The ability to predict the risk for serious drug-induced adverse reactions first requires a large patient database for characterization and validation of genetic markers. To date, the iSAEC has partnered with academic investigators to build a variety of SAE research cohorts. To obtain the research subjects for SAE research in the scale and diversity required, new collaborations with large integrated researcher groups and health networks (with electronic health records) will need to be developed. The Phenotype Standardization Project (PSP) was initiated to standardize phenotypic definitions, thereby facilitating recruitment without sacrificing the reliability of patient classification. Three phenotypes have been considered in this initial phase:
- Drug-induced liver injury (download standardization guidelines).
- Drug-induced skin injury (download standardization guidelines).
- Drug-induced torsades de pointes (download standardization guidelines).
Overview and RationaleThere is increasing appreciation that genomic variants can be used to successfully predict individual drug response, both in terms of efficacy and toxicity. Mechanisms most often recognized to date center on genes responsible for the metabolism and/or transport of the drug, and/or genes responsible for the mechanism of action of the compound. The increasing use of genome wide association scans, and in the near future, of whole genome sequencing technologies, are also likely to identify variants in genes that have not previously been considered to be involved in either the disposition or mode of action of a drug, thereby providing novel insights into the pathways involved in drug response.
Adverse drug reactions (ADRs) are common, but fortunately most are mild or moderate in intensity; however, some ADRs are serious and may have lasting or fatal effects.(1-3) Some ADRs are extensions of a drug’s desired pharmacologic effects, while others are the result of off-target effects often in patients with individual metabolic or immunologic variations that defy conventional modes of prediction.(2, 4) Many serious ADRs occur at a low incidence—accounting for only 1 in 3000 hospitalizations—and are statistically unlikely to emerge until after a drug not only achieves regulatory approval but is used in a sufficiently large pool of patients.(1) Indeed, the recognition of severe ADRs early in the evaluation of a new drug candidate makes regulatory approval difficult.
A Genetic Approach to Improving Drug Safety
Serious ADRs occur unpredictably and sometimes phenocopy genetic disease, suggesting a genetic contribution to risk. The completion of the human genome project and the advent of whole genome mapping techniques have allowed researchers to begin to characterize the genetic components underlying serious ADRs.(4) The identification and validation of these genetic markers will allow screening of patients at risk for serious ADRs and in some cases, tailor treatment regimens accordingly. The realization of this goal however requires a large and sufficiently diverse patient database and development of screening tests with sufficient sensitivity, specificity and predictive value.(5)
Accurate and standardized phenotyping is the vital initial step. For example, given the relatively rare incidence of these serious ADRs, only small numbers of well-phenotyped patients can be identified from a single center, necessitating the pooling of samples from multiple centers, often from different countries. A standardized phenotype then becomes crucial for accurate pooling of data. Without a sufficiently large patient database, it would not be possible to identify genetic factors of modest effect or by ethnicity. Another critical point is that phenotypic misclassification may dilute or even mask a significant association with a gene locus or alternatively provide a false positive association.
The International Serious Adverse Event Consortium, in collaboration with a number of other stakeholders, has initiated the PSP. The initial goal of the project is to develop standardized phenotypic definitions for 3 different ADRs: drug-induced liver injury (DILI), drug-induced skin injury (DISI), and drug-induced torsades de pointes (DITdP). Although the PSP was initiated to address the need for sufficient patient cohorts to study the genomic basis of specific drug-related adverse reactions and ultimately to identify individual at-risk patients a priori, important secondary goals include (a) ensuring epidemiologically correct patient classification to help clinicians, industry and regulators in identifying the clinical burden of different types of adverse reactions and (b) providing potential insight into novel pathways that would aid safer drug development. Ultimately, it may be possible to firm up individual patient diagnosis so that appropriate treatment/care can be provided to reduce morbidity and mortality. Furthermore, although spontaneous adverse event reporting systems represent an important form of pharmacovigilance, the quality and reliability of such reports is variable. Therefore, the implementation of standardized phenotypes for ADR reporting systems implemented in many jurisdictions would not only make classification easier but would also help with targeted follow-up in cases with inadequate information.
