Metabolomics is an FDA-identified Critical Path Opportunity (1) offering a toolkit which can be potentially applied to identification of safety biomarkers, diagnostic monitoring of patient response to drug treatment, lead optimization through toxicity assessment in non-clinical drug development, biochemical pathway studies in cells, animals and humans, patient stratification, and insights on and tracking of mechanisms associated with the onset of disease or following therapeutic intervention.
However, despite its above promise, the challenges in the complexity of the biological systems studies, the experimental spectroscopy methods used and the datasets generated has restricted to this point the full commercial application of metabolomics methods in the pharmaceutical industry and in healthcare.
The complexity of the interpretation of metabolomics data points to the need for improved data analytics and visualisation methods in decision making processes and situations (2). The importance of the value-added of the integration of metabolomics data with other proteomic and genomic data to provide a more integrated, accurate and broader view of the state of the biological system studied points to the importance of explicit knowledge management techniques in critical path areas of clinical research and drug development (3) to the application of semantic web, ontology and web service approaches (4) and to the adoption of these approaches in the day-to-day scientific research activity as supported by electronic laboratory notebooks and collaboration systems (5). It also points again to the importance of co-operation on the always difficult agreement area of the definition of standards and data integration.
I also find it quite interesting that two different fields (knowledge management and metabolomics) share a common critical concept: that of context. Context is critical in the human area of knowledge management and transfer in social ecosystems and its poor treatment a reason why many early IT and informatics approaches to knowledge management worked poorly. In the biological situation of a cell or animal, metabolomics provides the critical biochemical context and data to match it, and can do so in a dynamic way over time and can track perturbations in the behaviour of such a complex system.
Recent progress in the metabolomics field includes new advances in spectroscopic and statistical and analytical techniques that strengthen and expand the accuracy and scope of the analysis possible. But the progress also increasingly includes significant application experience which already has included a significant role in the 2005 Nobel Prize in Medicine award for assigning the causative role of H. pylori bacterium in peptic ulcers and gastritis, and more recently has been applied to patient stratification in Lou Gehrig’s disease, identification of off-targeted side effects for several drugs and new chemical entities, has been applied in diagnostic roles on serum samples of diabetics and non-diabetics, the development of biomarkers for prediction of drug-induced liver injury and to insight into toxic effects from urine analysis.
Richard Beger (FDA) has pointed out (6,7) that the development of NMR- based multi-dimensional quantitative spectrometric data-activity relationships (QSDAR) provides models which could be useful for estimating chemical toxicity, risk assessment of environmental contaminants and drug-lead identifications, and that such models of biological activity should be more objective and overcome some of the unreliabilities of traditional Quantitative Structure-Activity Relationship (QSAR) approaches (8). He also indicates that Metabolomics can play an important role in the creation of better evaluation tools and models for diseases, better identification and quantification of safety biomarkers, and improving the measurement of patient response. The voluntary submission of genomics data (VGDS) to the FDA is now accepting both proteomics and metabolomics data sets (9).
In response to such demand and interest, and in addition to significant activity in academic research, a number of companies including Metabolon, Leco, Blue Gnome, Bio-Rad and Chenomx are increasingly offering commercial solutions and services in metabolomics to industry.
I provide below a description of the presentation, discussion and workshop activity for the InnovationWell Session on Application of Metabolomics to Drug Discovery & Development which will take place on Wednesday 18th October ’06 at the InnovationWell meeting at Bryn Mawr. (Follow the Continuation…)
Barry Hardy
References
1. FDA Critical Path Opportunities Report, http://www.fda.gov/oc/initiatives/criticalpath
2. Decision Support in Drug Discovery & Development, http://barryhardy.blogs.com/theferryman/2006/09/decision_suppor.html
