Genome-wide association studies (GWAS) provide an unbiased approach for the discovery of potential mechanisms and pathways that underlie human characteristics, diseases and drug responses. As common diseases are frequently associated with genetic variants that exert small effect sizes, GWAS have increasingly focused on large cohorts of patients. For example, a recent mega-GWAS of schizophrenia included more than 37,000 individuals with schizophrenia and 113,000 controls, which enabled the identification of 108 genetic regions linked to schizophrenia, out of which 83 were novel, as discussed in a recent news article (Massive schizophrenia genomics study offers new drug directions. Nat. Rev. Drug Discov. 13, 641–642 (2014))1.This study was made possible through the Psychiatric Genomics Consortium, a large group of more than 500 scientists from 80 different institutions.
In the past decade, GWAS have led to a wealth of new information on the physiological and pathophysiological mechanisms that mediate many human attributes, such as height and weight, as well as an increased understanding of many common diseases. However, less than 10% of published GWAS have focused on the genetic contribution to variation in therapeutic drug responses and adverse drug reactions (ADRs): that is, pharmacogenomics (PGx) GWAS. The dearth of PGx GWAS has slowed our understanding of pharmacological mechanisms, specifically the mechanisms responsible for drug disposition, action and toxicity.
In this short article, we discuss published PGx GWAS data and highlight major areas in which PGx GWAS are needed to advance the fields of pharmacology, toxicology and clinical drug response. We conclude with a short description of recent efforts in PGx GWAS and suggestions for future directions.
PGx GWAS of drug response and toxicity
PGx GWAS have resulted in the identification of several actionable genetic variants that have been genotyped and used to inform drug selection and dosage. The most notable examples include genetic variants in CYP2C19, which guide the choice of antiplatelet drug and dosage2, as well as variants in interleukin-28B (IL28B; also known as IFNL3), which provide information as to whether additional drugs should be included in therapeutic regimens of pegylated interferon-α for hepatitis C infections3. Furthermore, examples of actionable genetic variants associated with toxicity are encoded in the human leukocyte antigen (HLA) loci. In cases such as this, information about specific HLA variants (for example, HLA-B*5701 and HLA-B*1502) is used by clinicians to reduce the incidence of hypersensitivity reactions to abacavir4, carbamazepine5 and allopurinol6. In fact, in the approved abacavir product label, the US Food and Drug Administration (FDA) recommends that genetic testing for HLA-B*5701 should be performed before abacavir is prescribed. However, despite their potential therapeutic impact, there have been very few PGx GWAS that focused on drug toxicities. Indeed, the US National Human Genome Research Institute and European Bioinformatics Institute GWAS Catalog lists only 69 such studies.
ADRs represent a major source of morbidity and mortality across most populations and affect virtually all organ systems and physiological processes. Drug-induced allergies, such as Stevens–Johnson syndrome (SJS), toxic epidermal necrolysis (TEN; also known as Lyell syndrome), drug-induced liver injury and agranulocytosis, are associated with an array of medications used to treat various human diseases. These allergies, which may be benign (for example, a mild skin rash) or severe (for example, agranulocytosis), are classed as type B ADRs, which result from a specific immunological response to a medication. As such, they involve HLA markers. Before the widespread use of GWAS, many HLA loci were discovered to be markers of drug-induced allergic reactions in candidate gene studies, which identified the HLA variants that led to risk of allergies efficiently7. Although the powerful, unbiased approaches of GWAS were not necessary for the discovery of HLA loci associated with drug-related allergies, GWAS provided some advantages: for example, single nucleotide polymorphisms (SNPs) in linkage disequilibrium with HLA variants were discovered6 and could be used as markers for the HLA risk variants, which are much more difficult to genotype directly.
However, the most significant PGx GWAS achievements for the prevention of ADRs have been related to the identification of non-HLA markers. Such examples include inosine triphosphatase (ITPA), a gene encoding an enzyme that is associated with protection against ribavirin-induced haemolytic anaemia8, CYP2C9, which is associated with phenytoin-induced SJS and TEN9, and nudix hydrolase 15 (NUDT15), which is associated with thiopurine-induced leucopenia10. Variants in these genes have sufficiently large effect sizes (odds ratios of >10) to offer clinical utility of diagnostics and provide new information on biological mechanisms that underlie ADRs. In the case of NUDT15, in vitro functional studies demonstrated that this hydrolase is involved in the metabolism and inactivation of thiopurine metabolites in lymphocytes and alters their cytotoxic effects. These findings have led to a new concept for intracellular pharmacokinetics of thiopurines, which can explain the susceptibility to thiopurine toxicity in patients who have loss-of-function variants in NUDT15.
Such studies also highlight the need to conduct research on multiple ethnic groups. For example, for several years, the most notable SNPs associated with thiopurine toxicities were thought to be located in the gene that encodes thiopurine methyltransferase (TPMT). However, whereas these SNPs have a large effect on individuals of European ancestry, SNPs in NUDT15 have been found to be critical for individuals of Asian descent. Clearly, more PGx GWAS that are focused on the identification of genes associated with drug toxicities need to be performed in all ethnic groups, in order to make better use of this powerful method for the discovery of novel toxicological mechanisms (Fig. 1; Supplementary information S1 (table)).
