INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It is now widely recognized that asthma is a disease of airway inflammation (1) in which a variety of cell types, including eosinophils (4), mast cells (5), and T lymphocytes (6), contribute to a complex pathologic process producing recurrent symptoms, intermittent airflow obstruction, and ultimately compromised baseline lung function.

Eotaxin is a chemotactic cytokine (chemokine) that recruits eosinophils by activating the CCR3 receptor. Activation of leukocyte chemokine receptors is part of the exquisitely regulated process that allows circulating leukocytes to move from the circulation into the tissues; this process involves selectins, integrins, and chemokines. Eotaxin is one member of a growing family of ligands for the CCR3 chemokine receptor present on eosinophils (7), basophilic leukocytes (8), and T lymphocytes (9). The discovery of eotaxin in the lung lavage fluid of allergen-challenged guinea pigs has led to speculation about its role in asthma (10). Eotaxin is produced by airway epithelial cells after stimulation by cytokines (11) and acts at the CCR3 receptor on eosinophils to induce chemotaxis (12). In asthmatic airways, eotaxin has been localized immunohistochemically to the epithelium where eosinophils are known to reside (13, 14). It is now known that airway eotaxin concentrations correlate with the sensitivity of asthmatic airways to contractile stimuli (14) and that the profusion of airway eosinophils and their activation products correlates with impaired lung function (15, 16). In addition, genetic markers near the eotaxin gene (chromosome 17q21.1) have been associated with asthma in one genome-wide search for asthma susceptibility loci (17).

To determine whether eotaxin plays a role in the asthmatic phenotype in the absence of specific provocation, we explored the relationship among plasma eotaxin levels, asthma diagnosis, and disease severity as measured by FEV₁. We compared eotaxin concentrations with eosinophil counts (15, 16) and with total immunoglobulin (Ig)E concentrations (18), which are established biomarkers of allergy as predictors of asthma and asthma severity. We found that plasma eotaxin levels are directly associated with asthma diagnosis and correlate inversely with lung function in a genetically and geographically diverse population.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

In a case-control study of 1,034 adults (515 asthmatics and 519 nonasthmatics), subjects assigned to the asthma case group were recruited at 40 sites in the United States for participation in a therapeutic trial with a novel antiasthma agent. Each had a physician's diagnosis of asthma and documented airflow reversibility of at least 15% with a short-acting -agonist. Subjects assigned to our normal control group either were recruited from participating centers or were part of a U.S. Army cohort of volunteers. Normal subjects had never had a physician's diagnosis of asthma, reported no respiratory symptoms on a standardized questionnaire, and had an IgE level of less than 150 kIU/L (21) or negative 12-allergen dermal testing with a positive control. Asthmatic subjects had had a chest radiograph within 1 yr before the study, with no significant abnormalities except those relating to asthma; symptoms of asthma were controlled with an inhaled short-acting -agonist (albuterol) and nasal symptoms with loratadine. All subjects had been free of respiratory system infections or asthma exacerbations for at least 4 wk at the time of the study. Asthmatic subjects either were former smokers, defined as having been abstinent for at least 1 yr and having less than 10 pack-years of cumulative exposure, or had never smoked. At the time of data acquisition, no subject had used corticosteroids (oral or inhaled), nedocromil, or cromolyn sodium for at least 1 mo, and none was taking long-acting -agonists, ipratropium bromide, astemizole, terfenadine, cetirizine, hydroxyzine, oral -agonists, aspirin, nonsteroidal anti-inflammatory drugs, or nasal steroids. All subjects gave written informed consent with the prior approval of the appropriate institutional review boards.

Study Measurements

Spirometry was performed in accordance with the standards of the American Thoracic Society. IgE levels were measured in duplicate with Uni-cap technology in accordance with the specifications of the manufacturer (Pharmacia, Kalamazoo, MI). The coefficient of variation for each assay was less than 2%. Plasma eotaxin levels were determined in duplicate (as previously described) in an ELISA able to detect 30 pg of eotaxin/ml with a coefficient of variation of 10% (11). This assay system accurately detects exogenous eotaxin added to ethylenediaminetetraacetic acid (EDTA)-preserved human plasma having varying levels of endogenous eotaxin, and therefore is likely to detect both bound and free eotaxin. Interplate variability was reduced by normalization to plasma samples of known concentration. Eosinophil counts were determined from Coulter Counter leukocyte measurements and manual differential counting by personnel who were unaware of the status of the subjects from whom the samples were obtained.

