The treatment of asthma remains a formidable task for clinicians, despite the development of new treatment options such as long acting β agonists, potent inhaled corticosteroids, and leukotriene receptor antagonists. The complexities in treatment arise not only from temporal changes in the clinical status of patients (such as viral infections and allergy exposures), but also from the well recognised variability between individuals in the response to treatment. This variability is found with the most widely used treatment for asthma—the β agonists—which act to dilate constricted airways by binding to the β₂ adrenoceptor on airway smooth muscle. It has been suggested that up to 50% of the variability in response to β agonists between individuals has a genetic basis.¹ Beta agonists evoke smooth muscle relaxation in airways that have become constricted from virtually any spasmogen, so it is no wonder they have a central role in treating asthma.

However, several aspects of β₂ adrenoceptor signalling have been lurking in the shadows for many years. Firstly, in virtually every cell type that has been studied, chronic exposure (∼8 hours or more) to β agonists results in a loss of receptor expression with accompanying desensitisation of the cellular response.² At the physiological level, desensitisation could be manifested as tachyphylaxis. Clinically, tachyphylaxis to β agonists could lead to a variety of outcomes such as progressive loss of the bronchodilating effect, a loss of the protective effect against bronchoconstriction, and increased exacerbations. It has also been recognised for over 25 years that corticosteroids increase β₂ adrenoceptor expression,³ which could mitigate against agonist promoted downregulation.⁴ The clinical response to corticosteroids in asthma may therefore involve both anti-inflammatory effects which reduce local spasmogens, as well as improvement in β₂ adrenoceptor signalling.

As these molecular studies were underway, epidemiological^5–7 and clinical^8,9,10 studies were beginning to reveal potential adverse effects of β agonists, particularly with regular usage compared with an as-needed schedule. One of the most recently published studies involved approximately 26 000 asthmatics treated with salmeterol or placebo (the SMART study).¹¹ Small but statistically significant increases in several secondary mortality related end points were noted in the salmeterol treatment group. This may have been due to an excess in events in African Americans treated with salmeterol.

The genomic era had its impact on β agonist pharmacology, beginning in 1993 when we defined three non-synonymous polymorphisms of the β₂ adrenoceptor coding region.¹² Subsequent studies in transfected cells¹³ and endogenously expressing airway smooth muscle cells¹⁴ indicated that the two common polymorphisms (at amino acid positions 16 and 27) have an impact on agonist promoted downregulation of receptor number. Furthermore, the variation at position 16 (where either Arg or Gly can be found) appeared to dominate the downregulation effect. What was not clear at that time was how these in vitro findings would be manifested in patients. A reasonable hypothesis was that the position 16 polymorphisms influence tachyphylaxis to chronic β agonists, and this has been explored using a number of different approaches. For example, Taylor et al¹⁵ retrospectively examined asthma exacerbations and peak expiratory flow rates (PEFR) in a three-way crossover study of 157 patients receiving regularly scheduled placebo, salbutamol (albuterol), or salmeterol and potential relationships to Arg or Gly16. In those treated with salbutamol, asthma exacerbations were greater in Arg16 homozygous patients than in Gly16 homozygotes, and Arg16 homozygotes treated with salbutamol had a higher exacerbation rate than Arg16 homozygotes treated with placebo. There was a strong trend towards lower PEFR in Arg16 homozygotes, but this did not reach statistical significance. The exacerbation rates during salmeterol treatment were low for all genotypes and did not differ between genotypes.

At about the same time, Israel et al¹⁶ published results from a retrospective analysis of a trial of as-needed versus regularly scheduled salbutamol in 190 mild asthmatics. The primary finding centred around changes in the PEFR. In the Arg16 group receiving regular salbutamol, a decrease of 30.5 l/min in morning PEFR was observed. In contrast, neither Arg16 patients on as-needed salbutamol nor Gly16 patients receiving regularly scheduled salbutamol had a decrease in PEFR. Regular salbutamol treatment in Arg16 homozygotes was associated with a decrease of 23.8 l/min in morning PEFR compared with Gly homozygotes. A somewhat similar association was found in a subsequent prospective trial with regularly scheduled salbutamol¹⁷ with a genotype attributable treatment difference of 24 l/min. Although retrospective and with a rather small number of subjects, another recent analysis has shown little benefit in patients with the Arg16 genotype during salmeterol treatment, either with or without concomitant inhaled corticosteroids.¹⁸ Taken together, these studies point to a segment of the population—those homozygous for Arg16—who do not appear to benefit and/or may deteriorate with chronic β agonists.

