Published ahead of print on March 18, 2005, doi:10.1164/rccm.200412-1635OE
© 2005 American Thoracic Society doi: 10.1164/rccm.200412-1635OE
How Pharmacogenomics Will Play a Role in the Management of AsthmaPulmonary and Critical Care Division, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts Correspondence and requests for reprints should be addressed to Elliot Israel, M.D., Pulmonary and Critical Care Division, Brigham and Women's Hospital, 75 Francis Street, Boston, MA 02115. E-mail: eisraelasst{at}partners.org A patient with a diagnosis of asthma tells her physician, "I'm really concerned about side effects and expense. Will these medications be safe and effective for me?" The physician orders an asthma pharmacogenomic profile and a week later prescribes medications according to the results of the profile. Science fiction? With new developments in the field of pharmacogenomics, this scenario is a distinct possibility over the next decade. Pharmacogenomics is the study of the relationship between patterns of genetic variability in whole sets of genes in individuals and the variability in the response to pharmacotherapy among individuals. Although most of the data we review are strictly defined as pharmacogenetic in nature (i.e., relating to variability in a single gene rather than multiple genes), we use the broader term "pharmacogenomics" to encompass studies at the level of a single gene, as well as interactions between genes. We will discuss the following: variability in response to asthma pharmacotherapy; potentially heritable mechanisms that explain a portion of that variability; patterns of genomic variability that may explain some observed phenotypic variability; and finally, pharmacogenomic progress as it relates to asthma therapy and the potential clinical applicability of these findings. Our goal is to give the clinician a working understanding of the vocabulary of this field as it applies to asthma and an ability to interpret future studies. WHAT TYPE OF VARIABILITY DOES PHARMACOGENOMICS SEEK TO EXPLAIN? As many as two-thirds of patients with asthma may not attain full control of their asthma (1, 2). Up to one-third of patients treated with inhaled corticosteroids (ICSs) may not achieve objective improvements in airway function or indices of airway reactivity (3). Even more may not respond to leukotriene antagonists (4). One-third of patients using oral corticosteroids develop osteoporosis (5–8); in addition, 3 to 5% of patients using a 5-lipoxygenase (5-LO) inhibitor develop increases in liver function enzymes (9). Some patients with asthma develop cataracts and/or glaucoma with ICS use (10–12). A very small percentage of patients with asthma may be at risk of increased mortality with use of long-acting ß-agonists (13, 14). An estimated 70 to 80% of variability in individual responses to therapy may have a genetic basis (15). To the degree to which the salutary and/or adverse effects are heritable, pharmacogenomics seeks to define the relationship between the variability in our genetic code and the variability in our responses to pharmacologic interventions. WHAT HERITABLE MECHANISMS ALTER THE RESPONSE TO THERAPEUTIC DRUGS? Table 1 shows the major mechanisms under genomic control that can alter drug efficacy or toxicity. The first four, which can affect both efficacy and toxicity, relate to factors that influence the pharmacokinetic characteristics of a drug. The next three mechanisms relate to influences on pharmacodynamic characteristics. The last, unintended, targets may help explain genetic determinants of adverse effects.
HOW DO WE DETERMINE GENETIC VARIABILITY? Pharmacogenomic studies associate variations, or polymorphisms, in the genome with patterns of response to therapy. We will not comprehensively review the complexities of genetic polymorphisms here but will provide an overview and illustrations in Figures 1A and 1B. Table 2 is a glossary of genetic terminology important to the practitioner's understanding of polymorphisms and genetic association studies. Polymorphisms can occur in coding and noncoding areas of the gene. As discussed later, the functional significance of the mutation, in terms of its altering effect on function, may or may not be understood. In many cases, mutations may not be functional in and of themselves but may be markers for other changes in the genome with which they are in linkage disequilibrium.
