Genomewide association between GLCCI1 and response to glucocorticoid therapy in asthma

Abstract

Background: The response to treatment for asthma is characterized by wide interindividual variability, with a significant number of patients who have no response. We hypothesized that a genomewide association study would reveal novel pharmacogenetic determinants of the response to inhaled glucocorticoids.

Methods: We analyzed a small number of statistically powerful variants selected on the basis of a family-based screening algorithm from among 534,290 single-nucleotide polymorphisms (SNPs) to determine changes in lung function in response to inhaled glucocorticoids. A significant, replicated association was found, and we characterized its functional effects.

Results: We identified a significant pharmacogenetic association at SNP rs37972, replicated in four independent populations totaling 935 persons (P=0.0007), which maps to the glucocorticoid-induced transcript 1 gene (GLCCI1) and is in complete linkage disequilibrium (i.e., perfectly correlated) with rs37973. Both rs37972 and rs37973 are associated with decrements in GLCCI1 expression. In isolated cell systems, the rs37973 variant is associated with significantly decreased luciferase reporter activity. Pooled data from treatment trials indicate reduced lung function in response to inhaled glucocorticoids in subjects with the variant allele (P=0.0007 for pooled data). Overall, the mean (±SE) increase in forced expiratory volume in 1 second in the treated subjects who were homozygous for the mutant rs37973 allele was only about one third of that seen in similarly treated subjects who were homozygous for the wild-type allele (3.2±1.6% vs. 9.4±1.1%), and their risk of a poor response was significantly higher (odds ratio, 2.36; 95% confidence interval, 1.27 to 4.41), with genotype accounting for about 6.6% of overall inhaled glucocorticoid response variability.

Conclusions: A functional GLCCI1 variant is associated with substantial decrements in the response to inhaled glucocorticoids in patients with asthma.

Trial registration: ClinicalTrials.gov NCT00000575.

Figure 1. Study Design

The family-based association test (FBAT) screening algorithm was applied to genomewide genotyping data in parent–child trios corresponding to probands randomly assigned to treatment with inhaled glucocorticoids (ICS) in the Childhood Asthma Management Program (CAMP), in order to identify the top 100 single-nucleotide polymorphisms (SNPs) in terms of power for replication (heritability) in association with changes in forced expiratory volume in 1 second (FEV₁). These SNPs were tested for association in the CAMP population, and 13 variants were genotyped in three independent replication populations (in the Salmeterol or Corticosteroids [SOCS] and Salmeterol ± Inhaled Corticosteroids [SLIC] trials, the Adult Study, and the Leukotriene Modifier or Corticosteroid or Corticosteroid–Salmeterol [LOCCS] trial). One variant, rs37972, was associated in each population. The potential for function of this variant and a closely correlated variant, rs37973, was assessed, and rs37973 was confirmed to alter expression of the glucocorticoid-induced transcript 1 gene (GLCCI1); rs37973 was then tested for overall association with the response to ICS therapy in four independent clinical-trial populations. CARE denotes Childhood Asthma Research and Education, and EMSA electrophoretic mobility-shift assay.

Figure 2. Changes in Lung Function with Therapy According to rs37972 Genotype

The association of genotypes from GLCCI1 rs37972 with changes in lung function is shown as the mean (± SE) change in forced expiratory volume in 1 second (FEV₁), expressed as the percent of the predicted value, after 4 to 8 weeks of therapy with inhaled glucocorticoids in four replication populations: participants in the run-in periods of the SOCS and SLIC trials (SS) (264 participants), the Adult Study (385 participants), the LOCCS trial (185 participants), and the CARE Network trials (101 participants). For each population, the additive model showed a significant association, with poorer responses being associated with the mutant (T) allele (Liptak combined P = 0.0007). The rs37972 minor-allele frequencies for the SS, Adult Study, LOCCS, and CARE populations were 0.40, 0.38, 0.38, and 0.37, respectively. Since rs37972 correlates perfectly (is in complete linkage disequilibrium) with rs37973, the within-trial clinical response was identical for the latter variant (data not shown). CC denotes homozygous wild-type allele, CT heterozygous wild-type and mutant alleles, and TT homozygous mutant allele.

Figure 3. Functional Analysis of GLCCI1 Allelic Variation

Panel A shows a graphical overview of the rs37972 and rs37973 single-nucleotide polymorphisms in relation to the first exon structure of GLCCI1. The position of the first methionine (Met), the start codon of GLCCI1, is shown. Indicated DNA fragments were cloned into luciferase vectors for the enhancer assay. Panel B shows the allelic differences in relative luciferase activity in Jurkat cells, Raji cells, and human acute monocytic leukemia (THP-1) cells, obtained with the use of a pGL4.23–luciferase vector. Data represent mean (± SD) values from one experiment performed in triplicate and are expressed relative to the luciferase activity of the mock transfectant, which was arbitrarily set at 1. P values were calculated with the use of Student’s t-test. Similar results were obtained in three independent experiments. Panel C shows results of an electrophoretic mobility-shift assay (EMSA) of rs37973 in Jurkat, Raji, and THP-1 cells. The intensity of each band on EMSA was analyzed by densitometry, and the intensity ratio of each G-allele band to each A-allele band is shown. Two independent experiments were performed with similar results. The genomic region containing the A allele of rs37973 was predicted to bind two transcription factors, CCAAT/enhancer binding protein (C/EBP) alpha and zinc finger E-box binding homeobox 1 (ZEB1). Although the signal intensity of the DNA–protein complex derived from the A allele was higher than that from the G allele in the presence of nuclear extract of each of the cell lines on EMSA, we could not confirm the binding of C/EBP alpha and ZEB1 to the sequence by supershift assay (data not shown). It is possible that other, unidentified transcription factors might bind to this sequence.