA and Apoptosis were developed in the PB laboratory. Both protocols were successfully applied in experimental studies by a collaborating laboratory at the University of Maryland.
Conclusion: The developed protocols will serve as the basis for development of a wide variety of other tissue markers requested by principal investigators for IH staining of experimental studies.
Analysis of Potentially Pathogenic Phenotypes of Cell Membrane Microparticles in Apheresis Platelets
M. P. Gelderman, L. B. Carter, J. Simak, LCH, CBER, FDA, Rockville, MD
Purpose: Cell membrane microparticles (MP) are released into the circulation from stimulated platelets, blood or endothelial cells. Their elevated counts have potential for different pathogenic activities.
Methods: We analyzed MP in apheresis platelet units (APU, n=26) on day 6 of storage using a three-color flow cytometry assay. Plasma from healthy volunteers (HVP, n=12) was used for comparison. To identify the cellular origin of MP, we used monoclonal antibodies to platelets (CD41), WBC (CD45), RBC (CD235a), and endothelial cells (CD105, CD144). Procoagulant phenotypes were also analyzed; MP expressing tissue factor (CD142) and binding of annexin V (AVB).
Results: We found in APU not only a higher CD41+MP count, but also a two-fold increase of CD235a+MP (p<0.001) and a 5x increase of CD45+MP (p=0.018) vs. HVP. In addition, MP of endothelial origin (CD105+CD45-MP) were 3x higher in APU (2790/µL; 840-9870) vs. HVP (930/uL; 710-1930; p=0.002). Surprisingly, the count of CD142+MP (4670/µL; 1830-11150) in APU was 3x higher vs. HVP (1580/µL; 1070-2510; p<0.001). CD142+MP were associated with WBC (20%), platelets (50%), or endothelial cells (20%) antigens. Counts of CD142+AVB+MP were 2080/µL (730-8040) in APU, and 1330/µL (720-1700; p=0.042) in HVP. In addition, the overnight incubation of washed APU MP with HUVEC resulted in a significant increase in cell expression of CD54 and CD142.
Conclusions: We found significantly elevated counts of platelet, WBC, RBC, endothelial cell derived MP, and MP of procoagulant phenotypes in APU when compared to HVP. Further studies to investigate a possible correlation between these findings and APU transfusion related adverse events are warranted.
 | Sigma Xi Outstanding Poster Award - 2005 FDA Science Forum |  |
A Subclinical Renal Injury Model in Rats for Detection of Nephrotoxic Compounds
P. L. Goering1 , E. F. Madden1 , R. P. Brown1 , B. A. Rosenzweig2 , K. L. Thompson2 , 1CDRH, FDA, Silver Spring, MD, 2CDER, FDA, Silver Spring, MD
Background: The objective of this study was to determine if rats with subclinical renal injury (SRI) are more susceptible than healthy control animals to nephrotoxic metals that exert their effect on different nephron segments.
Methods: To induce SRI, rats were dosed with gentamicin (sc) for 3 days. Twenty-four hr after the third injection, SRI- and saline-pretreated rats were injected with a single NOAEL or LOAEL dose of chromium (sc), mercury (iv) or lithium (ip) salts, metals which typically exert effects on proximal convoluted tubules (S1/S2 segments), proximal straight tubules (S3 segment), and distal tubules, respectively. The following biomarkers were evaluated 24 hours after metal injection: blood urea nitrogen (BUN) and creatinine; urinary N-acetyl-ß-glucosaminidase (u-NAG), creatinine, and total protein (u-TP); and kidney levels of kidney injury molecule-1 (KIM-1) mRNA.
Results: BUN levels were increased by 1.5-, 1.4-, 1.7-fold and NAG/Cr levels were elevated by 3.4-, 5.7-, and 2.2-fold in SRI rats challenged with chromium, mercury, and lithium, respectively, compared to control rats challenged with these metals. Kidney levels of KIM-1 mRNA were elevated due to gentamicin alone and elevated to a higher level in SRI rats treated with mercury, but not lithium and chromium.
Conclusions: SRI model rats demonstrated increased susceptibility to the effects of several nephrotoxic metals that target different nephron segments, thereby increasing the utility of the model for preclinical evaluation. U-NAG appeared to be the most sensitive biomarker to these metals, and kidney KIM-1 expression was elevated only in mercury-treated rats but not chromium- or lithium-treated rats.
Development and Validation of a High Content Cell Based Assay Using Imaging Technologies for the Detection of Neutralizing Antibodies to a Therapeutic Protein in Human Serum
J. J. Pennucci, S. J. Swanson, S. Gupta, Amgen Inc.
Purpose: To describe the development and validation of a cell-based bioassay for the detection of neutralizing antibodies to a therapeutic protein (TP) in human serum. The assay endpoint was based on cell imaging using fluorescence-based reagents incorporated into and at the cell surface.
Methods: The assay uses a human epidermoid carcinoma cell line that responds to a growth factor (GF) by signaling through a receptor that has tyrosine kinase activity (RTK) and autophosphorylates upon stimulation. Autophosphorylation is followed by activation and translocation of a transcription factor STAT-1 from the cytoplasm into the nucleus. The TP is an antagonist to the receptor and blocks GF binding resulting in lower RTK activity and STAT-1 translocation. A rabbit polyclonal affinity-purified neutralizing positive control antibody (NAb) to the TP restores GF binding and thus RTK activity and STAT-1 translocation. The cell line's response to the GF and TP was evaluated in the presence and absence of NAb at different concentrations of pooled human serum. TP inhibition curves and NAb titration curves were validated in 10% human serum. To determine the assay cutpoint,100 healthy donor sera (50 males and 50 females) were screened in the assay and the 95% prediction limit of the ratios of donor sera to pooled normal serum for the tested population was calculated. NAb spike-recovery experiments using 125 ng/mL NAb were conducted with 50 healthy donor sera.
Results: 4 ng/mL of GF elicited a robust tyrosine phosphorylation of the receptor and STAT-1 translocation in the cells. TP inhibited the GF-induced response in a dose dependent manner. A concentration of 100ng/mL TP was selected to hold constant in the assay for NAb detection. The limit of detection of the assay was determined to be 125 ng/mL NAb (in undiluted serum) which caused a significant inhibition of the RTK activation and STAT-1 translocation induced by 100 ng/mL TP.
Conclusions: Fluorescence based cell imaging methods for the detection of receptor tyrosine phosphorylation and STAT-1 translocation provided a novel multiplexed endpoint for use in a NAb assay.
The utility of monitoring cardiac troponin T to detect cardiac injury induced by low doses of isoproterenol in rats
E. H. Herman1 , J. Zhang2 , A. Knapton1 , N. Rifai3 , F. Sistare1 , 1DAPR, CDER, FDA, Silver Spring, MD, 2DAPR, CDER, FDA, Silver Sring, MD, 3Children's Hospital, Boston, MA
Backround: Cardiac troponin T (cTnT) is released from damaged myocytes and has been utilized as an biomarker of cardiac injury. In Previous studies high doses (4 to 80 mg/kg) of isoproterenol (Iso) caused acute cardiac lesions and increases in the serum levels of c TnT. The present study explored the utility of monitoring serum cTnT to detect cardiac injury at much lower Iso doses.
Methods: Groups of SD rats (8 wks) were dosed once (sc) with 0.008 to 0.5 mg/kg Iso and euthanized 3 to 48 hours later. Lesion severity was assessed on a scale of 0 to 5 (based on the degree of necrosis and apoptosis, inflammation, edema basement membrane damage and collagen deposition). cTnT levels were determined by Elecsys STAT immunoassay.
Results: cTnT levels were elevated in all animals (mean 0.20 to 0.28 ug/ml) 3 hours after administration of 0.008, 0.016, 0.032 or 0.064 mg/kg Iso but minimal cardiac lesions (severity score=1) were found in only 3 of 19 rats given the 0.032 and 0.064 mg/kg doses. However, by 6 hours cardiac lesions (severity score=1) together with increases in cTnT (0.13-0.18 ug/ml) were found in 20 of 22 animals given these same 2 Iso doses. Doses of 0.125, 0.250 or 0.50 mg/kg Iso caused both significant increases in in cTnT levels and marked myocardial alterations that reached a peak 3 and 6 hours post dosing. At 3 hours mean severity scores of 3.4, 4.2 and 5 and mean cTnT levels of 2.7, 6.8 and 13.6 ug/ml were observed in rats given 0.125, 0.25 and 0.50 mg/kg Iso respectively. The magnitude of myocardial alterations induced by these doses of Iso declined between 12 and 48 hours post treatment (mean lesion severity was 1.6-1.7 and cTnT levels were 0.06-0.20 ug/ml at 48 hours).
Conclusions: These results indicate that very small single doses of Iso are capable of causing time-and dose-dependent myocardial alterations and these alterations are associated with detectable increases in serum levels of cTnT.
Resolution of clofibrate efficacy and toxicity in rats using a combined LC/GC/MS metabolomics biomarker approach
A. Berger, I. Shah, A. J. Higgins, Icoria, Inc., Research Triangle Park, NC
Background: Clofibrate is a hypolipidemic drug (PPARα agonist) that has also been shown to produce hepatotoxicity. Our aim was to determine if the two actions could be distinguished using metabolomics.
Methods: Clofibrate or vehicle was administered orally (0, 50, 250 mg/kg/d) to groups of 6 rats and serum and livers were collected 6 and 24 h after either a single or 14 daily doses. Global biochemical profiles were determined by LC/MS and GC/MS.
