INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bronchial asthma, a chronic respiratory disease, affects over 17 million people in the United States (1). The prevalence of asthma is increasing in the United States and around the world (2, 3). Currently available data suggest that asthma has a hereditary component and that multiple genetic and environmental factors influence its development (4). No genes responsible for the transmission of asthma-related phenotypes, however, have been conclusively identified. Identifying the genetic determinants of bronchial asthma is of great importance, as this may ultimately provide clinicians with important tools in the diagnosis and management of this disease (4).

Even though numerous genetic studies of asthma have been conducted to date, comparing their results has been made difficult by the lack of a standardized definition of the asthma phenotype (5). To address this problem, an algorithm for the diagnosis of asthma was recently proposed for use in family-based genetic studies (9). Because this algorithm has been employed in a study of the genetics of asthma only in 92 families with asthma in a western population, we were interested in examining its use in a study of the genetics of asthma in 2,756 Chinese families living in a predominantly rural environment. Thus, we have classified members of these Chinese families using a slightly modified version of the aforementioned algorithm and have compared our results with those previously reported in a Dutch population (9).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Site

This study was conducted in collaboration with Anhui Medical University and the Anqing Health Bureau in Anqing, China. Anqing stretches for about 80 km along the northern bank of the Yangtze River. The annual average temperature is 15° C. Anqing has three urban areas and eight rural counties, with a total area of 15,000 km². The total population in 1990 was 5.8 million (urban 9%; rural 91%). Twenty nine percent of the population was younger than 15 yr of age, 63% between 15 and 59 yr of age, and 8% 60 yr of age or older. The average life expectancy was 65 yr. Anqing was selected for this study because of the homogeneity of its large population with regard to ethnicity, occupation, diet, and the environment and the stable resident population of its villages over several thousand years. In addition, medical care in each county is provided through a three-tier (county, township, and village) service network. Established 25 yr ago, this network includes 28,000 physicians and provides a unique opportunity for efficient and uniform identification of index cases.

Identification of Families with Asthma

Families with asthma from the eight counties of Anqing (Zongyang, Huining, Qianshan, Tongcheng, Taihu, Wangjiang, Susong, and Yuengsi) were enrolled in the study through a multistage process (10). First, investigators from Anhui Medical University and Anqing City Hospitals/Research Institutes held a 3-d workshop in each township to train township and village doctors and to collect information on families with asthma. The first day of this workshop was used to explain the purpose, scope, and procedures of the study. The definition of a family with asthma was discussed, and several examples were presented. Each village doctor was asked to go back to his or her own clinic to prepare a list of all families in which cases of asthma or bronchitis had been diagnosed by local doctors. This information on families was collected by the village doctors in a 2-d period. Pulmonologists then reviewed the family lists with the township/village doctors to exclude ineligible families. Cases of physician-diagnosed asthma had to meet the following criteria for a family to be eligible: (1) the patient had a history of repeated onset of wheeze with dyspnea but was asymptomatic between two events and (2) the patient's respiratory symptoms were relieved by bronchodilator use. Following the workshop, field team members and village doctors interviewed each eligible family using a short questionnaire to confirm the information provided by the village doctors. Information about family size, health status, respiratory symptoms, and medication use was also collected, and pedigree charts were drawn. The following criteria were used for inclusion of families with asthma in the study: (1) presence of at least two siblings with physician-diagnosed asthma who were at least 8 yr old (2) availability of both parents and (3) no more than one parent with physician-diagnosed asthma.

Procedures

Letters explaining the purpose of the study were sent to all eligible families. Local officials and health centers arranged for appointments to take place at the central office at a time convenient for the participants. Data were collected by faculty members from Anhui Medical University and trained interviewers between July 1, 1994 and January 26, 1998.

