- Genome databases
- Open Access
A useful tool for drug interaction evaluation: The University of Washington Metabolism and Transport Drug Interaction Database
© Henry Stewart Publications 2010
- Received: 9 July 2010
- Accepted: 9 July 2010
- Published: 1 October 2010
The Metabolism and Transport Drug Interaction Database (http://www.druginteractioninfo.org) is a web-based research and analysis tool developed in the Department of Pharmaceutics at the University of Washington. The database has the largest manually curated collection of data related to drug interactions in humans. The tool integrates information from the literature, public repositories, reference textbooks, guideline documents, product prescribing labels and clinical review sections of new drug approval (NDA) packages. The database's easy-to-use web portal offers tools for visualisation, reporting and filtering of information. The database helps scientists to mine kinetics information for drug-metabolising enzymes and transporters, to assess the extent of in vivo drug interaction studies, as well as case reports for drugs, therapeutic proteins, food products and herbal derivatives. This review provides a brief description of the database organisation, its search functionalities and examples of use.
- drug-drug interactions
- cytochrome P450 enzymes
Adverse drug reactions (ADRs) remain one of the leading causes of morbidity and mortality in healthcare. In January 2000 the Institute of Medicine reported that between 44,000 and 98,000 deaths occur annually from medical errors in American hospitals . Of this total, an estimated 7,000 deaths occur due to ADRs. It is estimated that drug-drug interactions (DDIs) represent 3-5 per cent of all in-hospital medication errors and that they are also an important cause of patient visits to emergency departments  Among the factors that contribute to the occurrence of a DDI are patient age, number and type of concomitant medications and disease stage. In recent years, while healthcare providers have been offered access to and have benefitted from numerous drug information tools that have provided them with guidance on how drugs can be co-administered, researchers within the drug development community have had access to a more limited portfolio of data repositories. These scientists need to browse the vast literature for primary scientific data (ie datasets on metabolic isozymes, transporters, substrates, inducers, and inhibitors) that will provide them with context for their research findings and help with their drug interaction programme.
Metabolism and Transport Drug Interaction Database (DIDB) users
Examples of database use
Pharmaceutical industry & CROs
DMPK Clinical pharmacology Clinical
Tool for IVIVE Modelling: to define acceptable input parameters and validate models
Helps optimize design of in vitro and in vivo drug interaction studies
Provides context for results obtained for candidate compounds
Provides access to labelling of recently marketed drugs
DIDB as a research tool: publications - Presentations
Provides context for results submitted for candidate compounds
Helps update guidance documents (DDI, pharmacogenetics)
DIDB as a research tool: publications - presentations
Resource for courses on DDI
DIDB as a research tool: publications - presentations
The database contains in vitro and in vivo kinetics information for drug-metabolising enzymes and transporters, pharmacokinetics parameters/pharma-codynamic measures and side effects reported in clinical drug interaction studies. Each dataset integrates both the experimental design and the primary results. The database can be searched not only by main concepts in the field of drug interaction (ie drug name, enzyme, transporter, etc.), but also by related topics such as QTc prolongation or impact of genetic variability on drug exposure in the context of a drug interaction. Even though the DIDB was initially designed for evaluation of drug interaction profiles of small molecule compounds, a new dataset related to therapeutic proteins has been added recently.
A menu of pre-defined queries allows users to analyse and integrate both preclinical and clinical data. In addition, drug and disease monographs (composed by the DIDB editorial team) add to the information mining and data retrieval power of the queries by highlighting the most relevant datasets. As shown previously,  the DIDB has been used extensively by researchers and clinicians interested in correlating in vitro and in vivo findings associated with metabolic enzymes and transporters. The database is also widely used in clinical programmes, including the management of drug interactions of new drugs in multicentre trials 
The DIDB application has a typical multi-tier architecture in a Microsoft®.NET environment. (The web part of the database, which is accessed by the user over the internet, is hosted on a Microsoft Windows 2003 server running IIS and version 2.0 of the ASP.NET framework. All data are stored on a Microsoft SQL Server 2005 database.) The use of the web facilitates worldwide access, as well as upgrades and updates; the DIDB is updated daily.
The current DIDB datasets are extracted from more than 8,300 published articles referenced in PubMed (from 1966 to the present), 70 new drug applications (NDAs) and 368 product labels (from 1998 to the present). The unit of information (citation) is either a published research article or the 'NDA Clinical Reviews' section available from the FDA Approved Drug Products website .
Detailed records are generated from each research article or NDA, highlighting study results as well as experimental conditions. The records are structured in the database according to a defined hierarchy; for example, relevant information collected from in vitro studies pertains to the role of particular metabolic enzymes in the various metabolic pathways of substrates and the inhibition and induction spectra of drugs toward metabolic enzymes. Particular attention is paid to experimental conditions used in determination of enzyme kinetics parameters, including K m , K i , IC50, KI-Kinact and EC50. In vivo studies include pharmacokinetic studies with blood level measurements, pharmacokinetic-pharmacodynamic studies, as well as case reports.