Results: There was a dose and time-related increase in liver weights and mitotic figures, but no changes in serum enzymes. Recovery of biochemical profiles from clofibrate exposure occurred at 24 h, and was more pronounced after 14 days, suggesting adaptation. Stepwise regression was used to determine components contributing most to liver hypertrophy; and these components were putatively identified and mapped to pathways, highlighting potential mechanisms involved in liver hypertrophy. When combined with GC/MS profiles, with NIST 2002 library matching, we detected components and pathways related to clofibrate efficacy, such as fatty acid oxidation. Other changes observed were in amino acid and polyamine metabolism, pathways that might relate more to toxicity.
Conclusions: Global LC/GC/MS metabolomics, in combination with multivariate statistics, detected sets of liver biomarkers that showed adaptive changes, and that were associated with both efficacy (lipid metabolism) and toxicity (liver hypertrophy). This demonstrates the potential for a functional metabolomics approach to separate on-target and off-target drug effects.
Supported in part by NIEHS SBIR Award# 291200445524C and NIST ATP Award# 7UNANB2H3009
Cdc42 as a potential therapeutic target for the reduction of cellular proliferation and migration as measured in MDA-MB-231 breast cancer cells
D. S. Hirsch, Y. Shen, W. J. Wu, OBP, DMA, CDER, FDA, Bethesda, MD 20892
Background: High proliferative index and invasive capacity are two indicators of poor prognosis in breast cancer. Cdc42, a member of the Ras superfamily of small GTPases, is a GTP binding protein that regulates cell proliferation and migration. Several studies indicate that increased Cdc42 protein levels are found in breast cancer tissue as compared to normal breast tissue. The biological significance of increased Cdc42 protein levels is not known. We tested the hypothesis that reduction of Cdc42 protein levels in an aggressive breast cancer cell line, MDA-MB-231, would inhibit cellular processes critical to the breast cancer progression.
Methods:
A series of breast cancer cell lines were screened for Cdc42 protein levels by western blot. Activated Cdc42 protein levels were determined using GST-p21 binding domain assay, an affinity chromatography assay that specifically detects activated Cdc42 protein levels. Migration was studied using boyden chamber and wound healing assays.
Results:
Higher levels of activated Cdc42 were found in MDA-MB-231 cells, an aggressive cell line, as compared to a less aggressive breast cancer cell line known as MCF7. Cdc42-specific siRNA reduced Cdc42 protein levels by greater than 50% in MDA-MB-231 cells. siRNA-mediated reduction in Cdc42 protein levels resulted in reduced proliferation and cell migration in MDA-MB-231 cells.
Conclusions:
Cdc42 is a critical regulator of both cell proliferation and cell migration. Targeted introduction of Cdc42-specific siRNA to cancer cells may offer a novel therapeutic approach to inhibit these processes. Future studies are aimed at determining the suitability of Cdc42 as a therapeutic target.
How Biomarkers can Improve Clinical Drug Development?
P. R. Jadhav1 , M. U. Mehta2 , J. V.S. Gobburu2 , 1CDER, FDA, Rockville, MD and MCV/VCU, Richmond, VA, 2CDER, FDA, Rockville, MD
Purpose. To provide an overview of the utilities of biomarkers to streamline important decisions during drug development supported by relevant examples.
Methods. Recently, use of biomarkers has acquired a lot of attention for its utility in drug development process. A critical literature review was undertaken to evaluate uses of biomarkers towards rational drug development.
Results. A wide variety of biomarkers are employed in the early drug development to screen compounds, in the diagnosis of patients and for follow- up of drug effects routinely. Additionally, biomarkers such as blood pressure, cholesterol, testosterone levels, tumor size and rate of gastrointestinal polyp formation have been the basis of drug approvals. As long as a biomarker has a reasonable mechanistic basis, there are several important contributions of biomarker( s) towards improving the efficiency of drug development, irrespective of whether the relationship with the clinical outcome( s) is formally established or not. Typically, biomarkers can aid in
Conclusions. Selection of the dose range and doses for further investigation in the pivotal trials is the single most important use of biomarkers. Utility of biomarker knowledge in identifying patients with important differences and in making various regulatory decisions was also reviewed. Better appreciation of the underlying mechanism of disease and drug action, routine collection of biomarker data and sophisticated analyses as early as possible in drug development might lead to optimal drug therapy and more efficient use of available resources.
Vaccinia Immune Globulin Treatment in a Mouse Model of Progressive Vaccinia
M. C. Kennedy1 , N. Eller1 , J. Manischewitz2 , H. Golding2 , D. E. Scott1 , 1Div of Hematology, OBRR, 2Div of Viral Products, OVRR
Progressive vaccinia is a severe life threatening complication of that can occur following the administration of the Dryvax smallpox vaccine. Vaccinia Immune Globulin (VIG) is a high titer anti-vaccinia virus product produced from immunized donors that is the only currently licensed product for the treatment of this disease. In order to evaluate the efficacy of VIG we have developed a murine progressive vaccinia model using dermal scarification with Dryvax virus in severe combine immunodeficient mice (SCID). This model displays a number of clinical characteristics seen in human progressive vaccinia such as progressive expansion of the primary lesion, escar formation, followed by lethal systemic spread of the infection. In our hands this model has shown a highly reproducible survival curves allowing us to test different post-exposure prophylaxis VIG treatment modalities. Results: We have found that VIG treatments given at 48 hours post-scarification is capable producing healing of the primary vaccinial lesion in 50% of SCID mice challenged with 106 virus/mouse, 67% of mice challenged with 105 virus and 89% of mice challenged with 104 virus. All mice treated with VIG showed a significant prolongation of survival time compared to virus challenged mice receiving no treatment. Also, this VIG treatment regime gives significant numbers of SCID mice surviving without any apparent disease symptoms for as long as four months at the lower virus challenge levels of 105 and 104 virus per mouse. We believe that the data generated by these studies is directly relevant to VIGIV therapy in human progressive vaccinia.
Characterization of a recombinant Interleukin-13 (IL-13) Receptor alpha 2 extracellular domain in blocking biological activities of IL-13
M. Kioi, S. Seetharam, N. Talwar, R. K. Puri, CBER, FDA, Bethesda, MD
Background: Understanding the mechanism of action of targeted therapeutics is critical for optimal development of biological products. Our group is involved in the identification of tumor-associated proteins to elucidate the mechanism of action and develop novel therapeutic approaches. IL-13 is a key cytokine involved in allergic diseases, inflammation, and cancer. IL-13 receptor alpha 2 (IL-13Rα2) is known to be a high affinity receptor, which is overexpressed in cancer cells and many inflammatory conditions. To characterize the biological significance of IL-13Rα2, we generated a soluble form of IL-13Rα2 extracellular domain (ECDα2).
Methods: We developed several techniques for expression and purification of ECDα2 protein including Escherichia coli and mammalian system. After purification, biological activities of ECDα2 protein were determined.
Results: First few attempts for expression of soluble fusion ECDα2 and stabilization of this protein after refolding failed. We later succeeded in expressing ECDα2 protein with high yield in 293FT cells. ECDα2 was mainly secreted as a monomer and heavily glycosylated. The purified ECDα2 inhibited IL-13-induced TF-1 (erythroleukemia) cell proliferation and protein synthesis, IL-13 binding on IL-13R-expressing cells, and IL-13-induced STAT6 phosphorylation. However, IL-4-induced proliferation and STAT6 phosphorylation were not affected.
Conclusions: These results indicate that ECDα2 protein is difficult to express and purify due to eight cysteine residues within the molecule. In addition, IL-13Rα2 is not involved in IL-4 signal cascade, and thus, this protein may be useful in blocking IL-13-specific effects for a variety of conditions.
Metabolic health status by metabonomics
D. Jacobs, Z. Ramadan, A. Fuerholz, S. Kochhar, Nestlé Research Center, PO BOX 44, Vers-chez-les-Blanc, CH-1000 Lausanne-26, Switzerland
Metabolites are products of cellular regulatory processes, and their levels can be regarded as the ultimate response of biological systems to genetic or environmental changes. The impact of metabonomics in food industry is potentially widespread. The technology can be applied to the following areas: identification of biomarkers leading to proprietary actives, taste and flavor chemistry, clinical trial efficacy and safety, and development of robust scientific concepts for nutritional claims. Identification of validated biomarkers holds a tremendous potential to substantiate specific claims. The primary objective of the study was to define metabolite profiles of urine, plasma and saliva in general human population and search for phenotypic (gender, age etc.) and for lifestyle (sports, smoking etc.) specific patterns.
A total of 150 healthy adults (about 50 % females and 50 % males) were recruited at Nestlé Research Center, Lausanne to donate one-time sampling of fasting plasma, urine and saliva. The subjects completed their habitual lifestyle questionnaire and one dimensional 1H NMR spectra were recorded on each of the samples, and analyzed by multivariate statistical methods. Mathematical models were developed in an attempt to understand inter-individual metabolome variations. The data show how metabonomic study could lead to understanding of mechanism underpinning the effect of specific diet and other parameters on the individual health status. A major strength of the new -omic technology is that it is minimally invasive, and can thus easily be applied to study subtle variations in human metabolic health status and metabolism's dynamic relationship to environment that includes typical lifestyle pattern including diets etc. In conclusion, the metabonomic approach employing 1H NMR of biological fluids (plasma, urine, saliva, etc.) examines comprehensive biochemical profiles of low-molecular-weight metabolites in various biofluids that are modulated in response to various habitual lifestyle stimuli. The study provides solid base to define inter-individual metabolite profile variability, and to construct multivariate boundaries of normality.