Statistical Analysis

Demographic data were compared by chi-square analysis, Wilcoxon's rank sum test, or Student's t test as appropriate. Plasma eotaxin levels from normals and asthmatics were analyzed with the Mann-Whitney rank sum test. Unless otherwise stated, correlative analyses were performed on data from our entire population. Before adjustment for demographic parameters, a Pearson product moment correlation analysis was performed. Stepwise logistic regression analysis assessed the efficacy of models using various allergic markers to predict asthma from data adjusted for age, race, sex, and smoking status. The efficacy of models using combinations of allergic markers was compared by logistic regression analysis with better models producing lower log likelihood scores. A multiple linear regression analysis of adjusted data compared models using various allergic markers to predict reduction in percent predicted FEV₁. A p value of less than 0.05 was considered statistically significant. Analyses were performed with SAS software version 6.12 (SAS Institute Inc., Cary, NC).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Demographic Information

The sex ratio of our asthmatic group was not significantly different from that of our normal group (Table 1). The asthmatic group was slightly older (p = 0.0041) and comprised more Caucasians and fewer African Americans (p = 0.012) than the normal control group. The asthmatic group had significant impairment of lung function, with a mean FEV₁ of 57% of predicted.

Plasma Eotaxin Levels

The median plasma eotaxin level in asthmatics (355 pg/ml; interquartile range, 237 to 536; n = 515) was significantly greater than that in the nonasthmatic group (287 pg/ml, 200 to 378; n = 519; p < 0.001).

Correlation of Plasma Eotaxin with Parameters of Asthma and Atopy

Plasma eotaxin levels were inversely correlated with percent predicted FEV₁ and directly related to peripheral blood eosinophil counts (Table 2). The correlation coefficient with circulating eosinophil counts was modest; that with IgE levels was very small and did not reach statistical significance.

Predicting Asthma from Plasma Eotaxin Levels

To quantify the relative risk of asthma associated with eotaxin, we divided our population by quartiles according to eotaxin levels and calculated the risk of having asthma in the three higher quartiles relative to that in the lowest quartile. Subjects in the lowest quartile had a median eotaxin level of 158 pg/ml. The median eotaxin level of the next highest quartile was 266 pg/ml. This group was 1.9 times more likely to have an asthma diagnosis than the lowest quartile group (95% confidence interval, 1.1 to 3.0; p = 0.012). The median eotaxin level in the next highest quartile was 368 pg/ml. In this group the odds of having asthma were 2.7 times higher than in the lowest quartile group (1.6 to 4.4, p < 0.0001). The highest quartile had a median eotaxin level of 603 pg/ml; this group was 5.4 times more likely than the lowest eotaxin group to have asthma (3.2 to 9.2, p < 0.0001; Figure 1).

View larger version (19K): [in this window] [in a new window]   Figure 1.   Odds ratios and 95% confidence intervals for having asthma: a comparison of asthma risk in quartiles of subjects with increasing eotaxin concentrations after correction for age, sex, smoking habits, and race. p Values reflect comparisons with the lowest eotaxin quartile, which had a median eotaxin concentration of 158 pg/ml.

Predicting Asthma from Allergic Markers

Logistic regression analysis was used to compare the predictive ability of models using combinations of biomarkers. Plasma eotaxin levels significantly improved all of the models we studied and added to the predictive power of the best model (p = 0.023). Part of the predictive power of eotaxin levels was independent of the predictive power of IgE levels, eosinophil counts, and demographic variables, which were included in the best model.

Correlations of Allergic Markers with Impairment of Lung Function (Reduction in FEV₁)

Plasma eotaxin levels were inversely correlated with percent predicted FEV₁, with a correlation coefficient of 0.23 (n = 761, p < 0.001) without adjustment for age, sex, race, or smoking status. Similarly, eosinophil counts (correlation coefficient, 0.22; n = 662; p < 0.001) and IgE levels (correlation coefficient, 0.24; n = 672; p < 0.001) were inversely correlated with percent predicted FEV₁.

