INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The assessment of genetic factors involved in the development of asthma is complicated because asthma is a heterogeneous, multietiologic disease (1); it has a high prevalence, estimated at 5% or more in the general population (2, 3), and it lacks a universally accepted definition (4). Thus, asthma is often considered a polygenic disorder with gene-environment interaction (5). Current research in the genetics of asthma and its related phenotypes (i.e., atopy and bronchial responsiveness) is focused mainly in three areas: the relationship between atopy/ bronchial responsiveness and some markers on chromosome 11q (6); the relationship between bronchial responsiveness, the production of total IgE, and some markers on chromosome 5q (7); and the association between allergic asthma and some class II human leukocyte antigens (HLA) on chromosome 6. Recent studies including wide genome searching have shown multiple linkages of asthma to other locations (8). Polymorphism in the HLA Class II molecules may lead to allelic forms that are more effective in binding allergenic peptides (i.e., epitopes) on the membranes of antigen-presenting cells, thus leading to allelic disequilibrium of HLA Class II alleles among sensitized individuals (9, 10). The most replicated and possibly the strongest association between the HLA system and allergic disease is that between increased IgE production in response to the ragweed Artemisia artemisiifolia pollen allergens Amb a V, Amb t V, Amb pV, and Amb a VI in individuals expressing the HLA-DR5 allele (11). From 1981 to 1987, 26 asthma outbreaks, affecting a total of 688 individuals, were detected in Barcelona, Spain. These outbreaks were caused by the inhalation of soybean dust. Although the exposure to airborne soybean dust was universal in Barcelona, only a small proportion of asthmatic individuals living in Barcelona expressed the epidemic phenotype. In a way, soybean epidemic asthma resembles occupational asthma in the community in terms of novelty of exposure, latency interval, intensity and avoidance to exposure (12). Because of these similarities, we intended to describe a relationship of the soybean epidemic with the HLA Class II system, just as others have described similar relationships for specific environmental or occupational allergens. We therefore investigated whether susceptibility or resistance to soybean epidemic asthma was associated to specific HLA-DR or -DQ genes.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Population

In 1995, we contacted a representative random sample of 316 adult patients treated in the emergency departments of hospitals in Barcelona from 1981 to 1987 for acute attacks of asthma (13). These individuals were seen on days of epidemics of asthma caused by soybean dust (169 patients in the epidemic asthma group) or on usual days (147 patients in the nonepidemic group). On the basis of a definition of asthma as current symptoms of asthma plus bronchial hyperresponsiveness (BHR) and/or atopy, the positive predictive value of a diagnosis of asthma in the hospitals' emergency rooms was 86.3% for epidemic and 91.9% for nonepidemic asthmatic individuals (14). Levels of low-molecular-weight soybean allergen in air samples collected during epidemic and nonepidemic days were 7,131.4 U/m³ on epidemic days (n = 3) versus 428.4 U/m³ on nonepidemic days (n = 17) (p < 0.001) (15). At the time of the study, 31 individuals were dead (18 with epidemic asthma [10.6%]; and 13 with nonepidemic asthma [8.8%], and 29 had moved away from Barcelona (12 with epidemic asthma [7.1%]; and 17 with nonepidemic asthma [11.5%].

Two hundred and three (203) of 256 eligible individuals were studied (78.98%), with no difference in the participation rates of epidemic and nonepidemic asthmatic individuals (p = 0.28). Blood samples suitable for genetic analyses were collected from 145 individuals. Demographic and clinical characteristics of the 58 participants from whom a blood sample was not available for genetic study are presented in Table 1. A control population from Barcelona, consisting of anonymous blood donors who provided 168 samples, was analyzed to compare the allele distribution in the general population with that in the asthmatic subjects. Field work was done from March to December 1995. All individuals, both asthmatic subjects and population controls, were caucasian and born in Spain. All field work and laboratory analyses were performed under conditions blinded for the status of cases and controls. The study was approved by the institutional committees on ethical practice of the participating insitutions, and all of the individuals in the study gave informed written consent.

Atopy

Total IgE levels were assessed by fluoroimmunoassay (CAP system, Pharmacia Diagnosis AB, Uppsala, Sweden), and values are expressed in kU/L using an IgE standard calibrated against reference preparation 75/502 of human serum IgE from the World Health Organization (WHO). Atopy was assessed through standard skin prick testing with common allergens, which included house dust, Dermatophagoides pteronyssinus, D. farinae, cat, dog, tree mix, graminaceae mix, weed mix, Aspergillus, and Alternaria (CBF LETI, Barcelona, Spain). A reaction was considered positive when a wheal diameter of 3 mm or more for one or more allergens was observed in an individual with a positive histamine control and a negative saline control (13). Testing of skin soybean sensitivity was done with a dry extract from the bean hull at a dilution of 1:10 of protein at a concentration of 100 µg/ml with 50% sterile glycerine (16).