Development of a multivariate statistical model for nephrotoxicity biomarker identification based on urinary 1H-NMR spectral pattern
M. Linder1 , I. Gustafsson1 , J. Lindberg1 , R. Torgrip1 , P. Ciaccio2 , S. Emeigh Hart2 , G. Betton3 , I. D. Wilson3 , E. Lenz3 , I. Schuppe-Koistinen1 , 1AstraZeneca R&D Södertälje, Sweden, 2AstraZeneca R&D Wilmington, DE, 3AstraZeneca R&D Alderley Park, UK
Methods
Male Han Wistar rats were treated with 3 compounds (mefenamic acid, N-phenylanthranylic acid and indomethacin) inducing renal papillary necrosis (RPN) and 3 compounds inducing renal cortical injury (gentamicin, cyclosporine and mercuric chloride). Approximately 1500 1H-NMR spectra were acquired on a 600 MHz NMR spectrometer. Spectral peaks were aligned using an in house alignment software and individual multivariate models were developed for each study. For biomarker identification all 6 studies were combined in one multivariate statistical model using Partial Least Squares- Discriminant Analysis.
Results
The model compounds for RPN induced different degrees of papillary toxicity. All compounds changed the endogenous metabolite pattern of rat urine as the treatment groups cluster separately from controls in each study. A combined multivariate model was developed including 1H-NMR spectra from animals treated with cortical toxins. Based on this model, potential biomarkers for RPN were identified including increased levels of glycolate, urocanate and phenylalanine as well as decreased levels of ornithine, betaine and choline.
Conclusion
Regiospecific renal damage, such as RPN, is difficult to monitor and predict. Using NMR-based metabolite profiling kidney toxicity-specific pattern were identified. A number of metabolites correlating with RPN have been tentatively assigned. These potential biomarkers for RPN require further validation.
| Sigma Xi Outstanding Poster Award - 2005 FDA Science Forum |
Effects of Implantable Cardioverter-Defibrillator (ICD) Shocks upon Human Blood
V. Krauthamer1 , R. Malinauskas1 , E. Mitrojorgji1 , M. S. Stratton1 , K. Higginson1 , J. Zivny2 , J. Vostal2 , 1CDRH, FDA, Rockville, MD, 2CBER, FDA, Bethesda, MD
The 100,000 patients annually undergoing ICD implantation with electrical defibrillation threshold testing have a relatively high risk of thrombo-embolic events (5%). It is unknown if the electric current of the ICD shocks causes platelet activation and hemolysis. Human blood from healthy individuals was shocked in vitro in electroporation vials using a defibrillator. Electroporation vials are ideal for studying blood damage because they accommodate small sample volumes, and have parallel electrode plates that provide a uniform current density. We applied energy levels from 0 - 35J, which generate up to twice the maximum equivalent current densities as used in commercial transvenous defibrillator leads. Blood temperature was measured to separate thermal from electrical effects. Platelet activation was evaluated by measuring expression of P-selectin on the surface of activated platelets using flow cytometry. Red blood cell (RBC) damage was assessed using blood smears, measuring hemoglobin released into plasma, and with a blood cell counter. Increased levels of hemolysis were noticeable at 8J (current density of 2.5 A/cm2), while platelet activation was increased only at 35J (6.7 A/cm2) in whole blood (but not in platelet-rich plasma). Alterations in RBC morphology, platelet count, mean RBC volume, and RBC distribution width occurred at the higher energy levels. Our preliminary findings support the results of a clinical study that threshold testing of ICDs is not associated with platelet activation. In vitro testing using electroporation vials may provide a new method to evaluate blood damage caused by electrical stimulation from medical devices.
A novel in vitro system for the evaluation of human drug adverse effects: The integrated discrete multiorgan co-culture (IdMOC) system
A.P. Li, Advanced Pharmaceutical Sciences Inc.
Primary cell culture systems, especially human cell system such as human hepatocytes, are now routinely used to evaluate drug effects, including efficacy, metabolism, and toxicity. Using a cell type derived from an organ (e.g., hepatocytes from a liver), one can obtain experimental system pertaining to the properties of the chosen organ. However, a drug administered to a whole organism will be subjected to multiple organ interactions. The drug may be metabolized by multiple organs, and the parent drug and its metabolites may be multiple organ effects. Further, metabolites from one organ (e.g. liver) may cause specific effects on cells of another organ (e.g. heart). Therefore, while the conventional single-organ in vitro system allows one to assess drug properties pertaining to a single organ (e.g. hepatic metabolism or hepatotoxicity), accurate prediction of in vivo effects requires an experimental system with multiple organ interactions as the whole organism in vivo. The IdMOC system consists of wells-in a well: multiple cell types can be cultured discretely in the inner wells, and when the cells are ready for integration, the wells are flooded with medium so that they are now connected by a common medium, akin to the blood connecting the organs in a human body. An application of IdMOC is demonstrated with tamoxifen, an anticancer agent for the treatment of estrogen-dependent breast cancer. Tamoxifen was evaluated for its cytotoxic effects on the following human cell types: MCF-7 breast adenocarcinoma cells, hepatocytes, renal proximal tubule epithelial cells, aortic endothelial cells, small airway epithelial cells, and astrocytes, representing the target cancer cells, and the normal organs liver, kidney, vascular system, lung, and central nervous system, respectively. Tamoxifen was found to be highly cytotoxic towards the MCF-7 cells, and expressed difffertial cytotoxicity towards the various normal cells. The results are similar to the known anticancer efficacy and normal organ toxicity of tamoxifen. The IdMOC represents an improved in vitro experimental system to assess human drug properties, including toxicity, metabolism, and drug distribution.
Testing biomaterials for stimulation of major inflammatory pathways
D. B. Lyle, J. C. Shallcross, V. M. Hitchins, J. J. Langone, CDRH, FDA, Silver Spring MD 20903
Background: Reactions to biomaterials present in implantable combination devices may include inflammation, such as mediated by macrophages, that may harm patient tissue or potentially interfere with proper function of the device. One of the substances produced by activated macrophages is nitric oxide (NO).
Methods: RAW 264.7 cells were grown in culture and treated at various times with drugs/reagents. Supernatants were assayed for nitric oxide via an increase in the fluorescence of the dye 2,3-diaminonapththalene (DAN) to nitrite.
Results: DAN allowed for detection of a very early (24 hours or less) initial response in nanomolar range of the inflammatory reaction of RAW 264.7 to E. coli 026:B6 lipopolysaccharide (LPS). By looking at early time points (20 and 24 hrs), an interferon-gamma augmentation of the LPS effect was readily discerned. The full interferon-only effect was studied by taking readings at 48 hrs or later. The cells produced a robust response of nitric oxide to uM zymosan comparable to that of nM LPS, which also was augmented by Interferon. Macrophage activation through the main inflammatory signaling pathway (nuclear factor-kappa beta) was confirmed in this system by parthenolide inhibition of zymosan and LPS stimulation. Interferon, which signals by a different pathway (the JAK pathway), was only partially inhibited by parthenolide.
Conclusions: A robust response of NO production by the murine macrophage cell line RAW 264.7 may serve as a convenient platform to confirm and identify cellular mechanisms and pathways to determine safety of implanted devices.
An In Vitro Test for Inflammatory Potential of Hyaluronic Acid Preparations
D. B. Lyle1 , J. C. Shallcross1 , V. M. Hitchins1 , C. N. Durfor2 , J. J. Langone1 , 1CDRH, FDA, Silver Spring, MD 20903, 2CDRH, FDA, Rockville, MD 20850
Background: Hyaluronic acid (HA) is a natural component of the body's extra-cellular matrix, contributing to the viscosity of skin. HA derived from different sources is used clinically as a tissue-filler to smooth skin wrinkles, and injected as a lubricant to treat the joint pain of osteoarthritis. Adverse events from injection of HA preparations have been reported, which may be inflammatory in nature and involve macrophages.
Methods: RAW 264.7, a well-characterized murine macrophage cell line, was treated at various times with HA preparations derived from bacterial, animal, and human sources (purchased from Sigma) and control reagents. Supernatants were assayed for nitric oxide (NO), an indicator of macrophage activation/inflammation, via an increase in the fluorescence of 2,3-diaminonapththalene (DAN) to nitrite. Endotoxin concentrations in HA were assayed using a Limulus amebocyte lysate (LAL) chromogenic kit (BioWhittaker).
Results: HA preparations directly stimulated NO production by RAW 264.7 cells. The NO stimulation was attributable to endotoxin contamination, as determined by the LAL assay. In 10 mg/ml HA preparations endotoxin ranged from 0.5 ng/ml to 15 µg/ml. "Cryptic" responses were revealed by simultaneous treatment with mouse interferon-gamma.
Conclusions: Inflammatory reactions attributed to HA preparations may be due not to the native molecule itself, but to possible endotoxin contamination. Ascertaining the potential of HA preparations to induce macrophage inflammation is an important preclinical and batch-certification test. These results were obtained with HA preparations not approved for clinical use. The validity of these results with HA products approved by the FDA for clinical use remains to be explored.