Single nucleotide polymorphisms (SNPs) are most commonly used to explore the pharmacogenetics of asthma (16). They are common (occurring at least every 1,000 base pairs) and are more stable over time than other polymorphisms (17, 18). Variable numbers of tandem repeats, or microsatellites, are another type of mutation that may produce functional changes (see LEUKOTRIENE MODIFIERS later) or serve as a marker for other changes in the genome. The potential effect of varying combinations of SNPs and/or variable numbers of tandem repeats on a single chromosome (haplotypes) has also been the subject of pharmacogenomic association studies. HOW DO WE LOCATE POLYMORPHISMS TO ASSOCIATE WITH VARIABLE RESPONSES TO DRUGS? There are two different approaches. First, according to one's understanding of the mechanism of action or of factors affecting a drug's pharmacodynamics or pharmacokinetics, one may choose a set of candidate genes to examine and determine whether the polymorphisms under investigation affect function. Such polymorphisms can be proposed as candidates for pharmacogenomic study. This "heuristic" or concept-based functional approach has been used in initial studies of the response to ß-agonists and leukotriene modifiers, as reviewed later. Although more labor intensive, it allows for pharmacogenetic studies in smaller numbers of patients, thereby reducing the factors that can confound genetic associations (see following section). An alternative "nonheuristic" approach can be quite powerful and permit faster discovery of candidate polymorphisms and can be applied at the levels of both gene and genome. At the gene level, a gene or pathway of genes believed to be important in a drug response is "mined" for polymorphisms. Most often, the polymorphisms are of unknown functional significance; they may even be in noncoding regions of the gene or be synonymous SNPs. The discovered SNPs, and sometimes the imputed patterns of SNPs on a single chromosome (the haplotype), are then compared with the variability in therapeutic response. This nonheuristic approach can be applied beyond the gene or pathway of interest and used for genomewide screening to identify candidate genes. Such an approach resulted in the identification of the association between the metalloproteinase ADAM 33 and asthma (19). WHAT FACTORS CAN ALTER PUTATIVE PHARMACOGENOMIC ASSOCIATIONS? Even variations in genes that have large effects on a pharmacologic response are not likely to produce an entirely uniform response to a drug because of additional sources of variability that can confound association studies and make false associations occur. Conversely, these sources of variability can introduce enough "noise" to mask real associations. They can be classified into three major categories: simultaneous genetic interactions, host factors, and environmental effects. First, simultaneous genetic alterations may enhance or detract from an examined pharmacogenomic effect. Coincident genetic alterations in the same gene may produce additive or countervailing effects. These simultaneous polymorphisms can occur within the same gene or among different genes. For example, if polymorphisms of the ß-adrenergic receptor predict patients with diminished response to ß-agonists, such patients who simultaneously possess a beneficial polymorphism in the corticosteroid pathway (e.g., corticotropin-releasing hormone receptor 1 [CRHR1], see later) may show an intermediate, rather than poor, response to a combination of long-acting ß-agonists and ICSs. These interactions can vary between populations of different genetic backgrounds, so that a particular polymorphism may produce an effect when overlaid on the genetic background of one population but not another. For example, the Arg/Arg genotype (homozygosity for arginine) at position 16 of the ß-adrenergic receptor predicts the response to ß-agonists in Puerto Rican but not Mexican populations (20), as described later. Host factors, such as age, disease severity, concomitant drugs, and disease etiology, can affect responses. Factors such as disease severity may have their own genetic associations that are an additional genomic source of variability, as outlined previously. Last, environmental effects, which can also alter the response in question, need to be considered. For example, data suggest that smoking modulates the response to ICSs (21), and a recent study suggests that polymorphisms of the ß-adrenergic receptor are differentially associated with airway responsiveness in smoking versus nonsmoking populations (22). Furthermore, a particular environmental effect may preferentially occur in a particular genetic group—because of cultural differences, for example. Because these sources of variability may confound apparent pharmacogenomic associations, it is important to reconfirm apparent associations, especially those without clear functional correlates, in multiple populations. WHAT PROGRESS HAS BEEN MADE IN THE PHARMACOGENOMICS OF ASTHMA? Three agents of asthma pharmacotherapy have been subjected to pharmacogenomic analysis: ß-agonists, leukotriene modifiers, and corticosteroids.
