INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Bronchial responsiveness (BR) is a central characteristic of asthma. It has been shown to be a risk factor for both the development and outcome of asthma (1). Cross-sectional studies have suggested an age-dependent relationship. In a study of children and adults living in different regions of Australia, the prevalence of BR was highest (16-18%) at age 7-9 yr, lower (7-8%) at age 11-14 yr, and higher again (12-14%) at an adult age (8). Another population study found cross-sectionally a progressive increase in prevalence of BR from age 14-24 to 55- 64 yr, independent of baseline lung size and airway caliber (9). In a study of nine girls (mean age, 13.6 yr) and seven female adults (mean age, 42.4 yr) who were matched for body size, the girls had larger decline in FEV₁ and FEF_25-75% after methacholine challenge (10). This suggests that female children have a larger airway responsiveness to inhaled methacholine than do female adults, and that this difference is not related to baseline lung size and airway caliber. A longitudinal study has confirmed cross-sectional results in also showing a decline in prevalence of BR from childhood into adolescence (11).

Notwithstanding a decrease in prevalence of BR with age in general populations, many patients with asthma retain their BR. It has not yet been elucidated what constitutes a risk for this persistence of BR. This is an important question because a deterioration in BR is associated with an increase in symptomatology and need for asthma medication (9, 12, 13). We examined longitudinal data from histamine challenge tests in a sample of patients with allergic asthma over a period of 30 yr. The first survey started in 1966, when patients were 5-14 yr old, follow-up studies were performed in 1983-1986 (visit 2) and 1995-1996 (visit 3). The primary aims of this study were to determine childhood (age 5-14 yr) and early adulthood (age 22- 32 yr) risk factors for persistence of BR in adulthood (age 32- 42 yr), and to investigate which childhood factors are related to either improvement or worsening of BR from childhood to adulthood. The secondary aim of this study was to determine cross-sectional factors associated with severity of BR in adulthood.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

We reexamined a cohort of patients with asthma, who were admitted to the outpatient clinic of the Pediatric Pulmonary Department of the University Hospital of Groningen (The Netherlands) and who participated in a study performed between 1966 and 1969 (visit 1). The study design has been published previously (1). Follow-up data were collected between 1983 and 1986 (visit 2) and most recently between 1995 and 1996 (visit 3). The study was approved by the medical ethics committee of the University Hospital of Groningen, and all participants signed a written informed consent form.

Patients

In the first survey, 119 children (5-14 yr of age) with asthma (113 allergic to house dust) who were attending the outpatient clinic were included to determine the relationship between house dust allergy and BR by means of a house dust and histamine provocation test (14). The patients lived in the northern region of The Netherlands. They were recruited if they had been diagnosed with asthma by a doctor, if the parents gave informed consent for a 5-d stay in the hospital (Pediatric Clinic of the University Hospital of Groningen), if the children were able to perform technically satisfactory lung function tests, and if their asthma was in a stable condition. Exclusion criteria included the presence of specific respiratory diseases such as cystic fibrosis and tuberculosis (identified by a physician), and the presence of other seriously interfering diseases. All therapy was withheld for at least 24 h before measurements were performed. None of the children used oral corticosteroids on a regular basis. Inhaled corticosteroids and cromoglycate were not available at that time.

Between 1983 and 1986, 101 of the 119 subjects (85%) were reevaluated. Sixteen refused to participate, and 2 were lost to follow-up. In 1995 and 1996, 101 of the 119 subjects (85%) were reinvestigated. All subjects were interviewed and 95 (80%) performed lung function and skin tests at visit 3. Eleven subjects refused, 1 could not participate because of pregnancy, 4 people were lost to follow-up, and 2 people had moved abroad. We analyzed data on 116 children, instead of 119, because data on 2 children were missing and 1 was excluded from the analysis because he appeared to suffer from an ₁-antitrypsin deficiency. In the initial and two follow-up studies measurements were performed to conform to the same protocol.

Questionnaire

The Dutch version of the British Medical Research Council standard questionnaire was used. This version is also comparable to the European Coal and Steel Community questionnaire (15, 16). During the last survey the European Community Respiratory Health Survey questionnaire was also applied (17). In this questionnaire additional questions about pets, medication, passive smoking, and housing are asked. In all three settings, subjects were interviewed by a trained physician.

In the analyses, subjects were regarded as symptomatic when they had symptoms of wheeze and/or asthma attacks. Nonsmokers were those who had never smoked. Ex-smokers were those who had stopped smoking at least 1 mo before examination. Current cigarette smokers were defined as those who smoked one or more cigarettes a day. Pack-years of cigarettes were calculated as the number of packs of cigarettes a day (20 cigarettes per pack) multiplied by the number of years smoking.

