Pharmacogenetics of Bronchodilator Response: Future Directions

Abstract

Purpose of Review

Several genome-wide association studies (GWASs) of bronchodilator response (BDR) to albuterol have been published over the past decade. This review describes current knowledge gaps, including pharmacogenetic studies of albuterol response in minority populations, effect modification of pharmacogenetic associations by age, and relevance of BDR phenotype characterization to pharmacogenetic findings. New approaches, such as leveraging additional “omics” data to focus pharmacogenetic interrogation, as well as developing polygenic risk scores in asthma treatment responses, are also discussed.

Recent Findings

Recent pharmacogenetic studies of albuterol response in minority populations have identified genetic polymorphisms in loci (DNAH5, NFKB1, PLCB1, ADAMTS3, COX18, and PRKG1), that are associated with BDR. Additional studies are needed to replicate these findings. Modification of the pharmacogenetic associations for SPATS2L and ASB3 polymorphisms by age has also been published. Evidence from metabolomic and epigenomic studies of BDR may point to new pharmacogenetic targets. Lastly, a polygenic risk score for response to albuterol has been developed but requires validation in additional cohorts.

Summary

In order to expand our knowledge of pharmacogenetics of BDR, additional studies in minority populations are needed. Consideration of effect modification by age and leverage of other “omics” data beyond genomics may also help uncover novel pharmacogenetic loci for use in precision medicine for asthma treatment.

Introduction

Asthma affects over 300 million people worldwide [1] and is characterized by wheeze, chronic airway inflammation, and reversible airflow obstruction. Asthma is also the most common chronic disease in children globally [2]. Individuals with asthma often require pharmacological treatment throughout the course of their illness. Short-acting beta-agonists (SABAs) are prescribed as a rescue medication for asthma, to be utilized as a bronchodilator to treat acute bronchospasm. SABAs, such as albuterol, act on beta-2 adrenergic receptors to relax bronchial smooth muscle [3]. However, like many of the pharmacotherapeutic options for asthma, SABA medications are not universally effective for all patients [4•], and some individuals are unable to derive much benefit from these treatments.

Part of an individual’s response to albuterol is determined by genetics. Bronchodilator response (BDR) measures the change in bronchoconstriction before and after the administration of a short-acting β₂-agonist, such as albuterol, which reduces airway constriction by stimulating the β₂-adrenergic receptors (β₂ARs) on airway smooth muscle. In fact, it is estimated that 31–92% of BDR is heritable [5-8]. This review article will equate BDR with response to albuterol.

Genome-wide association studies (GWASs) and candidate gene studies have identified genetic polymorphisms associated with BDR. These findings have been reviewed recently elsewhere [9-13] and include genetic associations within the ADRB2, ARG1, ARG2, THRB, SPATS2L, ASB3, SOCS, ADCY9, CRHR2, COL22A1, and CLOCK genes. ADRB2 and ARG1 are two of the most studied genes for BDR [12]. ADRB2 is a gene that encodes for the β₂ adrenergic receptor, which is involved in bronchodilator response and is the target of β₂ adrenergic agonists. Polymorphisms in ADRB2 have been extensively studied in multiple populations, although findings have been variable (some cohorts detect no associations with BDR, while others report associations with BDR and/or asthma exacerbations) [12]. ARG1 polymorphisms have been linked to lower BDR in several studies [12]. Overall, the total variability in BDR accounted for by all genetic polymorphisms across all associated loci is quite low. A further limitation of these genetic association studies is that they are mainly conducted in Caucasian populations, which means the findings may not be applicable across multiple racial/ethnic populations.

Pharmacogenetics of BDR in Minority Populations

Minority populations, in particular Blacks and Puerto Ricans, have the highest asthma morbidity as well as the lowest BDR [14, 15]. Genetic associations in minority populations often do not replicate in Caucasian populations, suggesting that genetic determinants of albuterol response may be partly population-specific [16•, 17, 18•] As reviewed by Jerome et al. in 2021 [19•], there have been few studies of BDR among Black cohorts. In a GWAS performed by Spear and colleagues, an intergenic SNP on chromosome 9q21 was associated with BDR among Black participants [16•]. After meta-analyzing data from both their Black and Latino cohorts, Spear and colleagues identified three polymorphisms in the intronic region of PRKG1 that were associated with BDR. However, these findings failed to replicate in additional Black and Latino populations. In a multi-ethnic meta-analysis by Finkelstein et al. in 2009, the beneficial effects of rs1042713 (A > G) in the ADRB2 gene on BDR were most pronounced in Black participants (as compared to whites) [17]. In 2018, Mak et al. conducted a whole-genome sequencing study in Black participants from the SAGE II cohort [18•]. While no genome-wide significant associations were observed within Black participants alone, a trans-ethnic meta-analysis of Black, Puerto Rican, and Mexican American participants (from SAGE II and GALA II cohorts) did identify 27 SNPs in 10 loci associated with BDR. Polymorphisms in these genetic loci explained 23%, 16%, and 18% of the variation in bronchodilator responder status in Puerto Rican, Mexican American, and Black participants, respectively [18•], In contrast to some of the genetic loci identified in Caucasian participants, these loci were within genes associated with lung capacity (DNAH5) immunity (NFKB1 and PLCB1). Similar to some of the findings in Caucasians, two of the loci identified in the Mak et al. multi-ethnic meta-analysis were in the β-adrenergic signaling pathway (ADAMTS3 and COX18). It is important to note that the investigators were not able to replicate these associations, a limitation which may stem from a lack of available minority populations with a sufficient sample size for validation.