It is worth noting that the aforementioned studies had relatively strict enrolment criteria and, as such, there was some degree of homogeneity in the patient populations within each study. This is a typical study design in that investigators attempted to minimise the noise from non-genetic heterogeneity. In the current issue of Thorax, Palmer et al¹⁹ have explored this important concept in a different manner. These investigators considered a more “real life” situation, using a cross section of children and young adults with asthma (546 patients) from asthma clinics in Tayside, Scotland. Of these, 164 were receiving regular salmeterol with an inhaled corticosteroid and 351 were receiving as-needed salbutamol. These two groups represented approximately 95% of the β agonist treatment regimens. When all patients were considered, the odds ratio associated with Arg16 and asthma exacerbations was 2.05 (p = 0.010) but, in salmeterol treated patients, the odds ratio associated with Arg16 homozygosity and exacerbations was 3.40 (p = 0.022). In contrast, in patients not being treated with salmeterol (which essentially represents those on as-needed salbutamol), the odds ratio for exacerbations associated with Arg16 was not statistically significant (regardless of whether homozygous status or a co-dominant model was used).

These results are therefore consistent with the previously cited studies, all pointing to Arg16 polymorphism as a genetic marker for a poor response to chronic β agonist treatment in asthma. In addition, like the aforementioned study,¹⁸ this phenotype is present during inhaled corticosteroid treatment. The potential ramifications of this concept are significant in that the frequency of homozygous Arg16 in the population ranges from ∼15% in white ethnic groups to ∼35% in those of Asian descent. What is not really clear from any of these studies is what percentage of the variability in a phenotype can be attributed to the position 16 polymorphism. If it is only a small fraction (10% or less), one has to wonder whether, alone, knowing the genotype at position 16 has any clinical usefulness. Similarly, what is the sensitivity and specificity of the test? The predictive value of genetic tests can potentially be increased if the gene has more than one polymorphism and only certain combinations of these polymorphisms (haplotypes) are responsible for a given cellular response. Genotyping at these multiple sites can therefore identify certain combinations more precisely. On the other hand, if a single polymorphism is the only variant that is controlling the phenotype, then partitioning the population into further subgroups will not improve the predictive power. For the intronless β₂ adrenoceptor gene, there are a number of polymorphic sites in the 5′ flanking region (up to nucleic acid −1023) which are found in 12 distinct haplotypes²⁰ (there may be additional polymorphic sites further upstream or in the 3′ untranslated region of the gene). Of the 12 haplotypes, five have frequencies of ∼5% or more in one or more ethnic groups. As can be seen in table 1, if an individual has an adenine (A, which encodes for arginine) at nucleic acid position 46 (amino acid 16), their haplotype could be #1 or #4. This is particularly so for those of African descent in whom haplotype #1 has a frequency of 25%. Furthermore, an individual with a guanine (G, which encodes for glycine) at position 46 could have haplotype #2 or #6. Also note the approximately sevenfold difference in the frequency of haplotype #2 between white ethnic groups and African Americans, and the ∼5% frequency of haplotype #9 which is exclusively found in African Americans. Within the β₂ adrenoceptor gene there is therefore substantial genetic variability outside the coding region. Adding these additional polymorphisms to the coding polymorphisms or the use of haplotypes may improve the predictive power. In addition, the β₂ adrenoceptor pathway has a number of signalling elements such as G_s, adenylyl cyclase, G-protein coupled receptor kinases, and protein kinase A. Genetic variability in any of these elements could also contribute to β agonist phenotypes.²¹

It is quite likely that a set of variants from multiple genes acts to set β agonist responsiveness. Ultimately, then, a “score card” may be developed which uses genetic variants from multiple genes and has substantial predictive power such that treatment can be individualised for asthma patients.