Age-related changes in intestinal crypt density may contribute to increased sensitivity of soy-fed female rats to acute toxicity of azoxymethane
A. Tracy1 , S. Francke-Carroll2 , B. A. Magnuson1 , 1University of Maryland, 2FDA
Soy isoflavone consumption by the public has significantly increased due to popularity of supplements and foods with added isoflavone for possible hormonal, cancer and heart disease-preventive properties. During one of our feeding studies for colon cancer prevention, we observed acute intestinal toxicity specifically in older female rats fed a soy-supplemented diet in response to treatment with the well-established colon carcinogen azoxymethane (AOM). Investigating possible mechanisms for this observation, we focused on two parameters: a) the expression of CYP2E1, an important enzyme responsible for metabolism of toxins and carcinogens, and b) on differences of the intestinal crypt density in older versus younger rats. We preformed real time PCR to quantify CYP2E1 in liver tissue. We assessed the intestinal crypt density stereometrically by applying a standardized square grid to standardized digital photographs of young, middle-aged and old rat colons that were sectioned in a standardized manner. The number of crypts crossing defined grid lines at defined points was calculated for each colon section from an individual rat and the average number of crypts/line was computed for each animal.There was no change in the gene expression of CYP2E1 due to either diet or age. Both age and diet affected colonic crypt density. Soy-fed rats had a statistically significant higher crypt density than the control rats, and young rats had a significantly higher crypt density than the mature and old rats (p<0.001). These findings suggest that age as well as a soy-supplemented diet decreases the colon crypt density, and thereby increasing susceptibility to acute toxicity following carcinogen injection.
Effect of age on azoxymethane-induced colonic apoptosis in rats
Y. J. Kwon1 , S. Francke-Carroll2 , B. A. Magnuson1 , 1University of Maryland, 2Pathology Branch
Aging processes may alter the regulation of apoptosis. This study investigated the effect of age on the time course of azoxymethane (AOM)-induced apoptosis in the colon. Young (6 week), middle-aged (12 month), and old (21 month) F344 male rats were treated with single s.c. injections of either saline or AOM (15mg/kg) and 0 (saline-treated), 4, 8, 16, or 24 hours later incidence of apoptosis (total stained cells/colon crypt) was estimated by TUNEL assay in the distal colon. In all age groups, apoptosis rarely occurred at the 0 hr time point and the apoptotic peak occurred at the 8 hr time point. After reaching the peak, apoptotic incidence decreased and there was no significant difference (p>0.05) between 0 hr and 24 hr time points in young and middle-aged rats compared to a significantly higher incidence in old animals. Old rats showed the highest and most rapid increase of apoptosis followed by middle-aged rats: old rats showed significantly higher apoptosis (4.1 ± 0.88 apoptotic cells/crypt) compared to young (1.13 ± 0.25) and middle-aged animals (1.73 ± 0.40) at the 8 hr time point. Apoptosis in old rats was also significantly higher (2.35 ± 0.63) than that of young rats (0.76 ± 0.18) at 16 hr. However, the magnitude of age-related difference in response to apoptosis decreased with time (significant age ´ time interaction): at the 24 hour time point, apoptosis was not significantly different among different ages of rats. The induction of apoptosis is greater and occurs for a longer time period in old animals. This age-related difference may be at least partially responsible for our previous finding of age-related difference in the inhibition of colon carcinogenesis by dietary interventions.
Quantitative morphometric analysis (QMA) of prion protein (PrPSc) for improved diagnosis of Transmissible Spongiform Encephalopathies (TSEs)
O. Maximova, R. Taffs, K. Pomeroy, D. McMahon, P. Piccardo, D. Asher, FDA
Background: TSEs are commonly diagnosed by histopathological examination and immunohistochemical (IHC) detection of disease-associated prion protein (PrPSc) in infected tissues. The traditional evaluation of IHC staining in tissues is based on visual inspection of sections and then grading by an arbitrarily chosen semi-quantitative scale. This approach is limited by observer subjectivity and bias, often yielding inconsistent and variable results. Our goal was to develop an objective method to facilitate reliable diagnosis of TSEs using image analysis of PrPSc immunostaining.
Methods: Paraffin-embedded sections from brains of hamsters infected with the 263K strain of scrapie were used to standardize detection of PrPSc by IHC. Commercial image analysis software (MetaMorph™) was then used to quantify PrPSc immunostaining.
Results: We first identified a "negative-chromaticity subdomain"-a range of hue, saturation, and intensity (HSI) that included all light from negative-control tissue sections. A chromaticity subdomain for positive PrPSc immunostaining was then defined by subtracting negative hue values from those of the entire visible light spectrum. This method allowed unambiguous detection of even minimal amounts of PrPSc in infected tissues. Performance of QMA was evaluated by comparing visual scores of distribution and intensity of PrPSc immunostaining with QMA data. QMA results correlated very well with traditional scores.
Conclusions: QMA of PrPSc provides a straightforward, reliable and objective measure of immunostaining that might aid in the diagnosis of difficult cases of TSE and offers a potentially powerful tool for many other applications requiring quantitative evaluation of immunohistochemical staining.
In Silico Discovery of Bridging Renal Biomarkers
D. Mendrick1 , M. Lawton2 , 1Gene Logic Inc., Gaithersburg, MD 20879, 2Pfizer Global Research & Development, Groton/New London Laboratories, Pfizer Inc, Groton, CT 06340
Background:The need for bridging biomarkers to predict nephrotoxicity and to follow its resolution in preclinical species and in humans continues to be of major interest to the FDA and pharmaceutical companies. Gene Logic has compiled large reference databases of gene expression data from normal and toxicant-treated rat tissues and from normal and diseased human tissues. This offers the possibility for in silico data mining to identify potential bridging biomarkers that might be useful for toxicity applications and/or disease diagnosis
Methods: Gene expression data generated on Affymetrix GeneChip® microarrays (RG_U34A, RAE 230 2.0, HG-U133) were evaluated. Examination was conducted on normal rat and human tissues to examine each gene's tissue distribution as this is important particularly if one chooses to monitor its encoded protein or peptides in urine or blood. To focus on the identification of predictive (leading) biomarkers, analysis was conducted on 1) toxicant- and vehicle-treated rat kidney tissue and 2) normal and diseased human renal tissue.
Results: Initial choices of genes focused on those reported in the public domain to be potential nephrotoxicity biomarkers including kidney injury molecule-1 (KIM-1,Havcr1) and cysteine rich protein 61 (CYR61). The analysis then was expanded to discover new potential biomarkers.
Conclusions: Large reference databases of normal and phenotypically altered tissue (caused by human disease conditions and experimental studies in preclinical species) afford the opportunity for in silico data mining to discover potential bridging biomarkers. Additional "wet lab work" will be needed to validate such markers.
Phenotyping Escherichia coli O157:H7
A. Mukherjee, K. L. McCutchan, E. W. Brown, J. E. LeClerc, T. A. Cebula, CFSAN, FDA, Laurel, MD
Background: As more genome sequences of microbial strains become available, it is becoming evident that variability among closely-related strains is prevalent. In order to investigate whether these variations are manifested in their phenotypes, a high throughput phenotypic microarray system capable of identifying more than 1200 phenotypes is being used to analyze a diverse collection of E. coli O157:H7 isolates.
Methods: Bacterial cells were inoculated into twenty 96-well plates containing various substrates: such as different carbon and nitrogen sources; varying conditions like pH and ionic environments; and various antibiotics. The plates were incubated at 37°C for 48 h. All wells contain a reducible tetrazolium dye, which, upon cell growth and respiration, forms color. The intensity of the color is recorded every 15 minute and analyzed.
Results: Eighty strains of E. coli O157:H7 have been analyzed by phenotypic microarray. Cladistic analysis using the phenotypic data grouped the strains into clades; phenotypes were also used to resolve clades of O157:H7 strains based on molecular markers. Additionally, we observed that O157:H7 strains can utilize sucrose but not D-serine as a carbon source, whereas most K-12 strains utilize D-serine but not sucrose. PCR analysis was used to confirm variability in the relevant chromosomal regions of these strains.
Conclusion: Cladistic analysis has enabled us to assign signature phenotypic changes that uniquely define clades or 'bins' of O157:H7. We are exploiting the ability of O157:H7 strains to utilize sucrose and not D-serine as a carbon source in an effort to distinguish O157:H7 from other pathogenic E. coli strains.
Candidate Cross-Species Biomarkers of Fibrosis and Hepatitis
M.S. Orr, Gene Logic Inc.
Background: There is a need in the toxicology field to develop diagnostic biomarkers that are relevant to multiple species such as rat, dog, and human. Genomics provides an opportunity to discover novel cross-species biomarkers for identifying phenotypes such as hepatitis and fibrosis, especially if the biomarkers are tissue specific secreted proteins that can be monitored in the blood.
Methods: Microarray analysis using Affyemtrix gene chips® was utilized to identify dysregulated secreted genes in human or rat samples. Comparisons of the dysregulated genes identified in either the human or rat studies were performed in order to identify candidate cross-species diagnostic markers of hepatitis and fibrosis.
Results: Microarray analysis identified 80 differentially expressed genes that coded for secreted proteins in human cirrhotic/fibrosis tissue samples and many of the genes were liver specific based on the gene expression profile across a panel of normal human tissues. Interestingly, APOA1 was one of the 80 secreted proteins identified by this genomic analysis and it is currently utilized as a biomarker for the early detection of fibrosis. In addition, Fetuin-B (FETUB), a secreted protein that can be monitored in serum, was identified as a novel candidate cross-species biomarker of fibrosis, as it is dysregulated in both the human hepatitis/fibrosis samples and rat DMN treated liver samples.