ß-Agonists The approach to predicting responsiveness to ß-agonists stemmed from work of Green and colleagues (24, 28) and Liggett (26), suggesting that several SNPs within the coding block of the ß2-AR gene significantly alter receptor downregulation. Two SNPs, those at the 16th (B16) and 27th (B27) amino acid position of the receptor, had a significant minor allele frequency and thus became the focus of initial pharmacogenomic studies. An early study examining the effects of ß2-AR genotype on responsiveness to ß2-agonists in children found that 60% of patients with asthma homozygous for arginine at B16 (B16 Arg/Arg) had a positive response to albuterol (29), compared with only 13% in individuals homozygous for glycine at that position (p = 0.05). A subsequent small study by Lima and colleagues (30) found a similar association (30). However, neither a larger study in adults (31) nor a family-based study in children (32) found such an association. The latter actually suggested an association between bronchodilator responsiveness and a single SNP at the 3' end of the gene. However, a recent family-based study in Latinos did find an association between Arg/Arg and bronchodilator response in Puerto Ricans but, importantly, not in Mexicans (20). The data concerning receptor downregulation also led to a search for a genotype-specific effect related to chronic use of ß-agonists. An analysis in more than 250 patients with mild asthma randomized to regular albuterol use versus as-needed use showed no association at B27 but an association of B16 Arg/Arg with a decline in peak expiratory flow of 24 L/minute with regular use of albuterol compared with B16 Gly/Gly subjects (31). B16 Arg/Arg patients participating in a study in New Zealand had more asthma exacerbations during regular treatment with albuterol than during treatment with placebo (33). These studies in different populations of different ages suggest a true association between the polymorphisms under study and ß-agonist response. A recent prospective study of patients randomized to regular versus minimal albuterol use confirmed genotype-specific altered responses in B16 Arg/Arg patients using albuterol regularly (Figure 2) (34). In addition to a differential effect on peak flow (a 24-L/minute difference between the two genotypes, p = 0.0003), similar clinically significant patterns of genotype-specific effects on FEV1, symptoms, and use of supplementary reliever medication occurred in this prospective study.
The effect of simultaneous polymorphisms in the ß2-AR gene on the same chromosome (haplotype) has also been examined. A study of 13 SNPs in the ß2-AR gene and its 5' segments found that the patterns of association of the polymorphisms on the chromosome differed significantly among different ethnic populations (25). In this case, one of the three most common haplotypes (one that contained B16 Arg and accounted for more than three-quarters of the B16 Arg alleles) was associated with a decreased response to ß-agonists. In vitro, this haplotype resulted in both lower production of ß2-AR mRNA and lower ß2-AR receptor density compared with other haplotypes. Illustrating the difficulties of potential confounding, two family-based association studies found increased bronchodilator responsiveness to this haplotype (20, 32) and, as with Puerto Rican Latinos, increased responsiveness in B16-Arg/Arg patients (20). These latter two studies illustrate how differential clinical effects may occur when a particular polymorphism (or set of polymorphisms) is overlaid on different genetic substrates. Although many of the data suggest that arginine at the 16th amino acid position of the receptor is associated with a major, clinically significant pharmacogenomic effect, how these effects are modulated by other simultaneous polymorphisms in the gene and areas upstream and downstream of the coding region on the same chromosome (haplotypes) adds additional complexity to pharmacogenomic analysis. Because B16 Arg/Arg occurs in one-sixth of whites and up to one-fifth of African Americans, clinicians should consider potential adverse pharmacogenomic effects on patients using high doses of ß-agonists and controller medications who continue to have poorly controlled asthma, or who experience adverse effects when using increased amounts of ß-agonist. Furthermore, if the apparent effects associated with B16-Arg/Arg are replicated in populations using long-acting ß-agonists, current recommendations for asthma pharmacotherapy may be significantly altered. Indeed, investigation of the effects of these polymorphisms on the response to long-acting ß-agonists is currently underway.
Leukotriene Modifiers
5-LO promoter polymorphisms. Thus, the absence of at least one copy of the wild-type allele creates a phenotype that is less responsive to leukotriene modifiers. However, this pattern of mutation only occurs in approximately 5% of patients with asthma, accounting for a small proportion of the variability in response to leukotriene modifier therapy.