Skin Tests, Eosinophils, and Serum IgE

Intracutaneous skin tests were performed with allergen extracts for house dust, grass pollens, tree pollens, weeds, animal dander, and feather (Diephuis Laboratory, Groningen, The Netherlands). Histamine was used as a positive control and phosphate buffer was used as a negative control. At all three visits skin tests were considered positive if the largest diameter of the wheal was at least 5 mm (1). At the last two visits identical allergen extract batches were used. Blood eosinophils were counted in a Bürker counting chamber (Scherf; Cecchinato, Venice, Italy). At visits 2 and 3 serum total and serum-specific IgE (IU/L) were measured by solid-phase immunoassay (Pharmacia IgE EIA; Pharmacia Diagnositics, Uppsala, Sweden). This test was not available at the time of visit 1.

Medication

At visits 2 and 3 inhaled short-acting ₂-agonists, anticholinergics, and cromoglycate were stopped at least 8 h before testing, theophylline and oral antihistaminics were stopped at least 24 h before testing, and long-acting ₂-agonists were stopped at least 48 h before testing. The use of oral and/or inhaled corticosteroids was continued. Individuals had to be in a stable condition without any upper or lower airway infection in the last 2 wk.

Lung Function

Measurements of lung function were carried out with a water-sealed spirometer (Lode spirograph type DL; Lode Instruments, Groningen, The Netherlands). Two valid measurements of FEV₁ and IVC (slow inspiratory vital capacity) were obtained, the highest value being recorded. Predicted values used at visit 1 are those of Zapletal and co-workers (18); the values of the two follow-up visits are according to the ECSC (19). The surveys took place in the same month plus or minus 1 mo.

Histamine Challenge Test

The method of Tiffeneau as modified by de Vries and coworkers (20) and Knol (21) was used in order to meet standardization guidelines (22). During the first survey, subjects inhaled nebulized distilled water from a Wiesbaden Doppel inhalator device after baseline measurements of pulmonary function. Phosphate-buffered saline was used in the second and third surveys. If there was a decrease in FEV₁ of 10% or more the test was stopped. Sequential aerosols of histamine biphosphate in concentrations of 0.25, 0.50, 1, 2, 4, 8, 16, and 32 mg/ml were inhaled for 30 s. Two FEV₁ maneuvers were performed after each challenge, the highest being recorded. The histamine concentration causing a decease in FEV₁ of 10% or more from baseline was taken as the threshold value (PC₁₀). The test was terminated when the threshold value was reached, or when the highest concentration had been given. In 1966, when the first survey took place, the PC₁₀ threshold of de Vries and coworkers (20) was a standard method. To ensure comparability, this method was used again in the later surveys. No test was performed in subjects with an FEV₁ < 1.5 L.

To define BR, a PC₁₀ of  16 mg/ml was used, a value comparable to a PC₂₀ of  8 mg/ml according to the 2-min inhalation method of Hargreave. The definition of the severity of BR in the analyses is expressed as log₂PC₁₀, because this reflects doubling concentrations (22). The change in BR is expressed as the change in log₂PC₁₀.

Statistical Analysis

All analyses were carried out by means of the statistical package SPSS/PC⁺ (version 5.0.2; SPSS, Chicago, IL). Variables were checked for normal distributions. To normalize the distributions, values of serum total IgE, blood eosinophils, and PC₁₀ histamine were log-transformed. In Tables 1234 these variables are presented with geometric means and percent standard deviations. Normally distributed variables are presented with arithmetic means and standard deviations, and not normally distributed variables are shown as medians and ranges.

To investigate the difference between adults with BR and without BR a univariate analysis was performed on those patients with and without a PC₁₀  16 mg/ml at visit 3. Differences between the groups were tested by student t test and ² test. Not normally distributed variables were tested by nonparametric methods (Mann-Whitney U test).