Unless otherwise specified, the following procedures were carried out in accordance with the NIH Collaborative Agreement on Asthma Genetics: (1) completion of a standardized questionnaire (modified ATS-DLD) including questions on respiratory symptoms, respiratory health status, occupational history, tobacco use, home environment, and family history of asthma and other chronic diseases; (2) pulmonary function testing (spirometry); (3) methacholine challenge testing; (4) bronchodilator testing; (5) skin testing of reactivity to 10 allergens along with a positive and a negative control; and (6) collection of blood samples for total serum immunogloblin E (IgE) level, eosinophil count, and DNA extraction. In addition, height and weight were measured by standard methods; subjects removed their shoes and outerwear before measurement. Height was measured to the nearest 0.1 cm on a portable stadiometer. Weight was measured to the nearest 0.1 kg with the subject standing motionless on the scale.

Spirometry

Pulmonary function tests were performed with ATS "Snowbird Guideline" approved criteria (Schiller, Switzerland) and with subjects seated and wearing a noseclip. As many as eight attempts were performed by each participant to obtain three acceptable measures. Spirometry was performed according to ATS specifications; the FEV₁/FVC from the best test effort and the highest FEV₁ value are reported (11).

Subjects were asked to avoid bronchodilator use for at least 4 h prior to spirometry unless respiratory symptoms required bronchodilator treatment. Spirometry was repeated 10 min after subjects inhaled 180 µg (2 puffs) of albuterol using a spacer device.

Methacholine Challenge Testing

Methacholine challenge testing, using the Chatham protocol, was performed in all subjects whose FEV₁ was > 60% of predicted (12). Subjects were asked to avoid bronchodilator use for at least 4 h prior to testing, unless symptoms required bronchodilator treatment.

Allergy Skin Testing

Skin testing was performed with a slightly modified version of the semiquantitative puncture method developed by Santilli and coworkers (13). In addition to histamine and saline controls, the following antigens were applied to the skin of the forearm: cockroach, house dust, mixed trees, mixed grasses, tobacco leaf, polyvalent molds, Dermatophagoides pteronyssinus, D. farinae, artemisia, and silk. A skin test was considered positive if the diameter of the wheal was  3 mm after subtracting the negative control.

Asthma Algorithm

We used a slightly modified version of the algorithm described by Panhuysen and coworkers to differentiate asthma from other obstructive airway diseases and from unaffected family members (9). Subjects were classified in five groups, as follows: definite asthma (class 1); probable asthma (class 2); unclassifiable airway disease (class 3); chronic obstructive pulmonary disease (COPD, class 4); and unaffected (no obstructive airway disease, class 5). The decision to assign a subject to a particular diagnostic class was made on the basis of five findings: increased airway responsiveness to methacholine (PD₂₀  50 mg/ml); cigarette smoking (> 5 pack-yr versus  5 pack-yr); asthma symptoms (cough, dyspnea, wheeze, and nocturnal symptoms); airflow obstruction (FEV₁/FVC < 0.75); and response to bronchodilator administration (increase of at least 9% in FEV₁%pred).