Bronchial Responsiveness

Bronchial responsiveness was assessed by means of a methacholine challenge or a bronchodilator test. Exclusion criteria for spirometry were tobacco smoking within 1 h before the challenge, use of a -agonist 4 h before the challenge, or respiratory infection during the week preceding the challenge. Exclusion criteria for the methacholine challenge were medical reasons (severe heart problems, epilepsy, pregnancy, breast feeding, or intake of a -blocker), a best FEV₁ below 70% of expected or less than 1.5 L, and inability to perform the maneuvers.

Methacholine was administered through a Hudson nebulizer at concentrations of 0.5, 1, 2, 5, 10 and 25 mg/ml. The test was stopped when a decrease of 20% or more in FEV₁ as compared with the baseline FEV₁ was recorded, or when the highest dose of methacholine was administered. All individuals not taking a methacholine challenge test (due in 95% of cases to an FEV₁ below the predicted value underwent a bronchodilator test with salbutamol (13). Four inhalations of salbutamol (400 mg) were administered, and the test was considered positive if after 15 min FEV₁ increased by 15% in relation to the basal FEV₁. Individuals with positive challenge-test results for methacholine or salbutamol were considered to have airway hyperreactivity.

HLA Class II Polymorphisms

Twenty milliliters of ethylenediamine tetraacetic acid (EDTA)-anticoagulated peripheral blood was removed from each individual. DNA was extracted according to the salting-out method (17). Handling was performed at 4° C. Two polymerase chain reaction (PCR) typing methods (single specific primer [SSP] and single specific oligonucleotide [SSO] were used simultaneously or alternatively to fully define the HLA-DRB1 alleles). Only INNO-LiPA HLA-DQB1 was used to define DQB polymorphisms.

PCR-SSP typing was performed according to Olerup and colleagues (18). Primer mixes, ready to use, were obtained from Dynal (Oslo, Norway). From 0.25 to 1 µg of sample DNA and 5 µl of each primer mix (including an internal control to amplify a segment of the human growth hormone) were added to the PCR buffer (500 mM KCl, 15 mM MgCl₂, 100 mM Tris-HCl; pH 8.3; 10 mM deoxynucleotide triphosphates (dNTPs); 5% glycerol; and 100 µg/ml cresol red) and 1 U Taq polymerase (GIBCO-BRL, Gaithersburg, MD). Amplifications were done in a Perkin Elmer Model 9600 DNA thermocycler (Perkin Elmer Cetus, Norwalk, CT) with a PCR profile of 94° C for 2 min; 10 cycles at 94° C for 10 s each; 65° C for 1 min; 20 cycles at 94° C for 10 s each; 61° C for 50 s; and 72° C for 30 s. PCR products were directly electrophoresed in 2% prestained agarose gels, and visualized under UV illumination. Samples were typed with the high-resolution technique.

The PCR-SSO method is based on the pattern of hybridization of a DRB1 universal PCR product with specific oligoprobes (19). During PCR, biotinylated nucleotides are incorporated into the amplified DNA fragments. The PCR product is then hybridized with specific oligonucleotide probes immobilized in a parallel-lines format on membrane-based strips. After hybridization followed by a stringency wash, streptavidin labeled with alkaline phosphatase is added and becomes bound to any biotinilated hybrid. Incubation with 5-bromo-4-chloro-3-indole phosphate/nitroblue tetrazolium (BCIP/NBT) chromogen results in a colored precipitate.

The oligonucleotide probes were designed by Innogenetics (Ghent, Belgium). Thirty-one oligoprobes were used to define 72 DRB1 allele, and 21 oligoprobes were used in defining 25 DQB1 alleles.

Statistical Analysis

A sample-size calculation was done prior to starting the study. We took into account the distribution of aspirin-sensitive asthma (20), with an HLA-DQw2 frequency of in 65% of cases and 32% of controls. Assuming an alpha of 5% and a power of 80%, we could detect a statistically significant association for a risk ratio of 1.97 if we were selecting a minimum of 43 cases and 43 controls.