Impact of Bronchodilator Response Phenotype Classification

In addition to racial differences across cohorts, another critical aspect to consider when examining genetic associations for response to albuterol is the distribution and classification of the BDR outcome variable. The majority of GWAS studies of BDR have modeled percent change in FEV₁ following administration of albuterol as a continuous outcome. However, some have utilized extremes of the BDR distribution in their study to classify “responders” and “non-responders” to albuterol. Differences in genetic association findings arise when different decisions are made about how to model the outcome variable. One such example is the twin set of BDR studies conducted in the same group of cohorts (GALA and SAGE) by Spear et al. [16•] and Mak et al. [18•]. In Spear et al., BDR was modeled as a continuous outcome variable, while in Mak et al., the extremes of the BDR distribution were used to classify “responders” and “non-responders” to albuterol. While the former study identified variants in PRKG1 as top hits, the latter study, based instead on a binary rather than a continuous outcome measure, identified a distinct set of SNPs as potential BDR response candidates. Both studies provide useful information about potential predictors of BDR. However, it may be difficult to translate the results of such findings to the commonly accepted “clinical threshold” of BDR, where response to albuterol is defined not as a continuous outcome or an extreme phenotype, but as whether an individual responds at or above a threshold of a 12% difference in FEV₁ after administration of albuterol [20-22].

Furthermore, there may even be subtleties in characterization of the continuous BDR phenotype when researchers consider changes in FEV₁ post albuterol as change in the % initial value vs. % of the predicted value [23]. BDR expressed as a change in percent predicted FEV₁ is adjusted for age and height, so that comparisons across individuals of various ages and sizes can easily be made. However, the % initial value may be most practical in countries where the lack of spirometric norms and inaccessibility of prediction software make calculation of age, sex, and height-adjusted percent predicted values difficult. Of course, an important caveat is that the easier to compute % initial has issues with bias with respect to sex and size [24].

These issues in BDR phenotype characterization should all be considered when comparing findings from across pharmacogenomic studies of response to albuterol. If possible, sensitivity analyses with multiple phenotypes should be explored to determine if associations are robust to choice of phenotype variable, or specific to a particular derivation of BDR phenotype.

Effect Modification of Pharmacogenetic Responses by Age

Variability in treatment response in children is an understudied area, and elucidating age-dependent treatment effects is important. Childhood is a unique time period with rapid growth and hormonal development. Extrapolating adult asthma pharmacogenetic studies to children does not adequately address the unique needs of child asthma sufferers. Statistical models of pharmacogenetic responses typically include the main effects of genotypes, with standard adjustment for covariates. Some genetic epidemiology studies of asthma stratify populations by age group prior to analysis. For example, in GWASs of asthma, populations are often stratified by childhood onset vs. adult onset asthma [25•]. In pharmacogenetic research, age is rarely considered an effect modifier of genetic responses [26]. Aging is correlated with changes in immune, hormonal and inflammatory responses [27, 28], which may alter the influence of genetic polymorphisms on response to asthma medications. As one example, ARG1 is a gene known to be associated with BDR in children and adults, but the effect is greater in adults than in children [29]. Furthermore, ADRB2, which works through β-adrenergic pathway, is one of the most studied genes for BDR and has more pronounced effects for children with asthma than for adults [30-33].

Our group has conducted several studies of age-related modifications of pharmacogenetics of asthma treatment responses [34-36], including studies of age by genotype interactions on BDR [37, 38]. In our analysis of repeated continuous BDR outcome measures in the CAMP clinical trial, we observed that polymorphisms in SPATS2L and ASB3 were most predictive of BDR in children prior to adolescence [38]. These same polymorphisms were no longer predictive of BDR after age 15 years. In another multi-cohort study, we observed effect modification by age for genetic polymorphisms in the genes PRAG1 and ADRB2 [37]. Both of these studies suggest that age should be considered an effect modifier in pharmacogenetic research, to enhance our understanding of how genetic effects vary with age, and also to aid in discovery of genetic variants that would not be identified without explicitly accounting for age-related heterogeneity. It should be noted, that Vin order for effective medication to be studied adequately, even larger sample sizes are needed, as an interaction effect is now the primary measure of interest. Therefore, either increase in the sample size of multiple cohorts throughout various important age ranges is necessary before significant scientific understanding in this area will occur.