Recently, a new section analysing DDIs in the context of specific diseases and their co-morbidities (Disease-Oriented Database) was added to the DIDB. This section allows users to retrieve overall summaries on DDIs related not only to drugs used to treat the disease, but also to drugs used to treat the main co-morbidities of that disease.
Defining the issue (background and question)
Selecting the search strategy
Analysing and interpreting the result.
Example 1. Notion of interchangeability of CYP3A substrates
Background and question
In its last guidance document,  the US Food and Drug Administration (FDA) proposed that CYP3A inhibitors be classified based on the magnitude of change in plasma area under the curve (AUC) of oral midazolam or other sensitive CYP3A substrate. For instance, if the ratio AUCinhibited/AUCcontrol (AUCR) of oral midazolam (or other sensitive CYP3A substrate) is ≥ 5, the inhibitor is considered a strong CYP3A inhibitor. If the ratio is ≥ 2 ≤ 5, the inhibitor is classified as moderate and, finally, if the ratio is ≥ 1.25 ≤ 2, it is considered a weak inhibitor. A similar classification has been proposed for the other CYP enzymes. By using a clear and consistent categorisation of drugs as substrates and inhibitors, the FDA hopes to facilitate analyses across DDI studies and to help healthcare providers to administer these drugs safely through a consistent labelling language. In addition to the CYP3A probe substrate midazolam, the FDA provides a list of other sensitive CYP3A substrates (ie that exhibit an AUCR of ≥ 5 when given concomitantly with a CYP3A inhibitor). These sensitive substrates are: budesonide, buspirone, eplerenone, eletriptan, felodipine, fluticasone, lovastatin, midazolam, saquinavir, sildenafil, simvastatin, triazolam and vardenafil.
Broad applicability of the above proposal rests on the assumption that the classification of a CYP3A inhibitor would be independent of the sensitive substrate used. In order to test the assumption of substrate independence, the DIDB was interrogated for: 1) a comprehensive list of sensitive substrates and 2) any discrepancies when classifying inhibitors with different sensitive substrates.
The AUCR (ie AUCinhibited/AUCcontrol ) of substrates was used as the metric to assess the degree of interaction and to classify inhibitors.
Step 1: Identify all inhibitors of midazolam and retrieve the maximal midazolam AUC R observed
Each precipitant (inhibitor) in the list has its own folder containing more detailed information: the exact value of the AUC change observed in the study; dosing regimen of the object (substrate) and the precipitant (inhibitor); and a link to the source article identified by either an accession number (PMID number) or an NDA number. By clicking directly on this number, the full description of the article can be retrieved (study design, population, the drug's dosing regimen, results of pharmacokinetic measurements, side effects, etc.). Two additional features are available next to the accession/NDA number: abstract of the article (visualised with the icon) and reference PK parameters for drugs (retrieved by clicking on the icon).
There are several options for displaying the results in a table and performing filter operations, as well as exporting capabilities into Microsoft Excel® or Microsoft Word®.
Step 2: Using CYP3A inhibitors to identify all other sensitive substrates
Step 3: Search for discrepancies when classifying inhibitors with different sensitive substrates
Examples of exceptions to midazolam classification for seven CYP3A4 inhibitors
Classification with sensitive substrate
Potent with midazolam
Moderate with midazolam
Analysis and interpretation
These discrepancies do not invalidate the proposed classification. Some of these differences could arise simply because of the absolute boundaries (2.0- and 5.0-fold) of the classification; some discrepancies could also be related to: (i) transporter effects in specific substrate-inhibitor pairs; (ii) intrinsic differences among substrates in sensitivity to inhibition (including fraction metabolised by CYP3A and intestinal metabolism); and/or (iii) inhibition of minor enzymes by CYP3A inhibitors . Similar findings have recently been reported by other groups [8, 9]. These differences in in vivo sensitivities of CYP3A substrates need to be considered when selecting a CYP3A probe substrate for clinical DDI studies.
Example 2. Analysis of drug interactions in the context of a disease and its co-morbidities
Background and question
Assessment of the DDI risk potential of a new molecular entity (NME) during drug development takes into consideration the clinical outcome of administration of the NME and focuses not only on the drugs used to treat the primary disease, but also on those used to treat co-morbidities. Moreover, questions arise regarding the roles of environmental factors (food, herbal medications) and patient characteristics (genotype, age, etc.) that may also alter drug disposition.