To examine which covariables in childhood and early adulthood are associated with the level of BR in adulthood, a multiple linear regression analysis was carried out on the log₂PC₁₀ at visit 3. Covariables included in this model were as follows: age (yr) at visit 1; sex (male, 1; female, 0); FEV₁, height adjusted, at visit 1 (dl/m²); natural logarithm eosinophils (10⁶/L) at visit 1; having pets at visit 1; passive smoking at visit 1; log₂PC₁₀ at visit 1; log serum total IgE at visit 2 (IU/L); smoking habits at visit 3; and use of oral and/or inhaled corticosteroids at visit 3. Furthermore, changes in age, FEV₁ (height adjusted), and eosinophils during follow-up were also included. Because the change in FEV₁ and eosinophils from childhood (visit 1) to early adulthood (visit 2) may have a different influence on the outcome than does the change from early adulthood (visit 3), changes during both periods were included in the analysis. Owing to the reported differences in IgE levels between smokers and nonsmokers, which may imply that a high level in a smoking person will mean something different than in a nonsmoking subject, the product terms of log serum total IgE and smoking habits were included in the analyses.

Another multiple linear regression analysis was carried out on the change in log₂PC₁₀ from visit 1 to visit 3 in order to investigate which covariables are associated with an increase or decrease in BR from childhood to adulthood. The same covariables as in the previous analysis were included in the model, except for the log₂PC₁₀ at visit 1. Including this covariable in the model would lead to regression to the mean, and because of the interdependence of all other covariables with the log₂PC₁₀ at visit 1 and with each other, this could lead to biased estimates (23, 24). To avoid the occurrence of this bias, the residuals derived from a regression of BR at visit 1 on the covariables of interest were used as the initial-level variable instead of an absolute level of BR. By the use of these residuals of BR, the initial-level variable is determined only by random errors and possibly by other factors not in the model. This procedure does not prevent bias in the estimated coefficient of the initial-level variable itself, caused by the "regression to the mean" phenomenon, but it does not lead to bias in the estimates of the effects of the other covariables.

A cross-sectional multiple linear regression analysis of the log₂PC₁₀ at visit 3 was carried out to examine which covariables are associated with the level of BR in adulthood, including only covariables of the third visit. These covariables were as follows: age (yr); sex (male, 1; female, 0); FEV₁, height adjusted (dl/m²); lymph node eosinophils (10⁶/L); log serum total IgE at visit 3 (IU/L); smoking habits; having pets; having symptoms of asthma (wheeze and/or asthma attacks); passive smoking; and use of oral and/or inhaled corticosteroids. Again, the product terms of log serum total IgE and smoking habits were included to control for a possible different effect of IgE in smoking and nonsmoking persons.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Comparison of Participants and Nonparticipants

The characteristics of subjects studied and lost to follow-up are summarized in Table 1. More females than males were nonparticipants. At visit 1, no significant differences existed between these groups in terms of mean age, lung function, number of eosinophils, and percentage of children with a PC₁₀ histamine  16 mg/ml, atopy, and symptoms.

Comparison of Level of BR at Visits 1, 2, and 3

Figure 1 shows the distribution of BR from childhood to adulthood. The percentage of subjects with a PC₁₀  2 mg/ml decreased from childhood to early adulthood and increased from early adulthood to adulthood. This pattern is also seen for subjects with a PC₁₀ between 2 and 16 mg/ml. The percentage of subjects with a PC₁₀ > 16 mg/ml increased from childhood to early adulthood and decreased from early adulthood to adulthood.

View larger version (28K): [in this window] [in a new window]   Figure 1.   Level of bronchial responsiveness to histamine in childhood and adulthood, measured in a longitudinal study of 119 allergic asthmatic children. Visit 1, age 5-14 yr; visit 2, age 22-32 yr; visit 3, age 32-42 yr.

Univariate Comparison of Responsive Subjects versus Nonresponsive Subjects at Visit 3

Subjects with a PC₁₀  16 mg/ml at visit 3 had a significantly lower level of lung function, expressed as FEV₁, FEV₁%pred, and FEV₁%VC at all three visits (Table 2). They also had a significantly higher level of serum IgE at visit 2, significantly more eosinophils at visit 2, and more symptoms of wheeze and/or asthma attacks at visits 2 and 3. They used corticosteroids more frequently both at visit 2 and visit 3, the later being significant (Table 2). In the group of subjects with a PC₁₀ histamine  16 mg/ml at visit 3, the proportion of subjects with BR at visit 2 was significantly higher in comparison with the group with a PC₁₀ > 16 mg/ml at visit 3. There were no significant differences in terms of age, sex, and smoking habits between both groups, altogether there were fewer smokers in the responsive group. These analyses were performed in 91 subjects because a histamine challenge was not performed in 4 subjects owing to low lung function and because 6 subjects were interviewed only (they refused further tests).