3. Knowledge Management in Translational Research, http://barryhardy.blogs.com/theferryman/2006/09/utilising_knowl.html
4. Semantic Web & Drug Development, http://www.innovationwell.net/COMTY_semweb/
5. KM in R&D and ELNs, http://www.innovationwell.net/COMTY_conferenceopenevent/
6. Richard D. Beger, Drug Discovery Today, Vol. 11, pp 429-435, May (2006).
7. R. D. Beger, D. A. Buzata, J.G. Wilkes, Drug Discovery Handbook, Ed. Shayne C. Gad, John Wiley & Sons, pp 227-285 (2005)
8. Predictive Toxicology, http://barryhardy.blogs.com/cheminfostream/2006/09/appyling_predic.html
9. FDA’s Critical Path Initiative: Opportunities for Metabolomics http://www.innovationwell.net/COMTY_mebegerr/
(continued…)
Application of Metabolomics to Drug Discovery & Development
http://innovationwell.net/COMTY_Metabolomics/
InterAction Meeting Session, Bryn Mawr, Philadelphia, USA
Wednesday, 18 October 2006
chaired by George G. Harrigan (Monsanto) & Bruce Kristal (Cornell University)
Presentation Abstracts
FDA’s Critical Path Initiative: Opportunities for Metabolomics
Richard D. Beger, FDA
Richard D. Beger, Branch Chief, Center for Metabolomics, Division of Systems Toxicology, National Center for Toxicological Research, Food and Drug Administration, Jefferson, AR 72079-9502, USA
Critical Path is the FDA's initiative to identify the most pressing developmental problems with medical products and the opportunities for rapid improvement in terms of public health benefit. The primary purpose of the initiative is to ensure that basic scientific discoveries translate into new and better medical treatments. On March 16, 2006, the FDA released a Critical Path Opportunities Report. The report listed 76 Critical Path Opportunities that can be found at http://www.fda.gov/oc/initiatives/criticalpath. The list contains many opportunities in new ‘omic’ technologies including metabolomics. Metabolomics can play a role in a number of these opportunities including better evaluation tools and models for diseases, better identification and quantification of safety biomarkers, and improving the measurement of patient response. The voluntary submission of genomics data (VGDS) to the FDA is now accepting proteomics and metabolomics data sets. Discussion on how the VGDS operates and what its goals are will be discussed. A general overview of the research strategies, goals and results of the Center for Metabolomics will be discussed.
Application of Metabolomics to Biomarker and Off-Target Side Effect Identification in Marketed Drugs and New Chemical Entities
Alvin Berger, Metabolon
Metabolon has developed a global metabolomics technology, using multiple analytical strategies to cover a wide range of polarities and masses, to quantitatively measure and identifiy the repertoire of biochemicals contained in a biological sample.We have used this technology to identify biomarkers for disease as well as to identify off-target side effects in marketed drugs and new chemical entities in development. The discussion will cover an introduction to the technology of metabolomics and a summary of results from several successful projects including patient stratification for Amyotrophic Lateral Sclerosis (ALS or Lou Gehrig’s disease), outcome prediction for pre-term labor and identification of off-targeted effects for several members of a class of commercial drugs. The integration of metabolomics data, through its biochemical context with proteomic, transcriptomic and genomic data is one of the aspects that make metabolomics data so useful.
By using metabolomics to examine the total chemical complexity of biological samples, one may gain a deep understanding of the physiological status of the samples at the time of collection. Thus, if an organism (or cell-line) is subjected to either biological or chemical stress, the specific biochemical abnormalities that are the direct result of that stressor will become apparent. The demonstration of these changes provides a unique way to understand pathological situations because of the ability to directly relate the data in the context of “normal” biochemistry.
Toward Diagnosis of Diabetes by NMR-based Metabolomics
Gregory Banik, Bio-RAD
Chen Peng (1) , Gregory M. Banik* (1), Scott Ramos (2), Tao Wang (3), Bin Xia (3)
(1) Bio-Rad Laboratories, Informatics Division; 3316 Spring Garden Street, Philadelphia, PA 19104, (2) Infometrix, Inc., Suite 250, 10634 East Riverside Drive, Bothell, WA 98011; (3) Beijing NMR Center, Peking University, Beijing, 100871, China
* indicates presenting author
Nuclear magnetic resonance spectroscopy (NMR) is becoming a key tool for understanding the metabolic processes in living systems. Among its many applications, advanced spectroscopic techniques are combined with multivariate statistical approaches to provide diagnostic information for diseases and to identify the changes in the metabolic pathways. This study demonstrates the potentials of this approach by the multivariate analysis of the 1H NMR spectra of serum samples from diabetic and healthy people.