LTC4 synthase polymorphisms. In summary, several candidate polymorphisms in genes related to the activity of leukotriene modifiers have been identified. Although the 5-LO polymorphism is infrequent, it seems to identify a group less likely to benefit from therapy with these agents. The differential response based on LTC4 synthase polymorphisms suggests that this locus, too, helps determine the response to this asthma therapy. Because variant LTC4 synthase genotypes are prevalent in patients with both aspirin-intolerant (76%) and aspirin-tolerant asthma (42–44%) (42), if the effects of this polymorphism are confirmed, its high prevalence may make it a useful predictor of response to this class of agents.
ICSs
The consistent association of different SNPs in the same gene with changes in lung function suggests that the actual causal variant in CRHR1 remains to be discovered but that the three variants studied are imperfectly correlated markers in linkage disequilibrium with a causal polymorphism. Nonetheless, recent data in knockout mice suggest that CRHR1 is involved in modulating endogenous glucocorticoid production and enhancing allergen-induced airway inflammation and lung mechanical dysfunction (44). These findings may explain the association with a differential response to ICSs but it is too early to tell whether the CRHR1 polymorphisms will be useful clinical predictors of response to ICSs. Interestingly, Tantisira and colleagues (45) recently reported on a functional variant in another gene that may be able to predict responsiveness to ICSs. They demonstrated that polymorphisms of TBX21, the gene coding for transcription factor T-bet (T-box expressed in T cells), are associated with significant improvement in methacholine responsiveness in children with asthma. In a large clinical trial spanning 4 years, inhaled-steroid users harboring a specific TBX21 polymorphism demonstrated a significant improvement in responsiveness to nonasthmatic levels. However, the minor allele frequency for this mutation (labeled H33Q) was only 4.5%, and there were no homozygotes observed in more than 500 subjects, suggesting that although the effect of such a mutation may be large, it may only affect a small number of individuals. WHAT ARE THE IMPLICATIONS OF PHARMACOGENOMIC FINDINGS FOR ASTHMA THERAPY? The previously described genetic associations with both therapeutic and detrimental responses to each of these classes of commonly used asthma medications constitute important steps in the development of individualized therapy for asthma. Although the effects noted at position 16 of the ß-adrenergic receptor may be of sufficient magnitude to affect therapy, most pharmacogenomic effects will probably be of smaller magnitude or, as with 5-LO polymorphisms, less common. Thus, we will most likely use "panels" of polymorphisms to calculate the relative risk–benefit ratio of a particular therapeutic course for an individual patient. Pharmacogenomic information may allow us to treat those who can benefit most from particular asthma medications and to avoid toxicity by administering medications to those unlikely to experience toxicity. For example, if pharmacogenomics fulfills its promise, we will be able to administer corticosteroids to those least likely to experience adverse effects. Furthermore, we will be able to introduce and/or develop drugs for asthma that were held back because of potential toxicity in a subset of patients. From a clinician's point of view, it is expected that pharmacogenomic assays will be readily available in clinical laboratories within the next 5 years. Considering the rapid fall in the cost of genotyping at multiple loci simultaneously, it is unlikely that the technology will limit the introduction of this methodology; rather, the design and execution of clinical trials in multiple populations will be the rate-limiting step. Thus, we advocate obtaining genetic material in all clinical asthma trials and consideration of prospective genotype-stratified clinical trials. Such association studies and biologically informative pharmacogenomic trials over the next decade should allow us not only to "Do no harm" but also to "Do much good." Acknowledgments The authors thank Dr. Benjamin A. Raby of the Channing Laboratory, Brigham and Women's Hospital, and Harvard Medical School for the development of illustrations for this article. FOOTNOTES Conflict of Interest Statement: M.E.W. has performed consultancies and lectures sponsored by Merck, GlaxoSmithKline (GSK) and Novartis, totaling less than $5,000/year in 2003–2005, and has received an unrestricted educational research grant from Merck for $25,000/year in 2003–2005; E.I. has served as a consultant and/or on advisory boards for Schering-Plough and Merck, and he received lecture fees from Merck and is receiving or has received grant support for clinical trials from GSK, AstraZeneca, Boehringer Ingelheim, and Wyeth. Received in original form December 6, 2004; accepted in final form March 16, 2005 REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
p = 0.0209). Reprinted by permission from Reference 34.