In the problem at hand, an NME is being developed for the treatment of hypertension. This NME is mainly metabolised by CYP2D6, with some contribution from CYP3A4. It was also found that this NME is a moderate CYP3A4 inhibitor, yielding an AUC ratio of midazolam of 3.2.
Because hypertension is a condition that often co-exists with hyperlipidaemia, the developer wanted to evaluate whether drugs that treat this condition would have any clinically relevant impact on the disposition of this new antihypertensive drug; in addition, given the inhibitory profile of this NME, the user wanted to determine whether any drugs used in hyperlipidaemia were likely to be affected by this new antihypertensive.
For each compound shown in Figure 9, the table provides the enzymes and/or transporters affected and a corresponding DDI risk level. Four risk levels have been created based on a combination of the following characteristics: (i) sensitivity to inhibition and induction of the involved enzymes and/or transporters; (ii) therapeutic range; (iii) documented clinical interactions.
Analysing results of the first query
The inhibitory profiles of the ten antilipaemics listed in Figure 9 show that most of these drugs do not exhibit any risk of increasing the exposure of co-administered drugs. Only one compound, gemfibrozil, exhibits a relatively high inhibitory risk potential towards CYP2C8 and OATP1B1. None of the drugs are expected to alter the disposition of CYP2D6 substrates such as the NME.
The table in Figure 10 shows that three statins (atorvastatin, simvastatin and lovastatin) are extensively metabolised by CYP3A4.
Analyzing the results of the second query
Analysis and interpretation
The new disease section has multiple uses and it allows a rapid assessment of the DDI potential of an NME in comparison with other marketed drugs used to treat the same disease, and also the DDI potential of this NME with drugs used to treat co-morbidities of that disease. Additionally, the complete DDI profile of a disease is provided in summarised tabulated views.
For almost a decade, the DIDB has been widely used by scientists and clinicians working in the field of DDI. The tool is constantly being optimised as a result of feedback from a large base of users, including requests for specific searches. These can be in the form of new queries or special reports tailored by the DIDB team. New features currently being developed include the addition of datasets pertaining to emerging areas (therapeutic proteins, pharmacogenomics). The DIDB is also enhanced with tools that allow users to focus rapidly on important DDI reports and sort through the large body of literature. Examples of such tools include: graphical displays of the extent of DDI (AUCR, changes in the clearance of substrates) and 'flagging' important drug characteristics (narrow therapeutic range drugs, probe substrates, potent inhibitors or inducers).
- Committee on Quality of Health Care in America Institute of Medicine: Errors in health care: A leading cause of death and injury. To Err Is Human: Building a Safer Health System. Edited by: Kohn LT, et al. 2000, National Academy Press, Washington, DC, 26-48.Google Scholar
- Leape LL, Bates DW, Cullen DJ, Cooper J, et al: Systems analysis of adverse drug events. ADE prevention study group. JAMA. 1995, 274: 35-43. 10.1001/jama.1995.03530010049034.View ArticlePubMedGoogle Scholar
- Hachad H, Ragueneau-Majlessi I, Levy R: Metabolism and transport drug interaction database: A web-based research and analysis tool. Drug-Drug Interactions. Edited by: Rodrigues AD. 2008, Informa Healthcare, New York, NY, 567-579. 2Google Scholar
- Hachad H, Ragueneau-Majlessi I, Levy R: Management of drug interactions of new drugs in multicenter trials using the metabolism and transport drug interaction database. Enzyme and Transporter-Based Drug-Drug Interactions Progress and Future Challenges. Edited by: Pang KS, et al. 2009, Springer, 371-386.Google Scholar
- FDA Approved Drug Products Drugs@FDA. [http://www.accessdata.fda.gov/scripts/cder/drugsatfda/]
- Guidance for Industry: Drug Interaction Studies -- Study design, data analysis, and implications for dosing and labeling. Draft guidance (Released September 2006). [http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM072101.pdf]
- Ragueneau-Majlessi I, Boulenc X, Rauch C, Hachad H, et al: Quantitative correlations among CYP3A sensitive substrates and inhibitors: literature analysis. Curr Drug Metab. 2007, 8: 810-814. 10.2174/138920007782798135.View ArticlePubMedGoogle Scholar
- Foti RS, Rock DA, Wienkers LC, Wahlstrom JL: Selection of alternative CYP3A4 probe substrates for clinical drug interaction studies using in vitro data and in vivo simulation. Drug Metab Dispos. 2010, 38: 981-987. 10.1124/dmd.110.032094.View ArticlePubMedGoogle Scholar
- Venkatakrishnan K, Obach RS, Rostami-Hodjegan A: Mechanism-based inactivation of human cytochrome P450 enzymes: Strategies for diagnosis and drug-drug interaction risk assessment. Xenobiotica. 2007, 37: 225-256.View ArticleGoogle Scholar