Identification of novel, minor lipids in total lipid extracts of Eschericia coli using Electrospray ionization mass spectrometry
Teresa Garrett, Duke University
The LipidMAPS consortium is a large scale collaborative effort to characterize the global changes in lipid metabolites. The consortium has developed methods for the extraction and analysis of lipid by mass spectrometry, a comprehensive lipid classification system, lipid database and computational tools for the analysis of lipids by mass spectrometry. As part of LipidMAPS we have initiated a project to identify the novel lipids in the mouse macrophage RAW 264.7. In order to develop techniques and methods to facilitate novel lipid identification we have begun surveying the Escherichia coli lipidome for novel lipid species. We have developed extraction and fractionation procedures and begun analyzing those lipid extracts using negative-ion electrospray ionization time of flight mass spectroscopy. Studies of DEAE fractionated lipids have led to the identification of several minor lipid species in the fractions which elute with high salt. We have identified several enterobacterial common antigen precursors in wild-type lipid extracts. In addition a novel lipid, which we propose to be phosphatidylserylglutamate, has been identified. The structure of this lipid has been confirmed by comparison of the collisionally induced decomposition mass spectrum and LCMS retention profile with that of a semi-synthetic standard. The strategy of combining pre-fractionation using anion exchange chromatography, LCMS and collision-induced decomposition mass spectroscopy will continue to lead to the identification of novel, minor lipids. The techniques and strategies developed using E. coli lipid extracts are translatable to the analysis of the minor lipids of from other organisms, cell lines or tissues. This work was supported in part by the LIPID MAPS Large Scale Collaborative Grant number GM069338 from the NIH.
A pharma perspective on metabolomics - the opportunities and realities
Susan Connor, GlaxoSmithKline
Metabolomics has now been used within the Pharmaceutical Industry for several years, although it is fair to say that it has still to reach its full potential and prove its worth commercially. This is in part because historically it has received only scant attention and resource until very recently, and in part because of the inherent complexity of the data such that unskilled interpretation of the data can be misleading, or even completely unsuccessful in providing novel biological information. This imbalance is now in-part being redressed and we are starting to see its full capability realised. An overview will be presented of where opportunities exist within Pharma for the application of metabolomics, with examples presented from recent work within GSK. The opportunities discussed range from Drug Discovery lead optimisation through toxicity assessment in non-clinical drug development, to the potential for biomarker discovery to aid clinical evaluation. Also to be presented will be an outline of the realities and challenges of doing this type of work in Industry and how we see the usefulness of the interface with academia and commercial suppliers of metabolomics data.
Tracer Substrate-based Metabolomics and the 2005 Nobel Prize award in Physiology & Medicine
Greg Maguire, SiDMAP
The Nobel Prize in Physiology or Medicine for 2005 was awarded to Barry J. Marshall and J. Robin Warren "for their discovery of the bacterium Helicobacter pylori and its role in gastritis and peptic ulcer disease." There is an important metabolic component to this disease, and the use of tracer-based metabolomics was important to assigning H. pylori as the causative agent in peptic ulcers and became an important diagnostic tool for the treatment of ulcers. Tracer-based metabolomics, or SiDMAP (stable-isotope dynamic metabolic profiling) as it is often called in the literature, in this case was used as a simple breath test to detect CO2 that had been labeled with 13C from a precursor molecule, urea that contained the original 13C label. This technique provided a simple clinical diagnostic tool allowing for the identification of the ulcer type and hence the correct treatment regimen for the ulcer. Tracer-based metabolomics continues to be an important and relatively easy to use tool for studying biochemical pathways in cells, animals, and humans, and provides a new and powerful method as a clinical diagnostic in the drug discovery and development process.
Serum Markers of Caloric Restriction
Bruce Kristal, Cornell University
Dietary or caloric restriction (DR) is the most potent and reproducible known means of increasing longevity and reducing morbidity in mammals. Exploratory studies previously identified 93 redox-active small molecules from sera (measured by HPLC coupled with coulometric detector arrays) with potential to distinguish dietary groups in both male and female rats. Classification and predictive power were addressed using megavariate data analysis approaches. The compounds weakly distinguished AL and DR samples by Principal Components Analysis (PCA) due to noise resulting from inter-cohort sampling. Soft Independent Modeling of Class Analogy, which builds independent PCA models of each class of interest, distinguished groups with 95% accuracy, but overfit models built on single cohorts. Partial Least Squares Projection to Latent Structures Discriminant Analysis, a projection method optimized for class separation, in contrast, built models with >95% accuracy in distinguishing groups without obvious cohort interference. Data processing choices of transformation, scaling, and winsorizing (outlier removal) each affected strength of the models, and, in some cases, revealed distinct metabolites to be of importance in building these models, often in gender-specific ways. Diets varying in extent and duration of DR were used to develop models for intermediate caloric intakes, which are more relevant for human studies. Partial Least Squares models had r2 values of 0.89 with respect to prediction of caloric intake at the individual level. We have now adapted these markers for use in humans. We will present these modelling approaches, the models and their ability to distinguish sera based on caloric intake, and the potential for moving these markers to epidemiological studies in human sera.
