Childhood Asthma Management Program Study Research Articles

Discovering Pediatric Asthma Phenotypes on the Basis of Response to Controller Medication Using Machine Learning

Pediatric asthma has variable underlying inflammation and symptom control. Approaches to addressing this heterogeneity, such as clustering methods to find phenotypes and predict outcomes, have been investigated. However, clustering based on the relationship between treatment and clinical outcome has not been performed, and machine learning approaches for long-term outcome prediction in pediatric asthma have not been studied in depth. Our objectives were to use our novel machine learning algorithm, predictor pursuit (PP), to discover pediatric asthma phenotypes on the basis of asthma control in response to controller medications, to predict longitudinal asthma control among children with asthma, and to identify features associated with asthma control within each discovered pediatric phenotype. We applied PP to the Childhood Asthma Management Program study data (n = 1,019) to discover phenotypes on the basis of asthma control between assigned controller therapy groups (budesonide vs. nedocromil). We confirmed PP's ability to discover phenotypes using the Asthma Clinical Research Network/Childhood Asthma Research and Education network data. We next predicted children's asthma control over time and compared PP's performance with that of traditional prediction methods. Last, we identified clinical features most correlated with asthma control in the discovered phenotypes. Four phenotypes were discovered in both datasets: allergic not obese (A+/O-), obese not allergic (A-/O+), allergic and obese (A+/O+), and not allergic not obese (A-/O-). Of the children with well-controlled asthma in the Childhood Asthma Management Program dataset, we found more nonobese children treated with budesonide than with nedocromil (P = 0.015) and more obese children treated with nedocromil than with budesonide (P = 0.008). Within the obese group, more A+/O+ children's asthma was well controlled with nedocromil than with budesonide (P = 0.022) or with placebo (P = 0.011). The PP algorithm performed significantly better (P < 0.001) than traditional machine learning algorithms for both short- and long-term asthma control prediction. Asthma control and bronchodilator response were the features most predictive of short-term asthma control, regardless of type of controller medication or phenotype. Bronchodilator response and serum eosinophils were the most predictive features of asthma control, regardless of type of controller medication or phenotype. Advanced statistical machine learning approaches can be powerful tools for discovery of phenotypes based on treatment response and can aid in asthma control prediction in complex medical conditions such as asthma.

Gene expression profiling of asthma phenotypes demonstrates molecular signatures of atopy and asthma control

Recent studies have used cluster analysis to identify phenotypic clusters of asthma with differences in clinical traits, as well as differences in response to therapy with anti-inflammatory medications. However, the correspondence between different phenotypic clusters and differences in the underlying molecular mechanisms of asthma pathogenesis remains unclear. We sought to determine whether clinical differences among children with asthma in different phenotypic clusters corresponded to differences in levels of gene expression. We explored differences in gene expression profiles of CD4(+) lymphocytes isolated from the peripheral blood of 299 young adult participants in the Childhood Asthma Management Program study. We obtained gene expression profiles from study subjects between 9 and 14years of age after they participated in a randomized, controlled longitudinal study examining the effects of inhaled anti-inflammatory medications over a 48-month study period, and we evaluated the correspondence between our earlier phenotypic cluster analysis and subsequent follow-up clinical and molecular profiles. We found that differences in clinical characteristics observed between subjects assigned to different phenotypic clusters persisted into young adulthood and that these clinical differences were associated with differences in gene expression patterns between subjects in different clusters. We identified a subset of genes associated with atopic status, validated the presence of an atopic signature among these genes in an independent cohort of asthmatic subjects, and identified the presence of common transcription factor binding sites corresponding to glucocorticoid receptor binding. These findings suggest that phenotypic clusters are associated with differences in the underlying pathobiology of asthma. Further experiments are necessary to confirm these findings.

Effect of Inhaled Glucocorticoids in Childhood on Adult Height

HW Kelly, AL Sternberg, R Lescher; CAMP Research Group. N Engl J Med. 2012;367(10):904–912 Inhaled glucocorticoids (ICS) have been shown to have a temporary decrease in linear growth velocity in children. The effect of ICS on adult height has not been well defined. Participants (943 of 1041) in the Childhood Asthma Management Program study were enrolled between December 1993 and September 1995. They were between 5 and 13 years of …

Lung Function Impairment Evidenced by Sequential Specific Airway Resistance in Childhood Persistent Asthma: A Longitudinal Study

Background. Specific airway resistance (sRaw) is virtually independent of lung growth, height, and gender, thus facilitating longitudinal follow-up. Objective. To assess whether a specific phenotype of asthmatic children with a decline in lung function can be evidenced using sRaw. Methods. The authors hypothesized that sequential sRaw measurements over a long period would detect subtle trends. Clinical and functional data of children with persistent asthma under inhaled corticosteroids, evaluated at least three times per year for at least 4 years, were retrieved from a database. Results. One hundred fourteen children (30 girls) were followed for (median [interquartile range]) 6.9 years [5.6–7.9]. Data from 1699 measurements of sRaw (median 14/child) allowed the calculation of individual slopes of sRaw plotted against time demonstrating stable values in the group as a whole between 4 and 18 years. A positive correlation between individual slopes and the degree of intraindividual variation of sRaw was observed (R2 = .16; p < .0001). Children with more than one positive skin test showed larger intrasubject variation of sRaw (p = .011). In 19/114 children (17%), a significant increase in sRaw of 12.3% per year (median) was observed. As compared to children without, those with a significant increase in sRaw were boys (p < .0001), had a lower initial (p = .008) and a higher final resistance (p = .025) but did not differ in terms of inhaled corticosteroid dose. Conclusion. This retrospective study identifies a specific phenotype of asthmatic children that develops an impairment of lung function, confirming the results of a post hoc analysis of the Childhood Asthma Management Program study.