Conclusions: This study illustrates a genomic approach for identifying candidate cross-species biomarkers of cirrhosis/fibrosis for humans and rats. The study identified a previously known biomarker of fibrosis (APOA1) and a novel candidate biomarker of fibrosis, FETUB.
Elevated Circulating Endothelial Microparticles in Acute Stroke Patients: A Correlation with Brain Lesion Volume and Outcome
J. Simak1 , M. P. Gelderman1 , H. Yu2 , V. Wright2 , N. Alberts-Grill2 , J. T. Stranix2 , A. E. Baird2 , 1CBER, FDA, Rockville, MD, 2NINDS, NIH, Bethesda, MD
Purpose: Elevated endothelial cell membrane microparticles (EC MP) in blood have been demonstrated in various diseases with a vascular injury component. The aim of this study was to investigate if circulating EC MP show a relationship with outcome after acute stroke and with the ischemic brain lesion volume measured by magnetic resonance diffusion-weighted imaging (DWI).
Methods: EC MP were analyzed in the blood of 42 acute stroke patients (AS): 20 patients with National Institutes of Health Stroke Scale (NIHSS) scores < 5 were classified as mild stroke (MS) (median NIHSS= 2; 25th-75th%: 0-2), while the other 22 patients with NIHSS >/= 5 (NIHSS=12; 6-21) were classified as moderate to severe stroke (SS). Peripheral venous blood samples were collected at a median time of 36 hours after the onset of clinical symptoms. Blood samples of 23 age matched control volunteers (CTRL) were used for comparison. EC MP were identified by antibodies to EC antigen CD105 (endoglin) and the highly specific CD144 (VE-cadherin) using a three-color flow cytometry assay. Platelet, white, and red blood cell MP were identified using cell specific antibodies to CD41a, CD45, and CD235a, respectively. Lesion volume was measured on magnetic resonance diffusion-weighted imaging (DWI) and clinical outcome was based on the Rankin score at hospital discharge.
Results: Plasma counts of CD105+CD41a-CD45- EC MP were elevated in SS (median: 840/µL; 25th-75th%: 565-1079/µL) as compared to CTRL (415/µL; 201-624/µL; p=0.014). Moreover, CD105+CD144+ EC MP were elevated in SS (261/µL; 137-433/µL) when compared to MS (154/µL; 99-182/µL; p=0.031) or the CTRL group (140/µL; 79-247/µL; p=0.031). Interestingly, CD105+CD41a-CD45- EC MP, but not CD105+CD144+ EC MP, exhibited in AS group a correlation (p=0.005; r=0.45) with brain lesion volume on DWI. In addition, CD105+CD144+ EC MP in the admission samples correlated (p=0.0007; r=0.54) with the Rankin disability score in the AS group at hospital discharge as did CD105+CD41a-CD45- EC MP (p=0.007; r=0.44).
Conclusion: Specific phenotypes of EC MP in the plasma samples of stroke patients were associated with stroke severity, ischemic lesion volume and clinical outcome. Analysis of EC MP in peripheral blood of stroke patients could be of diagnostic and prognostic use.
Effects of the Anesthetic, Ketamine, in Developing Rat and Monkey Forebrain Cultures
W. Slikker, Jr.1 , C. Hotchkiss2 , N. Sadovova3 , R. Devine3 , X. Fu4 , A. Scallet1 , J. Hanig5 , C. Wang1 , 1NCTR/Neurotoxicology, 2Bionetics Corporation, 3Toxicologic Pathology Associates, 4NCTR/Biochemical Toxicology, 5CDER/FDA
Ketamine, a noncompetitive N-methyl-D-aspartate (NMDA) receptor antagonist, has been used as a general pediatric anesthetic for surgical procedures in infants. Recent data from developing rats suggest that anesthetic drugsmaycause dose-dependent neurodegeneration.
Purpose: To compare ketamine-induced developmental neurotoxicity using rhesus monkey (postnatal day (PND) 3) frontal cortical and rodent (PND 7) forebrain culture and also to determine if dysregulation of NMDA receptor subunits promotes ketamine-induced apoptosis.
Methods: Primary brain cell cultures were incubated for 24 hrs with 1, 10 or 20 µM ketamine alone or co-incubated with ketamine and NR1 antisense oligonucleotides (2 µM). After washout of ketamine, cultures were kept in serum and glutamate-containing medium for 24 hrs.
Results: Ketamine (10 and 20 µM) caused a marked reduction in immunostaining for PSA-NCAM, a substantial increase in DNA fragmentation as measured by cell death ELISA, increased TUNEL-positive cells, and a reduction in MTT metabolism. NR1 antisense protected the neurons from ketamine-induced degeneration. Western analysis showed that neurotoxic concentrations of ketamine resulted in a decrease in PSA-NCAM expression. NR1 antisense prevented this effect of ketamine, suggesting that ketamine-induced cell death is associated with a compensatory up-regulation of the NMDA receptor.
Conclusions: NR1 offers neuroprotection from enhanced degeneration in vitro, and forebrain cell death induced by ketamine negatively impacts cortical synaptogenesis. Based on these in vitro results from developing brain tissue, cultured monkey forebrain is just as sensitive to the apoptotic effects of ketamine as the cultured rodent forebrain. Supported by NCTR/CDER/FDA, NTP and NICHD.
A REEVALUATION OF ANNEXIN I EXPRESSION IN NORMAL BREAST AND BREAST CARCINOMA
R. Speer1 , J. D. Wulfkuhle1 , V. S. Calvert1 , M. Raffeld2 , T. A. Braunschweig3 , S. M. Hewitt3 , L. A. Liotta1 , E. F. Petricoin1 , 1NCI/FDA Clinical Proteomics Program, Bethesda, MD, 2Laboratory of Pathology, Center for Cancer Research, National Cancer Institute, NIH, Bethesda, MD, 3Advanced Technology Center, National Cancer Institute, NIH, Bethesda, MD
Purpose: To reevaluate the annexin I expression in normal breast and breast carcinoma.
Experimental Procedures: We employed laser capture microdissection in order to select pure cell subpopulations of interest from heterogeneous frozen tissue sections of a human tissue set comprised of normal breast tissue, DCIS and breast cancer. The annexin I expression was analyzed using reverse phase protein microarray and Western immunoblotting. Tissue microarray staining and immunohistochemistry were used to verify the results and to localize the annexin I expression more specifically.
Results: Proteomic analysis of human breast tissue indicates the loss of annexin I expression during tumorigenesis. Lysates of 16 normal breast tissue samples, 1 DCIS sample and 45 breast cancer samples (partly matched) were analyzed by reverse phase protein microarray, showing dramatically decreased annexin I levels in cancer compared to normal breast tissue. Western blotting of 5 normal and 5 breast cancer lysates confirmed these findings. Immunohistochemical staining of both frozen breast sections and paraffin embedded tissue microarray supports the findings of a downregulation of annexin I during tumorigenesis.
Conclusions: Biological insights into the role that the annexin protein family plays in normal and aberrant cellular function may implicate this molecule as key mediator of cell growth and cell cycle modulation. We were able to show the downregulation of annexin I during tumorigenesis of breast cancer. This finding may reveal an important early event during carcinogenesis, namely that the annexin I down-regulation may contribute to this loss of structure and enable the cancer cells to invade adjacent local tissue structures, culminating in distant metastases.
Risk of Local Adverse Events Following Cardiac Catheterization by Hemostasis Device and Gender
D. R. Tavris1 , S. Deh2 , B. Albrecht-Gallauresi1 , R. E. Shaw3 , R. G. Brindis4 , W. S. Weintraub5 , K. Mitchell2 , 1FDA, 2American College of Cardiology, 3San Francisco Heart Institute, 4Sam Framcoscp Kaiser Hospital, 5Emory Center for Outcomes Research
Background:
Reports to the FDA of local complications resulting in serious injuries and deaths associated with the use of hemostasis devices following cardiac catheterization provided the impetus for this study of risk factors from a national cardiac database.
Methods:
Data was obtained from the American College of Cardiology-National Cardiovascular Data RegistryTM, updated to include supplemental information specifically for this research. It included information from 59 institutions and 13,878 cardiac catheterizations (diagnostic and therapeutic) performed during October-December of 2003. Multiple logistic regression, using ten different outcomes, was used to assess the risk associated with type of device and gender, while controlling for demographic, physiologic, and procedure related variables, and several indices of co-morbidity.
Results:
Serious adverse events were reported in 3.54 % of patients, the most common being hematoma (2.00 %).
In the multivariate analysis, factors that were positively associated with 'any vascular complication' included female gender (OR=1.69, p<.0001), sheath size (OR=1.24, p=.005), activated clotting time (ACT) (OR=1.02, p<.0001), renal failure (OR=1.85, p=.006), and emergency indication for the procedure (OR=1.59, p=.001). Compared with manual compression controls, there was one hemostasis device that demonstrated an association with 'any vascular complication' (OR=2.42[1.41-4.16], p=.001).
Conclusions:
Women have almost twice the risk of men for most local complications following cardiac catheterization. Other risk factors included sheath size, ACT, renal failure, emergency indication for the procedure, and one type of hemostasis device. Possible reasons for these associations (including unmeasured confounding variables) will be discussed.