Discrete data, including data for specific genes, were analyzed with the chi-square test, with the use of Yate's correction if an expected cell was lower than 5 (21). Values of p are presented as corrected for the number of comparisons according to the method of Svejgaard and Ryder (22). For continuous measurements, a classical two-tailed t test was used. The attributable risk was determined according to the method described by Thomson and colleagues (23). The strength of the association of the genetic factor and the disease (asthma, atopy, or BHR) was estimated by calculation of the odds ratios (OR) with SPSS-PC (SSPS, Inc., Chicago, IL) and EpiInfo software (Centers for Disease Control and Prevention, Atlanta, GA), and use of the Cornfield method for the calculation of 95% confidence intervals (CIs).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The characteristics of epidemic and nonepidemic asthmatics were similar to what they were when assessed in 1989 (Table 1). Epidemic asthmatic individuals were older and less frequently female (p < 0.05). No differences were observed in the percent predicted basal FEV₁, bronchial responsiveness, or use of inhaled corticosteroids. A trend among the epidemic asthmatic patients toward being more atopic was observed, with a greater geometric mean for total IgE (p < 0.05), although with a distribution of skin reactivity to a battery of common aeroallergens similar to that of the nonepidemic asthmatic subjects (60.9% versus 53.7%, p = n.s.). The main difference between the two groups of asthmatic subjects was a 10-fold greater sensitivity to soybean in epidemic asthmatic subjects (p < 0.05). HLA-DR gene frequencies are shown in Table 2. As can be seen, there was a different distribution of genes between epidemic and nonepidemic asthmatics, and also compared with controls from the general population. The DRB1*13 gene was particularly associated with the risk of epidemic asthma (p value corrected for multiple comparisons < 0.02). No HLA-DQ allele was associated with epidemic asthma (Table 2).

When individuals rather than allelic distributions are considered (Figure 1), DRB1*05 or DRB1*06 occurring in a given individual were independent risk factors, increasing to more than 100% the chance of being considered as having epidemic asthma, particularly for allele DRB1*13 (OR: 3.22 (1.38 to 7.50). This association was stronger in patients with the lowest risk levels of total IgE. Individuals in the lowest tertile of total IgE (2 to 46.3 kU/L) had a risk of 14.5 (1.6 to 130.8) of developing the epidemic asthma phenotype, as compared with an OR of 1.33 (0.3 to 6.7) for patients in the mid tertile of total IgE, and an OR of 1.93 (0.2 to 18.9) for those in the highest tertile of total IgE (Table 3).

View larger version (15K): [in this window] [in a new window]   Figure 1.   Association of presence of at least one specific DRB1 allele and epidemic asthma (OR and 95% CI). Broken lines indicate that upper limit of the CIs is greater than 4.

The haplotype combinations of homozygosity for DRB1*05-05 and DRB1*06-06, and heterozygosity for DRB1*05-06, were found only in epidemic asthmatic individuals and there was a trend toward susceptibility to developing the epidemic asthma phenotype in individuals heterozygous for DRB1*03-05 or DRB1*03-06. According to the Hardy-Weinberg equilibrium for the control population, we would have expected to find 5.99% of all individuals having alleles DRB1*05-05, DRB1*06-06, or DRB1*05-06. However, the three heplotype combinations occurred respectively in two, four and four of 77 epidemic asthmatic individuals (12.98%), whereas none of these haplotype combinations was found in 68 nonepidemic asthmatic individuals (p < 0.02). A recent report (24) suggests that the -chain gene is always coexpressed with genes DRB1*03, DRB1*05, and DRB1*06. Assuming this in our study, the ₃ gene was present 57.0% of haplotypes in epidemic asthmatic individuals and 45.0% of haplotypes in nonepidemic ones, producing an attributable risk of 27.9% of an individual risk for patients expressing the ₃ chain gene of 1.56 (0.8 to 3.0) of being considered epidemic asthmatic. When epidemic and nonepidemic asthmatic individuals were pooled, no differences in the HLA Class II distribution were observed for bronchial responsiveness or atopy.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have described an association of soybean epidemic asthma with specific HLA Class II alleles as compared with appropriate control subjects with asthma or the general population. Presence of allele DRB1*13 conferred the greatest risk of having epidemic asthma, of 3.22 (1.38 to 7.50), particularly for subjects with the lowest concentration of total IgE. On the contrary, no HLA-DQ association could be observed.