Predicting Impairment of FEV₁ from Plasma Eotaxin Levels

Eotaxin predicted impairment of FEV₁ independent of age, sex, race, and smoking status (p < 0.001). To determine whether this association was of a magnitude that could be clinically significant, we again considered our population according to quartiles. The median percent predicted FEV₁ in the highest eotaxin quartile was 13.9 ± 2.1 percentage points less than that in the lowest eotaxin group (p < 0.0001), that in the next highest eotaxin quartile was 9.4 ± 2.2 percentage points less than that of the lowest eotaxin group (p < 0.0001), and that in the third highest eotaxin quartile was 5.6 ± 2.2 percentage points less than that in the lowest eotaxin group (p = 0.011). This progressive reduction in lung function with increasing eotaxin level is presented in Figure 2.

View larger version (17K): [in this window] [in a new window]   Figure 2.   Reduction in percent predicted FEV1 and standard error in an adult population: a comparison of lung function in quartiles of subjects with increasing eotaxin concentrations after correction for age, sex, smoking habits, and race. p Values reflect comparisons with the lowest eotaxin quartile, which had a median eotaxin concentration of 158 pg/ml.

Predicting Asthma Severity from Allergic Markers

Multiple linear regression analysis was performed on our entire population to determine the usefulness of alternative biomarkers in models predicting impairment of baseline lung function. Eotaxin concentrations added significantly (p = 0.018) to the best model, which included IgE concentrations and demographic variables. The inclusion of plasma eotaxin levels was associated with a significant improvement in the predictive ability of all of the models studied. Similarly, eotaxin concentrations contributed to the predictive power of these models when our asthmatics were considered as a group.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Plasma eotaxin levels predicted asthma diagnosis and impairment of lung function independent of age, race, sex, and smoking status in a population of adult subjects who were not receiving treatment with corticosteroids. The study was conducted in steroid-free subjects because steroids reduce eotaxin expression in human cells (11) and affect airway and peripheral blood eosinophilia (15, 16, 22) and serum levels of total IgE (23). We found progressive, statistically significant, stepwise increases in the risk of having an asthma diagnosis with successive increases in plasma eotaxin level; the highest eotaxin quartile had a fivefold higher odds ratio than the lowest. Logistic regression analysis demonstrated that eotaxin added modestly to the ability of IgE concentrations and eosinophil counts to predict an asthma diagnosis. However, a significant part of this predictive power was independent of that provided by the established markers. This analysis demonstrates that plasma eotaxin levels are associated with asthma diagnosis independent of age, race, sex, or smoking status.

The importance of airway eosinophilia in asthma and the known biology of the CCR3 receptor ligand system suggested to us that the biomarker eotaxin would be related to impaired lung function in asthma. We found that percent predicted FEV₁ (which falls as asthma severity increases) progressively and significantly decreased as median eotaxin concentrations increased. The highest plasma eotaxin quartile had a 13-percentage-point reduction in percent predicted FEV₁ compared with the lowest. This degree of reduction is likely to be clinically significant, as a 10-percentage-point reduction increases the American Thoracic Society's suggested severity assessment category by one level (24). We explored this relationship, using a correlative analysis approach to better define the ability of plasma eotaxin levels to account for the variance in lung function in our population. Defining the factors that account for the observed variability in baseline lung function is central to establishing the disease relevance of markers reflecting processes that compromise lung function. Whereas airway eotaxin expression has been shown to correlate with the sensitivity of asthmatics to contractile stimuli, its effects on baseline lung function have not been reported.

We studied percent predicted baseline FEV₁ because it is the objective measure used by the National Institutes of Health Expert Panel to grade asthma severity (25) and because compromised lung function is the disease-relevant consequence of altered airway structure. We examined the ability of eotaxin to account for variance in lung function relative to IgE concentrations and eosinophil counts, as these biomarkers of sensitization and peripheral eosinophil mobilization are the best-known correlates of impaired lung function in asthma (15, 19). Knowledge of the relationship between plasma eotaxin and more direct indices of eosinophil recruitment such as bronchoalveolar lavage eosinophil measurements or physiological parameters such as methacholine sensitivity would be useful, but this information was not available for this study. Inclusion of eotaxin levels improved the predictive power of all of the models studied, including the best model, which considered IgE levels and demographic variables.