Serum vitamin D levels and severe asthma exacerbations in the Childhood Asthma Management Program study

Asthma exacerbations, most often caused by respiratory tract infections, are the leading causes of asthma morbidity and comprise a significant proportion of asthma-related costs. Vitamin D status might play a role in preventing asthma exacerbations. We sought to assess the relationship between serum vitamin D levels and subsequent severe asthma exacerbations. We measured 25-hydroxyvitamin D levels in sera collected from 1024 children with mild-to-moderate persistent asthma at the time of enrollment in a multicenter clinical trial of children randomized to receive budesonide, nedocromil, or placebo (as-needed beta-agonists): the Childhood Asthma Management Program. Using multivariable modeling, we examined the relationship between baseline vitamin D levels and the odds of any hospitalization or emergency department visit over the 4 years of the trial. Thirty-five percent of all subjects were vitamin D insufficient, as defined by a level of 30 ng/mL or less 25-hydroxyvitamin D. Mean vitamin D levels were lowest in African American subjects and highest in white subjects. After adjusting for age, sex, body mass index, income, and treatment group, insufficient vitamin D status was associated with a higher odds of any hospitalization or emergency department visit (odds ratio, 1.5; 95% CI, 1.1-1.9; P = .01). Vitamin D insufficiency is common in this population of North American children with mild-to-moderate persistent asthma and is associated with higher odds of severe exacerbation over a 4-year period.

Inhaled Corticosteroids for Young Children with Wheezing

In the Childhood Asthma Management Program study of school-aged children with established mild-to-moderate persistent asthma, twice-daily therapy with inhaled corticosteroids controlled symptoms of asthma but slightly reduced height growth and did not improve lung growth or affect the natural course of the disease.1 In this issue of the Journal, two studies involving young children2,3 test the hypothesis that the natural history of wheezing in early life may be altered by the administration of continuous or intermittent doses of inhaled corticosteroids when symptoms first occur but when the diagnosis of asthma is still uncertain. Both groups of authors conclude that . . .

Eotaxin polymorphisms and serum total IgE levels in children with asthma.

Eotaxin (chemokine, CC motif, ligand; CCL11) is a potent eosinophil chemoattractant strongly implicated in the pathobiology of asthma. Genetic variation at the CCL11 locus has been correlated with serum total IgE, blood eosinophil counts, and circulating eotaxin protein levels in several case-control asthma studies. Family-based association studies of CCL11 genetic variants have not been reported to date. To evaluate 9 common CCL11 single nucleotide polymorphisms (SNPs) in nuclear families ascertained through patients with asthma participating in the Childhood Asthma Management Program study. Single nucleotide polymorphism genotyping was performed by using minisequencing and probe hybridization platforms. Family-based association analysis for asthma and 4 asthma-related intermediate quantitative phenotypes was performed by using FBAT. One SNP, -384A>G, was associated with asthma among African American families (P = .01). CCL11 SNPs and haplotypes were not associated with asthma among white or Hispanic families. Two low-frequency alleles in strong pairwise linkage disequilibrium, -426C and IVS2+199A, were associated with lower serum total IgE levels (P = .0006 and P = .009, respectively) in white families, whereas 2 more common variants, -576C and g.4438C, were associated with higher IgE levels in African American families (P = .01-.04). Haplotype analysis in the white cohort provided additional evidence of association with serum total IgE, implicating 2 haplotypes. No single SNP or haplotype associations were observed with blood eosinophil levels, FEV(1), or airway responsiveness. These findings provide further evidence that genetic variation at the CCL11 locus is an important determinant of serum total IgE levels among patients with asthma.

EFFECTS OF AMBIENT AIR POLLUTION ON SYMPTOM SEVERITY AND MEDICATION USE IN CHILDREN WITH ASTHMA

Slaughter JC, Lumley T, Sheppard L, Koenig JQ, Shapiro GG. Ann Allergy Asthma Immunol . 2003;91:346–353 To investigate the short-term effects of ambient air pollution on asthma symptoms and medication use among children with persistent asthma. A total of 133 children (5–13 years of age) with mild/moderate asthma were studied. The mean duration of asthma was 5.3 years. The children were enrolled from 1 center participating in the Childhood Asthma Management Program study. During the run-in period of the Childhood Asthma Management Program study, before being placed on 1 of the study …