Uses and Abuses of Metabonomics in Pharmaceutical Preclinical Safety Assessment
Don Robertson, Pfizer
In recent years, metabonomics has received a great deal of attention as a potential tool for toxicologists in the pharmaceutical preclinical arena. We are at a point where we can ask if the early promise of the technology has been realized or not. Preconceived expectations of the technology and differing definitions of success can cause diverse conclusions from the same set of data. Furthermore, basic ignorance or neglect of basic biology leads to oft erroneous assumptions and conclusions. While the hope for a rapid throughput safety screen has not disappeared, the complexity of basic biochemistry stymies simplicity. Perhaps our animal models were never that simple in the first place and we now need to start understanding (if not embracing) biologic diversity in biologic response. The benefits of understanding this complexity may allow us to understand what we have always attributed to interanimal or interstudy variation. This in turn may allow us to run studies with fewer animals and to explain the odd responder that frequently reduces apparent safety margins. Metabonomics enables early and rapid assessment of safety related issues at a depth of biochemical analysis previously impractical for routine use with mechanisms and safety biomarkers a frequent product of its application. It also is opening new understanding of our animal models that we have taken for granted for much too long. This presentation will elaborate on the applications of metabonomics in preclinical safety assessment, what has worked and what still needs to be worked out with regard to screening, biomarkers and mechanistic toxicology.
The ‘Metabolomic Standards Initative’ and implementation of reporting requirements in database solutions
Oliver Fiehn, UC Davis Genome Center
Metabolite concentrations in cellular systems are very much dependent on the physiological, environmental and genetic status of an organism and are regarded as the ultimate result of cellular regulation, resulting in the visible phenotypes. Therefore, the comprehensive analysis of metabolite levels and fluxes renders a suitable tool for assessing the degree of perturbation in biological systems. Lessons derived from development of other –omics areas (genomics, proteomics, and transcriptomics) have shown that large scale comparisons and interpretations will require the re-use of data over long periods of time and by multiple laboratories with different expertise and backgrounds. Reaching this goal will require standardization of reporting structures of metabolomic studies both for journal publication purposes, regulatory deposition and database dissemination (e.g. MIAME). An initiative by the Metabolomics Society is presented that aims to define important aspects of metabolomic workflows. These include biological study designs, chemical analysis and data processing and the ontologies that are necessary in this framework.
The presentation will exemplify the incorporation of these standards and some results using metabolite profiling of spleen samples from UCP2 and UCP3 knockout mice that have dysfunctional variants of two genes important for energy metabolism in mammals. We have used profiling of primary metabolites by GC-TOF analysis concomitant with database annotations and a variety of statistical tools to analyze the phenotypic consequences of UCP knockout mice under different dietary regimes. Statistical evaluation and interpretation of data is highlighted for primary metabolites that were annotated through the laboratory’s combined database system, SetupX and BinBase. SetupX details the setup of experiments, also called biological sample context metadata or plainly ‘study design’. Such annotated study design metadata empower complex queries across many different diets, genotypes, interventions or combinations of these. Hence, metabolomic data receive increased value when studies can be compared across studies in order to distinguish unspecific stress responses from biomarkers that are highly specific for a certain disease or a defined organ. The second database was BinBase, our metabolite annotation repository that utilizes a multiple filtering system for validating genuine metabolic signals across experimental classes while discriminating noisy and inconsistent spectra. After deconvolution of co-eluting peaks, a chromatogram comprises some 400-800 spectra, or a daily output of some 20,000 spectra per day and instrument. BinBase then imports these spectra with accompanying metadata such as the ‘unique (model) masses’ that best describe the presence of a peak in the local environment. Compounds are reported by a variety of identifiers, most importantly the freely accessible PubChem identifiers that comprise links to the KEGG database but also hint to medical literature. Therefore, the interplay of SetupX and BinBase enabled tracking both the origin and the metabolic annotations of samples, with the goal of consistently storing and disseminating data on mouse metabolic phenotypes for use of the broader scientific community.
Studies of Drug-Induced Liver Injury using Comprehensive 2D Gas Chromatography with Time-of-Flight Detection
Eric Nemec, Leco Corporation
Tincuta Veriotti (1), Jack Cochran (1), Susan Sumner (2) and Eric Nemec (1)
(1) Leco Corporation, St. Joseph, MI, USA; (2) Drug Metabolism and Pharmacokinetics, RTI International, Research Triangle Park, NC, USA
The interest in studying the metabolom (metabolomics) has increased tremendously in the last few years. From species fingerprinting, studies of metabolites changes in response to external stimuli, network regulation, to metabolomic characterization of health versus diseased tissue, the study of the metabolom found its way in many areas of the life sciences. Together with genomics, transcriptomics, and proteomics, metabolomics can provide insights to understand mechanisms associated with the on-set of disease, and to stage disease and follow therapeutic intervention.