Challenges limiting PGx GWAS
PGx GWAS, particularly PGx GWAS focused on ADRs and in minority populations, lag far behind GWAS of common diseases. A probable explanation is that large sample sizes, which are needed to provide statistical power for meaningful associations in GWAS, are more difficult to obtain for PGx GWAS. In GWAS, a stringent threshold for statistical significance (that is, P < 5 × 10−8) is applied for examined variants; moreover, GWAS results must be replicated in independent cohorts.
Unfortunately, there are several factors that contribute to the challenge of obtaining large sample sets in PGx research. For the treatment of most common diseases, many drugs are available; therefore, the number of patients on a particular drug probably only represents a fraction of patients with that specific disease diagnosis. In addition, drug response measurements frequently need to include both baseline and on-treatment measurements to assess the effect of the drug, further limiting the number of patients that qualify for a PGx GWAS. Finally, replication studies with the same drug response measurements need to be conducted on patients who receive the same drug. Obtaining an adequate number of samples from patients who have experienced an ADR represents an even greater challenge for PGx GWAS, as such individuals probably represent a subset of patients on a particular drug. In the United States, these issues are amplified for minority populations, in which numbers are already lower than for populations of European ancestry.
Thus, there are substantial challenges for PGx GWAS to accrue sufficient numbers of samples, which include samples from appropriately characterized replication cohorts, minority populations and patients with ADRs.
Large consortia accelerate PGx GWAS
To obtain the necessary samples associated with drug response phenotypes, large collaborations, similar to those established for certain human diseases1, are crucial for the success of PGx GWAS. Indeed, international consortia of investigators focused on PGx GWAS of several drugs, such as warfarin, tamoxifen, selective serotonin reuptake inhibitors (SSRIs) and metformin, have been established and have contributed to the identification of the genetic factors that underlie variation in efficacy and toxicity among patients who receive these drugs (Table 1).
The largest subset of the pharmacological classes of PGx GWAS represents anticancer drug therapies (Fig. 1d), which accounts for approximately 25% of PGx studies reported in the GWAS Catalog to date. Such studies have been facilitated by a long-standing infrastructure for oncology clinical trials funded by the US National Cancer Institute (NCI), generally referred to as the 'cooperative groups'. Coordinated through the NCI's Cancer Therapy Evaluation Program (NCI-CTEP), the groups focus on phase III studies of FDA-approved and investigational agents in common malignancies, such as breast, colorectal, lung and prostate cancer.
Although the cooperative groups facilitate the collection of samples and phenotype information from patients on anticancer drugs, reliable genotyping, as well as expertise in pharmacology and statistics, is required to optimize these studies. To this end, a large international consortium named PGRN-RIKEN was established in 2008 and continues as an activity of the US National Institutes of Health (NIH) Pharmacogenomics Research Network (PGRN) Hub. To date, PGRN-RIKEN has conducted 37 PGx GWAS and has supported the largest number of PGx studies reported in the GWAS Catalog: 16 (7.4%) of 216 total PGx GWAS (see Fig. 1d).
During the twenty-first century, 'pharmacogenetics', which has evolved to become 'pharmacogenomics', has involved a migration from studies of candidate genes to the application of genome-wide research strategies, especially GWAS. The results of PGx GWAS have served to identify biomarkers for drug response and, in conjunction with functional genomic studies, have provided novel insights into both disease pathophysiology and molecular pharmacology. Notably, none of this progress would have been possible without collaboration across institutional and national boundaries. The success of PGRN-RIKEN and other collaborations makes a strong case for the creation and continuation of collaborative efforts.
Clearly, more PGx GWAS are needed, both for the identification of clinically actionable biomarkers for therapeutic and adverse drug responses and for the generation of novel mechanistic hypotheses.
- Massive schizophrenia genomics study offers new drug directions. Nat. Rev. Drug Discov. 13, 641–642 (2014).
- Clinical Pharmacogenetics Implementation Consortium guidelines for CYP2C19 genotype and clopidogrel therapy: 2013 update. Clin. Pharmacol. Ther. 94, 317–323 (2013). et al.
- Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines for IFNL3 (IL28B) genotype and PEG interferon-α-based regimens. Clin. Pharmacol. Ther. 95, 141–146 (2014). et al.
- Clinical Pharmacogenetics Implementation Consortium Guidelines for HLA-B genotype and abacavir dosing: 2014 update. Clin. Pharmacol. Ther. 95, 499–500 (2014). et al.
- Clinical Pharmacogenetics Implementation Consortium guidelines for HLA-B genotype and carbamazepine dosing. Clin. Pharmacol. Ther. 94, 324–328 (2013). et al.
- A whole-genome association study of major determinants for allopurinol-related Stevens–Johnson syndrome and toxic epidermal necrolysis in Japanese patients. Pharmacogenomics J. 13, 60–69 (2013). et al.
- Human leukocyte antigen-associated drug hypersensitivity. Curr. Opin. Immunol. 25, 81–89 (2013). , , , &
- ITPA gene variants protect against anaemia in patients treated for chronic hepatitis C. Nature 464, 405–408 (2010). et al.
- A common missense variant in NUDT15 confers susceptibility to thiopurine-induced leukopenia. Nat. Genet. 46, 1017–1020 (2014). et al.
- Genetic variants associated with phenytoin-related severe cutaneous adverse reactions. JAMA 312, 525–534 (2014). et al.
The authors acknowledge support from the following: U24 GM115370, R01 GM117163, R01 DK103729 (K.M.G. and S.W.Y.); R01 GM28157, U19 GM61388, U54 GM114838 (R.M.W.)