Neuroimaging Studies in CNS Drugs Approved from 1995 to 2004
R. S. Uppoor1 , S. U. Yasuda1 , P. Mummaneni1 , E. Cooper2 , H. Pien2 , G. Sorenson2 , J. M. Collins1 , M. U. Mehta1 , 1CDER, FDA, Rockville, MD, 2Harvard Medical School, Boston
BACKGROUND/AIMS: The objective of this project is to evaluate the use of the imaging modalities in development of Neuropharmacological drugs. Neuroimaging (NI) has the potential to link pharmacokinetics (PK) with pharmacodynamics (PD) and could therefore be an important aid in understanding mechanisms of action and adverse effects of drugs, as well as in identifying mechanisms for variability in drug response. Imaging can be an effective tool in optimizing Neuropharmacological drug development.
METHODS: NDAs approved from 1995 to 2004 in the Division of Neuropharmacological Drug Products were evaluated using a survey type approach. Information collected included the stated purpose of imaging study, PK, imaging, and PD results, and whether the study was used to support dosing or the proposed indication. Evaluation included review of the NDA submission, the final label, the FDA reviews and published literature.
RESULTS: There were 106 NDAs approved in the Neuropharmacology division. Fifteen of these NDAs included NI studies. Five of these NDAs had receptor occupancy studies and 2 were used to support the pharmacology/proof of concept. The receptor occupancy results are in reasonable agreement with the final selected doses/dose regimen.
CONCLUSIONS: A database capturing characteristics of NI studies in drug development has been initiated. Future efforts will focus on identifying those NI characteristics in the literature and in IND/NDAs that can be used to optimize drug development.
Mechanistic studies of the photocytotoxicity elicited by tattoo inks and component pigments
W.G. Wamer, CFSAN, FDA, College Park, MD
Background: A number of anecdotal reports suggest that tattooed skin can exhibit increased sensitivity to the sun. There is a need for systematic examination of tattoo inks and pigments for photoactivity.
Methods: Using an in vitro assay, 65 tattoo inks and 9 pigments were screened for photocytotoxicity. Three pigments and 3 tattoo inks were subsequently selected for additional mechanistic studies. To determine the wavelength dependence for photocytotoxicity, human skin fibroblasts, pre-treated overnight with tattoo inks or pigments, were irradiated with UVA light (320-400 nm) combined with visible light (400-800 nm) or with visible light alone. Photocytotoxicity was assessed as inhibition of colony growth. To gain insight into the mechanism of photocytotoxicity, we examined the connection between photocytotoxicity and oxidative damage, measured as lipid peroxidation.
Results: A significant number of tattoo inks and 3 pigments (Pigment Red 122, Pigment Red 170 and titanium dioxide) were photocytotoxic. The photocytotoxicity of titanium dioxide was fully dependent on the inclusion of UVA light during irradiation. The photocytotoxicity of Pigment Red 122 and Pigment Red 170 was partially reduced by filtering out UVA light. The photocytotoxicity of inks containing these three pigments showed a similar dependence on irradiation with UVA light. Titanium dioxide and Pigment Red 170 elicited lipid peroxidation concomitant with phototcytotoxicity.
Conclusions: These results suggest that sunlight may play a role in adverse reactions to tattoos. The roles of UVA light and oxidative damage need to be further investigated.
Application of A Model-Based Analysis to Improve the Drug Development Plan for A Life-Threatening Disease
Y. Wang, R. Powell, N. Beasley, P. Marroum, J. V.S. Gobburu, CDER, FDA, Rockville, MD
Background: Drug X was designed to delay the time to a disease event. A retrospective subgroup analysis of a failed pivotal trial (A) identified a responder population for the intent-to-treat patients in a second pivotal trial (B), which also failed. The purpose of the analysis was to explore the potential reasons for the failure of the two trials and provide guidance for future trials.
Methods: A time-dependent biomarker-disease event relationship was explored in both trials by the Cox proportional hazard model. The effect of drug X on biomarker level was analyzed in both trials by ANCOVA with baseline biomarker level as the covariate. Based on the Cox regression model, a simulation was performed to investigate the required biomarker level reduction to show significant clinical benefit.
Results: Biomarker level reduction was found to be correlated with a reduced risk of having the disease event. Drug X was shown to be able to lower the biomarker level significantly compared to placebo. The time-to-event analysis in both trials, however, failed to show the clinical benefit of drug X. The simulation results indicated that biomarker level reduction was not sufficient to show the expected clinical benefit given the sample size studied.
Conclusion: The analysis showed greater biomarker level reduction was needed to show the clinical benefit of drug X. To maximize the probability of success, higher doses should be studied in future trials given the observed dose-response relationship for the biomarker in dose-ranging studies. The sponsor accepted the recommendation and is studying higher doses.
Expression of hypoxia inducible factor (HIF-1a), erythropoietin (EPO) and heme oxygenase in guinea pig (Cavia porcellus) after exchange transfusion with hemoglobin-based blood substitutes.
L. Yeh, P. Buehler, N. Tayebi, A. I. Alayash, CBER, FDA, Bethesda, MD
Purpose: Expression of many mammalian genes is regulated by oxygen (O2) tension. This study was designed to test the concept that administration of cell free Hb-based oxygen carrier (OxyglobinTM) and its various oxidative and oxygenation states can be correlated to the expression of HIF-1α, a global transcriptional factor and other hypoxia sensitive genes such as erythropoietin (EPO), and heme oxygenase (HO-1) in a model of exchange transfusion.
Methods: We performed 50% and 75% blood substitute- OxyglobinTM exchange transfusion in guinea pig (N = 20). Animal without treatment served as baseline control. We excercised the kidney from these guinea pigs at specific time period (up to 72 hr) and organs were frozen in liquid nitrogen. Total RNA was extracted using Trizol. The cDNA for HIF-1α,EPO and HO-1 were generated by reverse transcription with primer pairs designed and synthesized based on the sequences of human in the GenBank. By semiquantitative RT-PCR, mRNA of HIF-1αwere detected using nested PCR technique.The genes were amplified using PCR and the products were separated using 1.5% agarose gel. Expression of G3DPH mRNA by a similar RT-PCR procedure was used as a positive control. The oxygenation and redox states of the Hb in sera obtained from animal at different time points were monitored by spectrophotometry.
Results: Ferrous Hb monitored during transfusion decline as a function of time and RBCs levels in animals. This was coupled with a proportional elevation of the ferric non-functional ferric form of Hb, indicating loss of the oxygen carrying capacity of Hb. Since only partial cDNA of HIF-1α in guinea pig was available in GenBank, the RT-PCR product of HIF-1α was identified and confirmed by restriction enzyme digestion as well as sequencing. Furthermore, the sequenced gene product of HIF-1α from the partial cDNA, was compared with known human HIF-1α sequence with 94% homology. The expression of HIF-1αwas increased three-folds over the control (Sham) after 4 hours of exchange transfusion and up to 72 hr. The levels of both EPO as well as HO-1 mRNAs were also elevated in the kidney tissues of these animals.
Conclusions: We demonstrate for the first time that a redox active "oxygen carrier" crosstalks with an "oxygen sensor" in a model of exchange transfusion. We also demonstrate that HIF-1α activates the transcription of genes whose proteins mediate adaptive responses to hypoxia, including EPO and HO-1.
Estimating the Uncertainty in the Estimated Mean Area Under the ROC Curve of a Multi-Feature Classifier
W. A.. Yousef1 , R. F. Wagner2 , M. H.. Loew3 , 1CDRH, FDA, Medical Device Fellows Program, Rockville, MD and George Washington University, Washington, D.C., 2CDRH, FDA, Rockville, MD, 3George Washington University, Washington, D.C.
This poster considers the problem of binary classification using any number of biomarkers and its assessment in a distribution-free approach. We estimate the area under the ROC curve of a classifier using estimators based on bootstrap methods of Efron and Tibshirani. We then use the method of the influence function to estimate the uncertainty of a particular estimate. Monte Carlo trials show that small-sample estimates can be obtained with little bias.
RhoGDI Protects Cancer Cells against Drug-induced Apoptosis
B. Zhang, Y. Zhang, E. Shacter, FDA, Bethesda, MD
Rho GDP Dissociation Inhibitor (RhoGDI) plays an essential role in control of a variety of cellular functions through interaction with Rho family GTPases including Rac1, Cdc42, and RhoA. RhoGDI is frequently overexpressed in different types of human tumors and chemoresistant cancer cell lines, raising the possibility that RhoGDI might play a role in the development of drug-resistance of cancer cells. Here, we found that overexpression of RhoGDI in MDA-MB-231 human breast cancer cells and JLP-119 lymphoma cells strongly inhibited apoptosis induced by different chemotherapeutic agents. Using DNA vector-mediated RNA interference, we were able to establish a stable MDA-MB-231 breast cancer cell line expressing siRNA targeting RhoGDI, rendering specific and persistent suppression of RhoGDI protein expression. These RhoGDI knockdown cells are significantly more susceptible to drug-induced apoptosis. Furthermore, we confirm that the observed knockdown phenotype is the result of silencing of the intended target by using a rescue plasmid to restore RhoGDI expression and drug-resistance. Taken together, these data suggest that RhoGDI may function as an antiapoptotic molecule that mediates cellular resistance to chemotherapy. In addition, we provide evidence that RhoGDI is a potent inhibitor of caspase-mediated cleavage of Rac1 GTPase. These results add a novel role for RhoGDI, showing that it shields Rac GTPase from proteolytic cleavage. Apoptotic cleavage of Rac1 has been shown to be required for maximal apoptosis to occur in response to cytotoxic drugs. Thus, RhoGDI-mediated inhibition of apoptosis may be partly through protection of Rac1 GTPase from caspase cleavage.