The complexity of the data obtained for metabolomics studies along with the need for very high sample throughput, creates a significant challenge for the analytical chemist. Fast acquisition rates and absence of spectral skewing make TOFMS the ideal detector for metabolomics analysis when gas chromatography is selected as the separation method. The additional peak capacity obtained from comprehensive two-dimensional gas chromatography (GCxGC) as well as the improved detection limits obtained from this technique can add more benefits to the analysis.
This study will be focused on the use of GCxGC-TOFMS technology to analyze urine from rats administered isoniazid at effect and no effect levels. Isoniazid is used in the treatment of tuberculosis, and is known to cause idiosyncratic liver injury in humans. Better biomarkers or marker profiles are needed in order to personalize the treatment of such drugs, or to enable clinicians to more accurately predict early onset of toxicity. Metabolomics provides a promise for the development of endogenous based marker profiles to predict the early onset of drug-induced liver injury, as well as to gain insights into mechanisms of idiosyncratic responses.
Workshop Abstracts
NMR-based Positional Isotopomer Analysis in Metabolomics
Andrew N. Lane, JG Brown Cancer Center, U. Louisville
Andrew N. Lane, Teresa W-M. Fan, JG Brown Cancer Center, U. Louisville. KY 40202
The ability to trace individual atoms through metabolic pathways is immensely powerful for understanding the control of metabolism and dysregulation in disease states. NMR offers the potential of following the fate of individual atoms from a precursor molecule to end products using stable isotopes, under a wide variety of conditions including in vivo and in cell culture.
Here recent developments in stable isotope editing of NMR spectra of mixtures of metabolites in crude extracts by 1D and 2D methods will be presented. Isotopomer distributions in metabolites in the extract mixtures can be determined using a variety of isotope-edited experiments (e.g. HSQC, HSQC-TOCSY and HCCH-TOCSY) and quantified by TOCSY. The advantages of the 2D H-1 detection approach include sensitivity and resolution with simultaneous identification of isotope distributions of all protiated carbon sites. The dependence of quantitative precision and accuracy on sample integrity, experimental design and instrumental considerations will be treated with concrete examples based on standard and real samples.The information content of these various experiments will be discussed, along with techniques for determining isotopomer distributions and quantification, and the relationship to pathway delineation and relative flux determination. The choice of labeled precursor compounds for specific and general metabolomics questions will be introduced. These principles will be illustrated using examples from cancer cell biology and recent results from human studies.
Linking metabolic profiles to biological outcome
S. Stanley Young, National Institute of Statistical Sciences
S. Stanley Young, Assistant Director of Bioinformatics, National Institute of Statistical Sciences
Paul Fogel, Consultant, France
Doug Hawkins, University of Minnesota
Microarray, proteomics, metabolomics, etc., all produce data sets where there are many more predictor variables than observations. There are correlations among these variables; indeed, the many variables/predictors can not all be independent of one another. The correlations can be utilized to improve the linking to outcomes (disease, drug effects, etc.) to predictors. New inference methods will be presented which combine statistical testing with matrix factorization. The methods will be demonstrated using a real data set.
Understanding Metabolomics Mixtures with Principal Components Analysis
Gregory Banik, Bio-RAD
L. Scott Ramos (1), Gregory M. Banik (2)*
(1) Infometrix, Inc., Suite 250, 10634 East Riverside Drive, Bothell, WA 98011; (2) Bio-Rad Laboratories, Informatics Division; 3316 Spring Garden Street, Philadelphia, PA 19104
In toxicology studies, alterations to the intrinsic in vivo biochemical profile of an animal can be monitored via NMR spectroscopic analysis of urine. Chemometric processing of the NMR data can enhance understanding of the toxic effects; in particular, examination of the principal component analysis (PCA) scores provides insight into these data in a time-efficient manner.
InnovationWell meeting workshop conference management executive Bryn Mawr Philadelphia Critical Pathbiomarkers metabolomics toxicology personalised medicine KM Life Sciences Pharma Drug Discovery Drug Safety Research and Development Drug Development Healthcare Innovation Knowledge Management events
Comments