The first step in this algorithm is to classify subjects on the basis of their airway responsiveness to a bronchoconstrictor; subjects with increased airway responsiveness are not considered to be unaffected (class 5), and those without increased airway responsiveness are not considered to have definite asthma (class 1) (9). Subjects with missing data on airway responsiveness are excluded from the algorithm. Obtaining a smoking history is the second step. Subjects with a history of  5 pack-yr of smoking are not classified as having definite asthma, unless there is clear documentation of asthma symptoms/attacks preceding the onset of smoking; nonsmokers are not classified as having COPD (class 4) (9). The third step is classifying subjects on the basis of the presence of symptoms (cough, dyspnea, wheeze, and nocturnal symptoms) and asthma attacks. The fourth step in this algorithm is assessing the presence of airflow obstruction. Although subjects could be classified as having definite asthma in the absence of airflow obstruction (if meeting other criteria), the presence of airflow obstruction is a necessary criterion for the diagnosis of COPD. Finally, if airflow obstruction is present, reversibility (FEV₁%pred, > 9%) is assessed to differentiate subjects with COPD (class 4) from those with either probable asthma (class 2) or unclassifiable airway disease (class 3) (9). The presence of two or more symptoms in subjects 16 yr of age or older or a clear history of recurrent asthma attacks (regardless of age) in nonsmoking subjects with increased airway responsiveness to methacholine (PD₂₀  50 mg/ml) is considered compatible with definite asthma (class 1). Nonsmoking subjects younger than 16 yr of age who have one or more symptoms and marked airway responsiveness to methacholine (PD₂₀  8 mg/ml) are also classified as having definite asthma (class 1) (9). Nonsmoking subjects with increased airway responsiveness to methacholine (PD₂₀  50 mg/ml) were classified as having probable asthma if they were (1) younger than 16 yr of age and had one symptom and a PD₂₀ > 8 mg/ml of methacholine; (2) 16 yr of age or older and had one symptom and reversible airflow obstruction; (3) asymptomatic and had an increase in FEV₁%pred  9% following bronchodilator administration, regardless of age (9). Smoking subjects with increased airway responsiveness to methacholine (PD₂₀  50 mg/ml) were classified as having probable asthma if they (1) had a history of recurrent asthma attacks and either reversible airflow obstruction or no evidence of airflow obstruction; and (2) had one or more symptoms and no evidence of airflow obstruction, but had an increase in FEV₁%pred  9% following bronchodilator administration. Nonsmoking subjects who had no increased airway responsiveness to methacholine (PD₂₀ > 50 mg/ml) were classified as having probable asthma if they (1) had airflow obstruction and either a history of recurrent asthma attacks and one or more symptoms or two or more symptoms; (2) had no evidence of airflow obstruction, an increase in FEV₁%pred  9% following bronchodilator administration, and either a history of recurrent asthma attacks and one or more symptoms or two or more symptoms; and (3) had reversible airflow obstruction and either a history of recurrent asthma attacks or one or more symptoms (9).

Our algorithm differs from the one proposed by Panhuysen and coworkers in two respects. First, we measured airway responsiveness to methacholine, not histamine. The response to methacholine challenge testing that would be equivalent to the cutoff value used in the original algorithm to define increased airway responsiveness (PD₂₀  32 mg/ml of histamine) is not known. Thus, we chose a cutoff value for increased airway responsiveness (PD₂₀  50 mg/ml of methacholine) that is reasonably similar to that used by Panhuysen and coworkers and that would allow us to identify nonresponders. Second, we chose an FEV₁/FVC ratio < 0.75 as a marker of airflow obstruction because 95% confidence intervals (CIs) for FEV₁ have not been validated in a Chinese population.

Statistical Methods

Student's t tests were performed and 2 × 2 contingency tables were examined using the SAS statistical package (SAS Institute, Cary, NC) on Sun microsystems (Sun Microsystems, Mountain View, CA).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject Characteristics

We tested 12,995 subjects in 2,756 families: 5,324 parents and 7,671 offspring; 1,227 (23%) of the parents and 1,247 (16.3%) of the offspring did not have an adequate measurement of airway responsiveness to methacholine (Table 1). Among these 1,227 parents, 581 (47.4%) had an FEV₁  60% of predicted, 216 (17.6%) had medical conditions precluding testing, 320 (26.1%) could not perform the test properly, and 110 (8.9%) could not be scheduled for testing. Of the 1,247 offspring, 209 (16.8%) had an FEV₁  60%, 134 (10.7%) had medical conditions precluding testing, 601 (48.2%) could not perform the test adequately, and 303 (24.3%) could not be scheduled for testing. In both parents and offspring, those missing information on airway responsiveness were significantly more likely to have physician-diagnosed asthma and airflow obstruction than those with complete information (Table 1).

Asthma Algorithm

Table 2 illustrates the results of implementing the asthma algorithm in 4,097 Chinese parents and 6,424 Chinese offspring with complete information on airway responsiveness, as well as in 320 relatives (255 offspring and 65 second-degree relatives) of 92 Dutch probands in the study by Panhuysen and coworkers (9).