Childhood Risk Factors for the Severity of Bronchial Responsiveness in Adulthood

A higher level of lung function in childhood is an important predictor of less severe BR at age 32-42 yr (Table 3). Furthermore, larger increases in the level of lung function, both during childhood and adulthood, were independently associated with less severe BR at visit 3. The longer the time interval between the first and last survey, the less severe was the BR at age 32-42 yr. Subjects who had pets in childhood had less severe BR at visit 3. Only 63 individuals were included in this analysis, because the number of eosinophils and/or lung function measurements were not available at all three visits, and measurements of serum total IgE at visit 2 were missing for 28 subjects.

Childhood Risk Factors for Changes in Bronchial Responsiveness over Time

Subjects with a low level of lung function in childhood and smaller increase in level of lung function with age, not only between early adulthood and adulthood but also between childhood and early adulthood, had a significant increase in BR from childhood to adulthood (Table 3). A longer time interval between visits and having pets in childhood were significantly associated with a reduction in the severity of BR. Compared with nonsmoking subjects, current and ex-smokers had a larger increase (or a smaller decrease) in BR. This was significant only for the ex-smokers. The effect of IgE is different for nonsmokers, ex-smokers, and current smokers: in nonsmokers higher IgE levels are related to an increase in BR, in ex-smokers higher IgE levels are related to a decrease in BR, and in current smokers IgE levels are not related to the change in BR from childhood to adulthood. The reduced sample size in this analysis is due to missing values on eosinophils, lung function, and/or serum total IgE (see above).

Cross-sectional Analysis of the Severity of Bronchial Responsiveness at Visit 3

Table 4 shows that a lower height-adjusted FEV₁ and the presence of symptoms of wheezing and/or asthma attacks at visit 3 were significantly associated with more severe BR. Older subjects and subjects who had pets had less severe BR (p = 0.076 and p = 0.085, respectively). IgE was significantly related to more severe BR in nonsmokers. In ex- and current smokers IgE levels were not related to the severity of BR. In this analysis 79 subjects were included. Data concerning the number of eosinophils and/or on serum total IgE at visit 3 were missing for the other 16 subjects.

In the subgroup that was hyperresponsive (PC₁₀  16 mg/ml) at visit 3, subjects with good lung function, subjects who were exposed to environmental tobacco smoke, and subjects who had pets had less severe BR. Again, higher IgE levels were related to more severe BR only in nonsmoking subjects (results not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This longitudinal study of children with asthma over 30 yr of follow-up showed that a better lung function at age 5-14 yr and a larger increase in lung function over time, as well as having pets in childhood and longer time interval between visit 1 and visit 3, were associated with less severe BR at age 32-42 yr. The same factors were significantly associated with improvement of BR from visit 1 to visit 3. Furthermore, smokers and ex-smokers had a larger increase in BR compared with nonsmoking subjects and higher IgE levels in nonsmokers were related to an increase in BR whereas in ex-smokers higher IgE levels were related to a decrease in BR. In the cross-sectional analysis a lower level of lung function, younger age, and having symptoms at visit 3 were significantly associated with more severe BR at that time, whereas having pets at visit 3 was associated with less severe BR. IgE levels were related to more severe BR only in the nonsmoking subjects.

We found that a low level of lung function in childhood is associated with more severe BR at age 32-42 yr. This finding reinforces the results of the second survey of this study, performed in 1983-1986, when the subjects were 22-32 yr of age (1). A population-based sample of children, 5-9 yr old, who were monitored for up to 12 yr showed that persistent BR was associated with low levels of FEF_25-75% at the first survey (25). Another important finding in our study is that an improvement in FEV₁ between visit 1 and visit 2 is significantly associated with the level of BR in adulthood. Thus, the level of lung function in early life and the growth in lung size in childhood are both independently related to the outcome of asthma at age 32-42 yr as assessed by BR. The underlying pathophysiology of this observation is still unclear. One interpretation is that low lung function is caused by more severe BR. However, in the present population, BR in childhood was not associated with the level of lung function in childhood. Moreover, the level of BR in childhood was not related to the level of BR in adulthood, even after exclusion of the lung function variables from the model (results not shown). This indicates that low lung function in childhood is related to other factors than BR. Because the results suggest that low lung function in childhood is related to the severity of BR in adulthood and to the course of BR from childhood to adulthood, it is of great importance for future studies to assess which factors are related to low lung function in childhood and to investigate which interventions and/or treatment might change the course of FEV₁ into adulthood, as this might theoretically improve BR in adulthood.

We must realize that this is a selected group of children with asthma, and data may not be applicable to all patients with asthma in the population; the sample is unique because objective data are available from the onset of the study.