A significant part of the predictive power of eotaxin was independent of the predictive power of the other variables. This independent association has mechanistic implications: it would not exist if the dominant source of plasma eotaxin was from peripheral eosinophils, if the chief function of plasma eotaxin was to mobilize peripheral eosinophils, or if eotaxin was a surrogate marker for sensitization. The association of increasing eotaxin concentrations with clinically significant decreases in baseline percent predicted FEV₁ could occur only if plasma eotaxin levels reflect local events that impair lung function. Our current knowledge of eotaxin biology implicates plasma eotaxin as a systemic index of local eosinophil chemotaxis, because compromised lung function in asthma must depend both on the availability of circulating eosinophils and on their recruitment into the lung. The independent portion of this association could also be explained by links between eotaxin and alteration in lung structure and raises the possibility that eotaxin may have a role beyond its known effects on eosinophil chemotaxis.

The direct participation of eotaxin in asthma is suggested by the localization of eotaxin messenger RNA (mRNA) and protein to the airway epithelium, where CCR3-bearing eosinophils are found in asthma (13, 14). Airway epithelial cell lines mobilize eotaxin mRNA and secrete eotaxin protein in response to cytokines that are known to be available in asthmatic airways (11), and airway eotaxin expression is correlated with airway eosinophilia and with sensitivity to the contractile effects of histamine in asthmatics (14). Many studies of eosinophil biology support a role for this cell type and the CCR3 receptor ligand system in asthma. Eosinophils are present in the airways of persons with mild as well as severe asthma (4) and are reduced in number after asthma treatment (22). Eosinophils and their products correlate with asthma severity (5, 15, 26), and eosinophil-derived granular proteins (27), oxidants (28), and lipid mediators (29) cause airflow obstruction and changes in airway structure. Eotaxin may be more relevant to human asthma than is suggested by allergen challenge experiments in animals, where eotaxin abrogation by antibody neutralization, gene targeting, and receptor antagonism have demonstrated 30 to 60% reduction in airway responsiveness or antigen-induced eosinophil recruitment (30). The chronic nature of the disease and period of years required for impaired baseline lung function to be manifest are not well reflected in these animal models. That asthmatic airways are chronically exposed to eotaxin is inferred from studies demonstrating its constitutive airway expression in the absence of acute symptoms (13, 14); taken together with our findings, these studies imply that eotaxin contributes to impaired lung function through its effects on target cells, including eosinophils.

Recent advances in our understanding of immune alterations in allergic disease (34) have fostered the notion that T helper cell, type 2 (Th2) lymphocytes augment IgE production by producing cytokines, including interleukin-4 (IL-4). The recognition that some human lymphocytes polarized to the Th2 phenotype express the eotaxin (CCR3) receptor (9) has led to speculation that cytokines (including IL-4) produced by CCR3 receptor activation promote the production of IgE (35). Our data show that a significant but weak correlation exists between total IgE and plasma eotaxin and that some of the predictive power of plasma eotaxin levels is accounted for by IgE levels; thus eotaxin may contribute to but is not a dominant determinant of IgE production. Eotaxin may exert its effects through interaction with other Th2 cytokines, such as IL-5, which is important for eosinophil maturation, release, and activation (32). The independent predictive power of eotaxin levels suggests that activation of the CCR3 receptor ligand system itself is relevant to asthma pathogenesis. Our data do not allow us to further define this role, but we speculate that elevated eotaxin concentrations are an index of inflammatory cell recruitment to the airway and/or act as a marker for alterations in airway structure that are associated with impaired function.

In summary, we have demonstrated an association between elevated plasma eotaxin levels and asthma diagnosis and compromised lung function that is independent of age, race, sex, or smoking status. We have also shown that part of the predictive power of eotaxin is independent of that of eosinophil counts and total IgE levels. In the context of the established role of eosinophils in asthma and the eosinophil chemoattractant effects of eotaxin, the association between asthma severity and eotaxin concentrations implies that eosinophil chemotaxis is relevant to impaired lung function in asthma. In short, our data in a geographically dispersed and ethnically diverse population of asthmatics demonstrate that eotaxin is associated with asthma diagnosis and severity.