Of the 4,097 Chinese parents, 2,208 (53.9%) were women. Sixty (2.7%) of these 2,208 female parents smoked > 5 pack-yr, whereas 1,325 (70.1%) of 1,889 male parents smoked > 5 pack-yr. Among the 6,424 offspring, 3,196 (49.7%) were female. Only 11 (0.3%) of these 3,196 female offspring smoked > 5 pack-yr, whereas 705 (21.8%) of 3,228 male offspring smoked > 5 pack-yr.

Of the 4,097 Chinese parents, 1,527 (37.3%) had increased airway responsiveness to methacholine (PD₂₀  50 mg/ml) (Figure 1). Five hundred eighty-five (38.3%) of these 1,527 parents smoked > 5 pack-yr. Of these 585 smoking parents with increased airway responsiveness, 231 (39.5%) and 344 (58.8%) were classified as having unclassifiable obstructive airway disease and COPD, respectively. Three thousand eighty-two (48%) of 6,424 offspring had increased airway responsiveness to methacholine, and 316 (10.3%) of these offspring were smokers. Of these 316 smoking offspring with increased airway responsiveness, 167 (52.8%) and 140 (44.3%) were classified as having unclassifiable obstructive airway disease and COPD, respectively.

View larger version (18K): [in this window] [in a new window]   Figure 1.   Diagram showing the classification of 4,097 parents after the decision steps. AHR = increased airways responsiveness to methacholine (PD20  50 mg/ml); class 1 = definite asthma; class 2 = probable asthma; class 3 = unclassifiable obstructive airway disease; class 4 = COPD; and class 5 = unaffected (no obstructive airway disease).

Relationship between Airway Responsiveness, Physician Diagnosis of Asthma, Respiratory Symptoms, and "Definite Asthma"

The relationship between airway responsiveness to methacholine, physician diagnosis of asthma, and either two or more respiratory symptoms or a history of recurrent asthma attacks is illustrated in Table 3. Among subjects who had a history of recurrent asthma attacks or two or more respiratory symptoms and increased airway responsiveness (PD₂₀  25 mg/ml), 617 (57.6%) of 1,072 Chinese offspring and 319 (50.6%) of 631 Chinese parents had a physician's diagnosis of asthma. Among subjects who had complete information on airway responsiveness and a significant bronchodilator response, 263 (72.5%) of 363 Chinese offspring and 181 (68.6%) of 264 Chinese parents also had increased airway responsiveness (PD₂₀  25 mg/ml).

Among Chinese subjects classified as having "definite asthma" by the asthma algorithm, 676 (63.5%) of 1,065 offspring and 239 (59.2%) of 404 parents had both increased airway responsiveness to methacholine (PD₂₀  8 mg/ml) and a history of recurrent asthma attacks or two or more respiratory symptoms. Among Chinese subjects who had both increased airway responsiveness to methacholine (PD₂₀  8 mg/ml) and a history of recurrent asthma attacks or two or more respiratory symptoms, 676 (91.5%) of 739 offspring and 239 (56.2%) of 425 parents were classified as having "definite asthma" by the algorithm.

Number of Affected Sibling Pairs Available for Linkage Analysis Using the Asthma Algorithm versus Alternative Definitions of Asthma

After implementing the asthma algorithm in the Chinese offspring, 150 sibling pairs were classified as having "definite asthma." If asthma were defined as a combination of airway hyperresponsiveness (PD₂₀  25 mg/ml of methacholine) and either two or more respiratory symptoms or a history of recurrent asthma attacks, 157 sibling pairs would be classified as having asthma. If asthma were defined as a combination of a physician's diagnosis of asthma, airway hyperresponsiveness to methacholine (PD₂₀  25 mg/ml of methacholine), and either two or more respiratory symptoms or history of recurrent asthma attacks, 95 sibling pairs would be classified as affected.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Phenotype assessment is a crucial issue in gene mapping studies of complex diseases, as misclassification of the trait of interest can lead to spurious results of genetic analysis (14). The lack of a standardized definition of the asthma phenotype has hampered the comparison of genetic studies of asthma conducted to date. Although an algorithm for the diagnosis of asthma was recently proposed for use in family-based studies of the genetics of asthma, it has only been used in a longitudinal study of 92 Western (Dutch) families ascertained on the basis of asthma (physician-diagnosed asthma, respiratory symptoms, and increased airway responsiveness) in a proband 45 yr of age or younger (9). We were interested in using this algorithm in a cross-sectional study of the genetics of asthma conducted among 2,756 non-Western (Chinese) families ascertained on the basis of the presence of both physician-diagnosed asthma and intermittent respiratory symptoms (wheezing and dyspnea) relieved by bronchodilator use in two or more siblings and no more than one parent.