EXAMINATION OF THE VASCULAR INJURY IN SPRAGUE-DAWLEY (SD) RATS INDUCED BY THE PHOSPHODIESTERASE (PDE) IV INHIBITOR SCH 534385: A COMPARISION WITH SCH 351591.
J. Zhang1 , E. H. Herman1 , A. Knapton1 , T. J. Miller1 , P. Espandiari1 , R. Snyder2 , J. Hanig1 , J. L. Weaver1 , 1CDER, FDA, Silver Spring, MD, 2Schering-Plough Research Institute
Our previous study showed that the PDE IV inhibitor SCH 351591 induced vascular injury in SD rats and suggested that a panel of serum proteins could potentially serve as biomarkers for detecting and monitoring vascular lesions (Toxicologist 78: 375, 2004). The present study was begun to determine whether the structurally distinct PDE IV inhibitor SCH 534385 would induce similar vascular injury and, if so, whether these same biomarkers would again be observed . SD rats were given SCH 534385 (20 or 40 mg/kg/day for 3 days) by gavage. Microscopic evaluation of tissues obtained 24 h after the final dose revealed dose-dependent vascular injury. Arterial hemorrhage and necrosis, periarterial inflammation, and microvascular injury (fibrin insudation and fibrin exudation) were found in the mesentery, pancreas, kidney, liver and small intestine. Activation of immune system cells (endothelial cells, mast cells, and macrophages) and proliferation of fibroblasts were also noted at sites of inflammation. Lymphocyte numbers were markedly decreased in splenic and thymic cortex. In peripheral blood, granulocytes were elevated and lymphocytes decreased. SCH 534385 induced increases in the serum proteins alpha-1-acid glycoprotein, haptoglobin, GRO/CINC-1 (homolog of human interleukin-8), vascular endothelial growth factor, and tissue inhibitor of metalloprotease-1. No significant changes were found in C-reactive protein. The pathological changes in the splanchnic vasculature, elevations in serum proteins, and alterations in leukocyte numbers in rats treated with this compound were similar to those seen in rats given SCH 351591. The data show that two structurally different PDE IV inhibitors, SCH 351591 and SCH 534385, induced similar vascular injury and comparable changes in biomarkers. Studies are underway to further characterize the etiology of vascular damage induced by PDE IV inhibitors and the clinical utility of serum biomarkers.
Bioinformatics Analysis of Possible Biomarkerfor the Evaluation of Biologics Product Quality
B. G. Zaslavsky1 , J. Han2 , J. Tiwari3 , R. K. Puri1 , 1CBER, FDA, Rockville, MD, 2CBER, FDA, Bethesda, MD, 3BER, FDA, Rockville, MD
To develop biomarkers of cellular quality, we utilized DNA microarray technology to determine gene expression during various growth conditions of 293 human embryonic kidney (HEK) cell line. These cells were cultured under different confluence states (40%, 90%, and over confluence), RNA isolated and labeled targets were hybridized with high quality ~10K cDNA microarrays. The image was captured by Axon scanner and data analyzed by GenePix software. The data were further analyzed by various tools available at NIH Center for Information Technology microarray database.
The logistic regression method of SAS software was applied to verify the genes identified by gene expression profiles. Two-group classification (over confluence vs. 90% confluence) was used to select the subset of differentially expressed genes. Because the number of studied genes was too large relative to the number of cases available, the leave-one-out cross-validation procedure was used to estimate the subset of most informative genes. For each leave-one-out training set a different logistic classifier was built; therefore the estimated error rate applies to the procedure used to build the classifier. The normalization procedures were performed in the preparation of highly processed raw data for statistical analysis. In order to evaluate the robustness or stability of the logistic classifier we created two different sets of data applying different normalization procedures to the set of raw data. We used one of the normalized datasets as training set for the logistic predictor and another dataset as the test sets. Application of the logistic predictor to the test datasets confirmed the robustness of the predictor to the normalization procedures. This predictor accurately predicted the class membership on the basis of the expression level of key genes. These analyses identified gene clusters that may serve as biomarker for assessment of the quality of 293 cell substrates in different confluence states.
An Analysis of Genetic Toxicity, Reproductive and Developmental Toxicity, and Carcinogenicity Data: I. Identification of Carcinogens Using Surrogate Endpoints
E. J. Matthews, N. L. Kruhlak, R. D. Benz, J. F. Contrera, US Food and Drug Administration, Center for Drug Evaluation and Research (HFD-901), 5600 Fishers Lane, Rockville, MD 20857
A retrospective analysis of standard genetic toxicity (genetox) tests, reproductive and developmental toxicity (reprotox) studies, and rodent carcinogenicity bioassays (rcbioassay) was performed by the FDA Center for Drug Evaluation and Research (CDER) Informatics and Computational Safety and Analysis Staff (ICSAS) to identify the genetox and reprotox endpoints whose results best correlate with rcbioassay observations. Criteria for a good surrogate for predicting carcinogenicity was high (>73%) positive predictive value, specificity, and correlation indicator (the average of the former two parameters), and low (<25%) false positives. The goal was to expand upon the evidence of a correlation between Salmonella t. mutagenicity and carcinogenicity test results established by the National Toxicology Program. ICSAS compiled genetox, reprotox, and rcbioassay databases on 5938, 2115, and 1488 chemicals, respectively; 1112 of the chemicals have both genetox and rcbioassay data and 721 chemicals have both reprotox and rcbioassay data. This study revealed that 11 genetox endpoint results provided a good correlation with rcbioassay findings: 8 gene mutation tests, 2 in vivo clastogenicity tests, and the unscheduled DNA synthesis assay. In addition, the results of 3 endocrine organ reprotox studies correlated with detection of carcinogens. In contrast, the results of many other genetox and reprotox tests were not successful by our criteria in predicting carcinogenicity.
An Analysis of Genetic Toxicity, Reproductive and Developmental Toxicity, and Carcinogenicity Data: II. Identification of Genotoxicants Using In Silico Methods
E. J. Matthews, N. L. Kruhlak, R. D. Benz, J. F. Contrera, US Food and Drug Administration, Center for Drug Evaluation and Research (HFD-901), 5600 Fishers Lane, Rockville, MD 20857
This investigation is part of program to develop in silico methods to provide decision support information for US Food and Drug Administration (FDA) regulatory and research activities. A retrospective analysis of standard genetic toxicity (genetox) tests, and reproductive and developmental toxicity (reprotox) studies was performed by the FDA Center for Drug Evaluation and Research (CDER) Informatics and Computational Safety and Analysis Staff (ICSAS) to create genetox and reprotox endpoint modules using MC4PC computational toxicology software program (MultiCASE, Inc.) that can provide good in silico quantitative structure activity relationship (QSAR) predictions for those endpoints. The modules created correspond to the 14 genetox and endocrine organ reprotox endpoints that produced results that correlated with those of carcinogenicity studies. Of these, the plant mutation data were not suitable for QSAR modeling. The Salmonella mutation module had a sensitivity of 70.1%, a specificity of 89.9%, a positive predictivity of 85.1%, a correlation indicator (the average of specificity and positive predictivity) of 87.5%, and a false positive rate of 10.1% for this same endpoint. The other 12 QSAR modules also exhibited a high correlation indicator (>75%) and low false positive rate (<15%) for their respective laboratory findings. In addition, for 1112 chemicals having both rcbioassay data and also genetox data at at least some of the 10 endpoints, experimental results, and, where these were lacking, in silico predictions for genetox and reprotox data were pooled to determine cumulative predictions of carcinogenicity.
Beyond the LFER Paradigm: Harnessing Atomic Descriptors and Artificial Neural Networks to Predict pKa
R. Fraczkiewicz, G. Fraczkiewicz, B. Steere, M. B. Bolger, Simulations Plus
A vast array of chemical and biological properties of molecules strongly depend on ionization in water - the fundamental solvent in living systems. Consequently, the knowledge and understanding of ionization constants (pKa) is of chief importance to the pharmaceutical and environmental industries. The cheapest and easiest way of obtaining pKa for new molecules is in silico prediction. Almost all computational methods for this purpose are based on a perturbational approach utilizing the Linear Free Energy Relationships (LFER) of Hammett and Taft. An alternative route to pKa prediction, successful in the area of QSAR/QSPR, involves a direct, non-linear correlation between the observed property, pKa, and calculated descriptors. However, unlike straigtforward molecular property-molecular descriptors relationships (e.g., log P, solubility), the localized nature of ionization requires creation of a new class of atomic descriptors. Another serious problem: unlike measured log P or solubility which, essentially, describe single reactions, the measured pKas of a polyprotic molecule are a net effect of (sometimes) very large and complex networks of microscopic equilibria. We have solved both problems and will present the resulting model of pKa prediction.
Development of a Structure-Searchable Subchronic Studies Database for Use in Predictive Toxicology Modeling
C. Nelson, J. Mayer, K. Arvidson, A. McDougal, E. Lee, M. Twaroski, CFSAN, FDA, College Park, MD
Background: Toxicology data are being compiled to create comprehensive, structure-searchable databases with the goal of improving regulatory safety reviews and developing better toxicology guidance. FDA/OFAS is exploring the use of predictive toxicology as a tool in making safety decisions, based on chemical structure, when little or no toxicity data are available.
Methods: Subchronic study information and FDA reviewer conclusions were extracted from toxicology memoranda and original study reports stored in microfiche, paper copies, and electronic files.Electronic dictionaries in the form of pull-down menus were used to standardize language when possible. An MS Access platform based on FDA/CFSAN's Redbook 2000 was developed in order to facilitate data entry and to allow for export of the data into other software programs. The subchronic database was imported into software that allows more comprehensive data visualization and structure searching capabilities.