Leveraging “Omics” Data to Identify Pharmacogenetic Targets

Results from traditional GWAS approaches in pharmacogenetics, where millions of individual polymorphisms are considered “agnostically” without regard to functional importance, may be insufficient to predict an individual’s BDR. While larger GWAS studies based on whole-genome sequencing data may enhance our knowledge of pharmacogenetic responses somewhat, they are unlikely to completely capture the full spectrum of genetic effects. A “bottom-up” approach that leverages associations of BDR with other “omics” data may help focus pharmacogenetic studies on the most relevant genetic predictors. Metabolomics and epigenomics are two types of “omics” data that can be leveraged in this context. Modifications of the metabolome and epigenome are more proximal physiologically to asthma medication response and, as intermediary biomarkers, may highlight the most relevant pathways for consideration in pharmacogenetic testing.

Metabolomics studies of albuterol response have been conducted in airway and serum samples. In a study by Gaugg et al., metabolomics analysis of exhaled breath condensate (EBC) collected 10–30 min after albuterol administration revealed 131 mass spectral features that were altered as compared to placebo [39]. The changes in these features were strongly correlated within a given chemical class, suggesting that changes in biochemical processes in response to albuterol can be monitored using EBC. Fatty acids associated with lipolysis (acetic, propionic, and butyric acid) were increased following albuterol administration. Similarly, serum metabolomic profiling has also shown alterations in fatty acid levels directly following albuterol administration [40]. Metabolomic profiles may be influenced by administration of a drug, but that does not necessarily mean that the altered metabolic pathways play a role in the drug’s effectiveness. Future studies that examine whether metabolomics varies by BDR profile may help pinpoint specific metabolic pathways (and the genes that encode for them) for consideration in pharmacogenetics studies. We conducted one such study in the Childhood Asthma Management Program (CAMP) and Genetics of Asthma in Costa Rica Study (GACRS) populations [35]. Findings from our group show that response to albuterol is associated with a butyrate metabolite (2-hydroxyglutarate) in these two independent cohorts, further supporting the role of fatty acid metabolism as a potential pathway to consider for potential genetic influences of BDR using a pharmacogenetic framework.

Epigenomics is another type of “omics” that may prove useful for pharmacogenetic prediction [41]. The idea of epigenomics for precision medicine has been hypothesized previously [42]. Epigenetic signals, particularly those in airway cells, may highlight the most relevant genes involved in asthma treatment response for pharmacogenetic studies, or may in and of themselves serve as a predictive biomarker of drug response. The nasal methylome demonstrates a distinct DNA methylation pattern for BDR that is entirely independent of the altered DNA methylation signals observed for allergy, fraction exhaled nitric oxide (FeNO), or asthma status. In our 2019 Nature Communications study [43], we observed differential DNA methylation of 130 CpG sites in a cohort of over 500 adolescents with BDR data. Some of the top differentially methylated CpGs were in LGALS8 (gene for an integrin-like protein involved in cell–cell adhesion and apoptosis), STIM1 (a mediator of Ca + 2 influx channels), ACTR3 (actin-related protein 3), and KCNJ4 (a potassium ion conductance gene). Nasal samples were collected prior to assessing BDR. Therefore, these DNA methylation marks are predictive of response to albuterol, as opposed to being affected by administration of the drug. Additional studies are needed to determine whether airway epigenomics will be valuable as a tool for guiding our interrogation of the genome, or whether some combination of genetic polymorphisms and epigenetic marks may contribute to precision medicine for BDR in asthma.

Polygenic Risk Scores in Pharmacogenetics

Polygenic risk scores (PRSs) provide the opportunity to combine small effects of individual genetic polymorphisms into an overall genetic risk score for prediction of a phenotype. The ability to predict individual responses to asthma medications through PRSs has great potential to guide treatment and may also provide information for sub-phenotyping of asthma. PRSs have been used successfully to predict respiratory disease phenotypes, including COPD [44]. This same concept can be applied to pharmacogenetic responses as well. Pharmacogenetic polygenic risk scores have been developed for drug responses in diabetes and cardiovascular disease [45-47]. Our group recently published a proof of concept study for a pharmacogenetic PRS for BDR [48] that prioritized GWAS variants with high combined annotation dependent depletion (CADD) scores for functional relevance. For PRS computation, we used results from a GWAS of BDR and ranked the top GWAS polymorphisms by CADD scores. CADD scores integrate multiple annotations into one score, in a way that prioritizes functional, deleterious, and disease causal variants across a wide range of functional categories. We observed a 0.63% decrease in BDR in response to albuterol for a standard deviation increase in the PRS. Additional studies in larger and more diverse populations are required to validate this PRS for BDR and to further develop a PRS with higher accuracy, so that it may be used clinically for albuterol response prediction.

Conclusion

In order to expand our knowledge of BDR pharmacogenetics, larger studies with higher proportions of minority participants are needed to identify population-specific as well as transethnic loci. Replication of initial findings in minority populations has been challenging, due to the limited number of pharmacogenetics studies in these populations. Sensitivity analyses by BDR phenotype characterization may also be helpful, to determine whether genotypes are linked to incremental changes in BDR along with a continuous distribution and/or whether they are predictive of BDR above a certain threshold. Pharmacogenetic studies that incorporate effect modification by age, as well as those that leverage additional “omics” data types to identify the most relevant genetic pathways, may also help uncover novel pharmacogenetic loci for use in precision medicine. Developing pharmacogenetic PRSs also may have significant clinical value.