Airway responsiveness to methacholine was not measured in 1,227 (23%) of 5,324 parents and 1,247 (16.3%) of 7,671 offspring in the 2,756 Chinese families participating in this study. Because measuring airway responsiveness is an essential step of the diagnostic algorithm for asthma proposed by Panhuysen and coworkers, 2,474 (19%) of 12,995 subjects would have to be excluded from linkage analysis of asthma, resulting in significant loss of statistical power. In addition, subjects lacking information on airway responsiveness were more likely to have a physician diagnosis of asthma and a lower FEV₁ level than those with complete information. Because FEV₁ level is correlated with asthma severity and airways responsiveness (15), some of the subjects who were unable to perform methacholine challenge testing probably had severe asthma and increased airway responsiveness; excluding these severely affected individuals from linkage analysis of asthma could result in further loss of statistical power.

Because we ascertained families on the basis of a physician diagnosis of asthma and respiratory symptoms in at least two siblings, we predictably observed a higher proportion of subjects with increased airway responsiveness (PD₂₀  50 mg/ml of methacholine) in the current study among Chinese offspring (48%) than in a previous study of Dutch subjects (35.3%), which used a different ascertainment scheme. In spite of these differences, the proportion of subjects with "definite asthma" in Chinese offspring was the same as that observed in Dutch subjects after the application of similar algorithms for the diagnosis of asthma (Table 2) (9). In addition, we observed a lower proportion of subjects with "definite asthma" and a higher proportion of subjects with "COPD" and "unclassifiable obstructive airway disease" in Chinese parents than in Dutch subjects (Table 2) (9). Our findings are likely due to the higher prevalence of smoking among Chinese parents and offspring with increased airway responsiveness than among Dutch subjects with increased airway responsiveness (9). In the absence of radiological and histological data, a significant number of smoking Chinese parents and offspring who truly had asthma may have been misclassified by the algorithm.

Because asthma is a clinical syndrome, it is subject to potential diagnostic bias by physicians (16). In western populations, factors such as age, sex, presence or absence of allergy, presence or absence of reversible airflow obstruction, a family history of the disorder, and the presence or absence of cigarette smoking may influence whether a physician will actually diagnose asthma or another condition (16). Our results in a study of a predominantly rural population in China suggest that the exclusive use of a prior physician's diagnosis of asthma in family-based genetic studies may result in significant misclassification of the asthma phenotype in this population as well. In particular, about a third of the subjects with a physician's diagnosis of asthma did not have increased airway responsiveness (PD₂₀  50 mg/ml; Table 3). Although some of these subjects may have had asthma in the past (19), it is likely that a significant number of subjects without asthma would be classified as truly affected ("false positives"), potentially reducing the statistical power of linkage analysis of asthma. Furthermore, over 40% of the subjects who reported a history of two or more respiratory symptoms or asthma attacks and who had increased airway responsiveness (PD₂₀  25 mg/ml) were not diagnosed as having asthma by a physician. Some of these undiagnosed symptomatic individuals likely truly had asthma, but they would be classified as "unaffected," making them undistinguishable from subjects with incomplete penetrance and potentially distorting the results of genetic analyses.