Results: Subchronic studies submitted to FDA/CFSAN in support of submissions were data mined from multiple formats and consolidated into a structurally searchable platform. This platform allows for efficient data management and identification of similar compounds to compare safety data, such as target organ toxicity.
Conclusions: The FDA is organizing its toxicity data from multiple formats into structure-searchable databases. For new chemicals with limited toxicological data, searching for chemicals with similar structures and substructures will aid in identifying appropriate analogs for structure-activity relationship (SAR) analysis. The compilation of this subchronic database will also be useful in building models that can more efficiently predict toxicity in chronic studies, as well as aid in the design of those studies.
Does the lack of genetic diversity in animal models currently used for safety testing put the public at risk?
S. H. Nye1 , N. V. Cozzi1 , J. Baye1 , D. Evans1 , Y. Evrard1 , S. Korb1 , H. Vernon1 , A. Wittenburg1 , M. Hessner2 , X. Wang2 , H. J. Jacob2 , R. J. Roman2 , 1PhysioGenix, Inc., Milwaukee, WI, 2Medical College of Wisconsin, Milwaukee, WI
Purpose: Safety testing in Pharma is currently done by using inbred or outbred rat strains. Inbred strains offer reproducibility but have limited predictive value to other rat strains let alone humans. Outbred strains lack the perceived degree of genetic diversity and test populations cannot be reproduced in reasonable sample sizes. To address this, PhysioGenix developed genetically diverse PharmGenix rat panels that capture 82% of the commercially available genetic diversity in the rat and are faithfully reproduced in a controlled fashion.
Methods: PharmGenix rats were tested for sensitivity to the toxic effects of the antibiotic gentamicin, the analgesic acetaminophen and the Alzheimer's drug tacrine. Biomarkers in urine and blood were analyzed, organ pathology was assessed and gene expression by microarray analysis was performed.
Results: The differential response exhibited by PharmGenix rat strains revealed genetic components underlie toxicity to gentamicin. PharmGenix rats also responded differentially to acetaminophen, but the pattern of toxicity was dependent on the target organs (liver, kidney). Tacrine elevated serum transaminases in three PharmGenix strains, but hepatotoxicity was missed by the Sprague-Dawley and F344. For mechanistic studies, knowing that PharmGenix rats respond differently to drugs enables 20-fold enrichment in finding relevant susceptibility and resistance genes by microarray and this number is further refined when combining haplotype information.
Conclusions: Genetic background has a large impact on the response to known renal and liver toxicants and suggests that results of efficacy or safety testing performed in any single strain of rats will not predict the range of responses expected in humans.
Comparison of Felbamate and Fluorofelbamate Intermediary Metabolism by Human Liver S9 Fractions Leading to Reactive Metabolites
R. J. Parker1 , N. R. Hartman1 , B. A. Roecklein2 , H. Mortko2 , H. J. Kupferberg3 , J. Stables4 , J. M. Strong1 , 1FDA, 2MedPointe Pharmaceuticals, NJ, 3Consultant, Potomac MD, 4NIH / NINDS
Background: Previous studies have demonstrated that metabolism of felbamate (FBM) in humans and animals generates a reactive intermediate ATPAL which may be responsible for the toxicities observed with this drug. Formation of ATPAL from its immediate precursor requires loss of a 2-position hydrogen and it has been postulated that substitution of this atom with fluorine would prevent the formation of ATPAL. Based on this hypothesis fluorofelbamate (F-FBM) was synthesized and is presently undergoing drug development.
Methods: We compared the metabolism of selected metabolic precursors of FBM and F-FBM in human liver S9 fractions using glutathione (GSH) as a trapping agent for any reactive metabolites formed.
Results: When incubated with human liver S9 fraction, the FBM precursor MCF was oxidized to CCMF and further oxidized to CPPA. In contrast, the F-FBM precursor F-MCF was stable and no metabolites were observed. When CCMF was incubated with human liver S9, both oxidation to CPPA and reduction to MCF were observed and a new atropic acid GSH adduct was identified. When F-CCMF was incubated under the same conditions, both reduced and oxidized metabolites, F-MCF and F-CPPA, were identified.
Conclusions: Our results support the hypothesis that F-FBM and F-CCMF are not metabolized by human S9 fraction in vitro to the known FBM reactive metabolite ATPAL.
The teratogen, valproic acid, alters homeobox gene expression during retinoic acid-induced nerve cell differentiation
D. Reese1 , M. Ramos-Valle2 , J. Breger3 , 1CFSAN, OARSA, DMB, Laurel, MD 20708, 2CFSAN, OARSA, DTNPS, Laurel, MD 20708, 3JIFSAN Student Intern, University of Maryland
Background: A short-term gene expression assay is being developed for detecting potential teratogenic agents in food and dietary supplements. This assay monitors an agent's ability to alter the expression of a key class of developmental control genes, homeobox (Hbox) genes, during retinoic acid (RA)-induced nerve cell (NC) differentiation in P19 mouse stem cells. Known teratogens are being evaluated in this system to assess its effectiveness in "flagging" potential teratogens. Valproic acid (VPA), used to treat seizures and bipolar disorder in humans and a potent teratogen, causing neural tube and limb defects, lower IQs, and developmental delays, was evaluated.
Methods: Hbox gene expression was monitored in P19-cell embryoid bodies exposed 4 days to RA or RA+VPA. Hbox-containing genes were selectively amplified from total RNA by RT-PCR and detected on an oligonucleotide array.
Results: VPA selectively upregulated a subset of homeobox genes (Hoxa9, Hoxc8, Hoxc9, Hoxd8, Hoxd9) that are not expressed during NC differentiation in this system. Upregulation of these genes, which are expressed in the lumbar region of the developing embryo, the site of spina bifida, was dose dependent and occurred at concentrations generated in the blood by therapeutic doses of the drug. Although NC differentiation was observed in valproate-treated cultures, these cultures had a significant population of proliferating/undifferentiated cells not seen in control cultures.
Conclusions: Agents that selectively alter the normal expression of key developmental control genes are potential teratogens and candidates for further evaluation in animal bioassays. Valproate would have been "flagged" by this assay for further evaluation.
Metabonomics: Bringing Safety Assessment into Early Drug Discovery
D. G. Robertson1 , L. Egnash2 , D. Wells1 , L. Robosky2 , M. Manning2 , M. D. Reily2 , J. M. Stanislawski1 , K. Datta1 , 1Department of Worldwide Safety Sciences, Pfizer Global R&D, Ann Arbor, 2Department of Discovery Technologies, Pfizer Global R&D, Ann Arbor
PURPOSE: Metabonomics has significant potential for bringing safety assessment early into the drug discovery process. Development of a candidate compound was hampered by sporadic and unexplained mortality.
METHODS: A rat in vivo toxicity study with a metabonomic component was conducted to evaluate the phenomenon. Twenty-four rats were divided into 4 dose groups (3/sex/dose) and administered 0, 100, 300 or 1000 mg/kg Compound X, once daily for 7 days. Urine was collected for metabonomic profiling. Routine hematology, clinical chemistry, gross pathology and limited histopathology were evaluated at termination.
RESULTS: All 300 and 1000 mg/kg animals died or were sacrificed in moribund condition by Day 3. A single 100 mg/kg female died on Day 6. Traditional toxicity endpoints, (toxicokinetics, clinical and microscopic pathology) were uninformative as to cause of death. Metabonomics evaluation identified a profound glucosuria accompanied by increased urinary ketone bodies and creatine that was completely reversed by Day 7, despite continued dosing. A second study was conducted at doses of 0, 30, 100 and 300 mg/kg (6 males/group) with more comprehensive serum and urine collection. The second study confirmed the findings of the first study, established the time-course of effects and identified transient decreases in serum triglycerides and elevations in serum glucose, insulin and corticosterone as concurrent findings.
CONCLUSIONS: Metabonomics evaluation identified heretofore-unknown in vivo biochemical sequelae of Compound X treatment that may have implications in understanding the pharmacology of the compound, as well as providing a simple biomarker (urine or serum glucose) and an approach for screening molecules in this class.
The Prediction of Maximum Recommended Therapeutic Dose (MRTD) with a Neural Network Ensemble Model
B. Steere1 , R. Fraczkiewicz1 , E. J. Matthews2 , R. D. Benz2 , N. L. Kruhlak2 , J. F. Contrera2 , 1Simulations Plus, Inc., Lancaster, CA 93534, 2ICSAS, OPS, CDER, FDA, Rockville, MD 20852
Background: MRTD is correlated to the toxic dose threshold in humans. The prediction of toxicity has been identified as a priority of the Critical Path Initiatives.
Methods: The ICSAS group reviewed MRTD data for orally-delivered drugs from clinical trials reports. The structures and MRTD values for 1,220 non-proprietary drugs were collated in a publicly-available database. 1,039 of the structures in this database were used by the Simulations-Plus group to generate molecular descriptors. The data was divided into training and test sets, then selected descriptors were combined with the MRTD values to build an ensemble of 32 neural network models.
Results: The output of the model was binned into categories to characterize the compounds in a 115-member sequestered test set as "Active", "Inactive", or "Inconclusive". The optimal neural net architecture had 5 hidden neurons and employed 40 descriptors. The ensemble correctly identified the inactive compounds 76% of the time, and misidentified them as active 14% of the time. Conversely, the model correctly identified the active compounds 67% of the time and misidentified them as inactive 10% of the time. Performance