A critical issue in the study of associations between the HLA system and specific immunoresponse is the need for testing allergens with a high degree of purity, often greater than 99.5% (25). Although several efforts have been made to characterize the molecule responsible for soybean epidemic asthma (26, 27), it has so far not been fully characterized; we tested individuals in our study with a crude soybean extract likely to include a large number of epitopes (16). However, alike the finding by Marsh and colleagues for HLA-A2 and Ra3 of Ragweed (28), our results for an association between epidemic asthma and DRB1*13 were stronger in individuals with low levels of total IgE. The lower the concentrations of total IgE, the more specific epitopes were recognized, such that the association with HLA was reinforced. Thus, it can be hypothesized that in future studies, when using purer extracts or even the causative molecule, the associations observed can be strengthened.

Most studies on the association of genetic factors and asthma have methodologic problems relating to sample size and the selection of asthma cases and controls (29). By studying a sample of moderate size and selecting asthmatic individuals attending hospital emergency departments on days without epidemic asthma (5.6% skin positive to soybean) and individuals from the general population of Barcelona (about 1% of whom without epidemic asthma have a measureable serum-specific IgE to soybeans [CAP method > 0.35 kU/L] according to our own unpublished data from the European Community Respiratory Health Survey in Barcelona) as controls, we tried to overcome this caveat. The possibility of a bias due to misclassification of the asthmatic and population controls must be discussed. We must keep in mind that we studied the association of HLA Class II genes and soybean epidemic asthma 8 yr after the elimination of outbreaks of this disease (30, 31), although sporadic releases of high amounts of soybean dust were reported on 29 October 1994 (32) and 12 June 1996 (33). Assessing clinical asthma, BHR, or atopy was not possible in the population control sample, which consisted of anonymous healthy blood donors. Because of protocol restrictions by the ethics committee, we were unable to identify blood donors and invite them for a full clinical assessment; consequently, no phenotypic or demographic data are available. However, it is our opinion that the HLA gene frequencies of the blood donors would have (Table 2) more closely resembled the gene distribution of the nonepidemic (a pooled mix of asthmatic individuals) than of epidemic asthmatic patients. In any case, if we estimate that about 5% of the general population has asthma (2), a certain degree of underestimation could be expected. As with the approach taken in occupational asthma, pooling cases affected only by a single exposure is an appropriate way of searching for susceptibility genes, especially in the case of the Barcelona epidemics, when the city was globally exposed but only a few hundred asthmatic individuals were affected. Although soybean epidemic asthma has a necessary cause (i.e., exposure to soybean dust), we cannot yet estimate the proportion of disease due to genes, if any. Indeed, the association reported here for DRB1*13 accounts for about 12% of the attributable risk of epidemic asthma only, and for the ₃ gene, which is always coexpressed with genes DRB1*03, DRB1*05, and DRB1*06, for an attributable risk of 27.9%. Interestingly, the haplotype combinations DRB1*05-05, DRB1*05-06, or DRB1*06-06 were present only in epidemic asthmatic individuals such that an unquantifiably high risk for these genotypes could be calculated in terms of the control asthmatic subjects. We must keep in mind that any genetic assessment in the population of soybean asthmatic individuals implies a necessary but not sufficient cause of asthma, in the form of exposure to airborne soybean dust. Although other factors have been reported to be associated with soybean epidemic asthma, such a as older age, male sex, current cigarette smoking, and area of residence close to sites of unloading of soybeans (12, 34), complete avoidance of exposure would lead to a safe environment and a risk of zero for any individual with genetic susceptibility. Because there was no a priori hypothesis about the HLA Class II genes involved in soybean epidemic asthma, replication of our results, or lack of it, should come from an analysis of the entire cohort of Barcelona (about 500 epidemic athmatic individuals who are still alive), or from an assessment in other areas where soybean epidemic asthma has been documented (e.g., New Orleans, LA [35] or Naples, Italy [36]).

New studies will further assess the gene-environment interaction in soybean epidemic asthma, including a high-resolution HLA gene search, associations with a candidate epitope of the low-molecular-weight soybean allergen, a wide-genome search of candidate genes in a sample of soybean epidemic asthmatic individuals, and polylmorphisms of other molecules involved in peptide transportation to coupling with HLA Class II molecules (DMB). To date, the HLA findings described in this report partly explain why only a proportion of asthmatic individuals in Barcelona developed the soybean phenotype. Because inhabitants in Barcelona affected by the soybean asthma epidemics have been so thoroughly characterized, this sample may be suitable for quantifying the separate contribution of genes and of the environment in asthma.