Many of the subjects classified as having "definite asthma" by the asthma algorithm proposed by Panhuysen and coworkers are likely to truly have asthma and not smoke. Thus, implementation of the asthma algorithm in family-based genetic studies of asthma would work relatively well if linkage analysis were planned in a population with low prevalence of smoking. In addition, the use of "definite asthma" as the asthma phenotype in genetic analyses offers the advantage of reducing the number of subjects with COPD incorrectly classified as subjects with asthma. However, there are several limitations in using this algorithm in family-based genetic studies of obstructive airway diseases. First, implementation of the algorithm in populations with a high prevalence of smoking (such as the Chinese) results in loss of statistical power, as true subjects with asthma who smoke cigarettes cannot be classified as having "definite asthma," except under special circumstances that are absent in most studies of the genetics of asthma, which are cross-sectional. Second, the algorithm implicitly assumes a lack of interaction between smoking and genetic factors in the pathogenesis of asthma. Since long-term cigarette smoking has been found to increase nonspecific airway responsiveness (15, 20), exploring the interaction between genetic factors and cigarette smoking in the pathogenesis of asthma may be of great importance. Third, subjects who have increased airway responsiveness, smoke > 5 pack-yr, and have nonreversible airflow obstruction would be classified as having COPD by the algorithm. Although a majority of subjects with asthma have reversible airflow obstruction, this is certainly not true in all cases (23, 24). In addition, subjects who have mixed (restrictive/obstructive) lung disease could be classified as having COPD in the absence of complete data on pulmonary function and radiological or histological confirmation of this illness. Thus, the algorithm may be suboptimal for use in family-based genetic studies of COPD. Finally, because the first step of the algorithm requires an assessment of airway responsiveness, subjects with severe asthma are likely to be excluded from further study, resulting in additional loss of statistical power.

Reasonable alternatives for defining asthma in family-based genetic studies are (1) a combination of respiratory symptoms and a more stringent definition of increased airway responsiveness (e.g., PD₂₀  25 mg/ml or PD₂₀  16 mg/ml) and (2) a combination of physician-diagnosed asthma, respiratory symptoms, and either increased airway responsiveness (e.g., PD₂₀  25 mg/ml) or a significant response to bronchodilator administration (e.g., an increase in FEV₁%pred  12% following bronchodilator use) (25). The first definition of asthma is relatively simple and likely to yield results similar to those obtained using the asthma algorithm in populations in which cigarette smoking is uncommon (e.g., affected sib-pair linkage analysis in Chinese offspring). However, if used in populations with a high prevalence of smoking, it may lead to significant misclassification of subjects with COPD and increased airway responsiveness (26) as having asthma. The second definition, which has been previously used in linkage analysis of asthma (6), may lead to loss of statistical power. However, when used in populations with high prevalence of smoking, it offers the advantage of reducing the misclassification of subjects with COPD as having asthma while avoiding the exclusion of severely affected subjects in whom airway responsiveness cannot be assessed. In addition, it allows the examination of potential interactions between tobacco use and genetic factors (27) in family-based genetic studies of asthma. Thus, this "stringent" definition of asthma would be preferable if linkage analyses were conducted in populations with a high prevalence of smoking (e.g., linkage analysis of family pedigrees in our Chinese population).

Implementation of the asthma algorithm proposed by Panhuysen and coworkers in a large cross-sectional, family-based study of the genetics of asthma in rural China resulted in the exclusion of subjects with asthma who smoked or who had severe airflow obstruction from linkage analysis, as well as in an inability to examine potential interactions between cigarette smoking and genetic factors in the pathogenesis of asthma. In the absence of a "gold standard" for the diagnosis of asthma, definitions that incorporate a combination of increased airway responsiveness, respiratory symptoms, and a physician's diagnosis of asthma are reasonable. The choice of a particular algorithm for the diagnosis of asthma for a family-based genetic study should be made according to factors such as the prevalence of smoking in the study population. Ultimately, however, it is plausible that the "best definition" of asthma may not coincide with the best definition for mapping genes that contribute to asthma. Genetic studies of intermediate phenotypes related to asthma, which are objectively defined and may be influenced by a smaller number of genes, continue to be of great importance.