The PSP is a collaborative effort between the iSAEC, the Wellcome Trust and the U.S. Food and Drug Administration. The overall PSP program is being chaired by Prof. Munir Pirmohamed (NHS Chair of Pharmacogenetics, Department of Pharmacology, University of Liverpool). Three expert working groups (EWGs) have been established to develop the standardization guidelines for the three initial phenotypes of interest.The EWGs have the following chairmanships:
Drug-induced skin injury (DISI) EWG – Munir Pirmohamed
Drug-induced liver injury (DILI) – Ann Daly (Newcastle University), Guru Aithal (University Of Nottingham)
Prolonged QT/Torsades de Pointes (DITdP) – Dan Roden (Vanderbilt University), Elijah Behr (St George's Hospital, London)
In each case, the Expert Working Groups are comprised of individuals with varied backgrounds, representing specialists in each relevant field (Munir Pirmohamed, Ann Daly, Paul Watkins (UNC), Guru Aithal, Dan Roden, Elijah Behr), electronic medical database managers, regulatory agencies (FDA, EMEA and the MHLW), researchers and other interested parties such as the Wellcome Trust and the iSAEC pharmaceutical members.
The PSP project leader is Julian Arbuckle (GlaxoSmithKlein), consultant to the iSAEC, and supported by a Coordinating Committee which includes Arthur Holden (iSAEC), Michael Dunn (Wellcome Trust), Munir Pirmohamed (Clinical Chairman), and ShaAvhree Buckman (FDA)
Each of the three ADR Expert Working Groups held meetings and teleconferences in relation to their phenotypic areas before a meeting was organized at the Wellcome Trust Hinxton Campus in Hinxton, England on March 16th, 2010. The goal of each Expert Working Group was to identify the phenotypic requirements that would allow accurate identification of patients with a given serious ADR and to develop a corresponding clinical classification algorithm to assist in the recruitment of such cases. In addition, participants provided recommendations for key elements to include in the patient record (for use as covariates in a genomic analysis, for example) and the prioritization of patient sample collection.
Participants considered the recruitment of patients using multiple methods, noting their various limitations. Clearly the most accurate methodology is to recruit patients at the time they are experiencing the reaction (i.e., in the acute state). However, this approach alone is not practical and will prolong case recruitment due not only to the relatively uncommon incidence of severe ADRs but also the lack of proper infrastructure for recruitment in resource-poor settings. Therefore, we also considered that there may be some novel methodologies for identifying and recruiting patients with serious ADRs, for example using electronic medical records. Where a patient is identified through electronic medical records as having an ADR, and can have the phenotype verified using standardised criteria, they should be recruited. However, the challenge will be in identifying these patients in the first place given that the coding used in different databases may be variable in terms of sensitivity, specificity, and accuracy. In addition, diagnostic criteria that may be sought in prospective studies may be absent in a retrospective approach since documentation of the nature of the reaction in case notes is often incomplete. In such circumstances, a pragmatic approach may be required so that valuable patients are not lost from studies, while at the same time ensuring that the researcher does not over-compensate to such a degree that it leads to over-inclusion, patient heterogeneity, and in the end, the inability to identify any genetic associations.
In conclusion, the PSP has undertaken phenotype standardization for three serious forms of ADRs affecting the skin, liver and heart, which collectively cause significant morbidity and mortality and have been responsible for the withdrawal of drugs from the market. All three standardized phenotypes will appear in issues of the Clinical Pharmacology and Therapeutics journal. The PSP will continuously review and re-evaluate these standardized phenotypes in response to new clinical or molecular findings and is committed to outlining standardized phenotypes for other types of drug-induced adverse reactions as well. Once available, the adoption of common phenotypic criteria is likely to benefit future research efforts, whether conducted by academic institutions, pharmaceutical companies or regulatory bodies, with the ultimate aim of creating a robust - and ultimately predictive - pharmacogenomic database that is relevant for all ethnicities.
(1) Pirmohamed, M., James, S., Meakin, S., Green, C., Scott, A.K., Walley, T.J. et al. Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. BMJ 329, 15-9 (2004).
(2) Riedl, M.A. & Casillas, A.M. Adverse drug reactions: types and treatment options. Am Fam Physician 68, 1781-90 (2003).
(3) Wester, K., Jonsson, A.K., Spigset, O., Druid, H. & Hagg, S. Incidence of fatal adverse drug reactions: a population based study. Br J Clin Pharmacol 65, 573-9 (2008).
(4) Pirmohamed, M. Pharmacogenetics of idiosyncratic adverse drug reactions. Handb Exp Pharmacol, 477-91.
(5) Hughes, A.R., Brothers, C.H., Mosteller, M., Spreen, W.R. & Burns, D.K. Genetic association studies to detect adverse drug reactions: abacavir hypersensitivity as an example. Pharmacogenomics 10, 225-33 (2009).