To our knowledge, this is the first study that has investigated both cross-sectionally and longitudinally the association of having pets in childhood and adulthood with the severity of BR at age 32-42 yr. Eighty-two percent of our children with allergic asthma did have pets, as did 56% of the adults. We expected that having pets would be a risk factor for more severe BR in childhood and adulthood, because Meijer and coworkers found that this contributed to an increased circadian peak expiratory flow (PEF) amplitude (26). To our surprise we found in both the cross-sectional and longitudinal analysis that subjects with pets had significantly less severe BR. In childhood, however, no differences in BR were found between children with and without pets. The most likely explanation, in our opinion, is that subjects without pets had more severe airway wall inflammation imposed on their atopic constitution, and therefore they were unable to hold pets. This was indirectly supported by the finding that adults with asthma and without pets had a higher prevalence of eosinophilia at visit 3, which in itself was not significantly associated with more severe BR at that time (data not shown). In childhood, no differences were found between children with pets and children without pets. We cannot exclude other influential factors that we did not measure, e.g., children with pets may have lived in a better home environment without carpets.

Cross-sectional as well as longitudinal studies have found atopy to be one of the most important risk factors for BR (27, 28). In our univariate analysis on the presence (PC₁₀  16 mg/ml) or absence (PC₁₀ > 16 mg/ml) of BR at visit 3, total serum IgE at visit 2 was significantly higher in the hyperresponsive group. Sears and colleagues (27) found in a cross-sectional analysis of a birth cohort of 562 children, 11 yr of age, that the prevalence of BR significantly increased with higher total serum IgE levels. Burrows and co-workers (28) found in a longitudinal study of children (between 9 and 15 yr of age) a close relationship between high serum IgE and the severity and persistence of BR. Because our population is somewhat older, we must deal with the possible modification of IgE by smoking. In both the longitudinal and the cross-sectional analyses high serum IgE is related to more severe BR only in nonsmokers. In smokers IgE levels are not associated with BR whereas in ex-smokers IgE is associated with the change in BR from childhood to adulthood in a positive way, i.e., the higher the IgE levels, the less severe BR becomes. Our results do not contradict the results of Sears and Burrows because their populations are most likely nonsmokers, given their age (27, 28). The effect of smoking habits on IgE levels is a consistent finding (29, 30). However, the mechanisms by which tobacco smoke affects serum total IgE is not known. Future studies that investigate the mechanisms of this elevated IgE level in smoking persons are needed.

In our study we found an increase in BR over time in both smokers and ex-smokers. Effects of smoking on BR and atopy have been explained by increased bronchial epithelial permeability, which is increased in smokers. This may allow easier and greater access of allergens to subepithelial inflammatory cells (31). Our finding is in agreement with the previous findings reported in the literature. However, in our study higher IgE levels in smokers and ex-smokers were associated with a decrease in BR over time. Similar findings were shown in subjects with chronic obstructive pulmonary disease, in which a higher IgE level was associated with a slower decline in PC₂₀, especially in smokers (32). These observations need further study.

Many investigations have shown sex-based differences in the prevalence of asthma and atopy. The incidence of clinical asthma is 1.5 to 2 times higher in boys than in girls. An explanation for this sex-based difference could be that the prevalence of atopy is higher in boys (33), because this is known to be a risk factor for the development of asthma. A population study found a higher prevalence of BR in boys (33); another study found that male sex increased the odds ratio for BR at any age from 9 to 15 yr (34). Studies assessing the effect of sex on the severity of BR in adulthood found either no sex-based difference or more severe BR in females (35). We did not find a sex-based difference in the severity of BR in adulthood, or in the change in the level of BR from childhood to adulthood. This might be due to the relatively low number of females in our study, the high prevalence of atopy, and the relatively small sample size.

Current smoking at visit 3 had no influence on the severity and changes in BR from childhood to adulthood. Ex-smokers showed a significant increase in BR from childhood to adulthood. Smoking has been considered in the literature as a risk factor for severity of BR; the opposite has also been frequently reported. This may be the result of the "healthy smoker effect," in that those with more susceptible airways do not take up smoking or quit at an early age. To test the hypothesis that subjects with better lung function and less severe BR take up smoking and continue smoking, we reanalyzed our data. We could not find significant differences with respect to sex, lung function, and BR in childhood between subjects who started smoking during follow-up and those who never smoked. However, significantly more males and subjects with a good lung function in childhood continued smoking into adulthood (data not shown). Thus, the "healthy smoker hypothesis" is supported by our longitudinal data, in that individuals with asthma with better lung function persist in their smoking habits.

In summary, this study showed that a lower FEV₁ in childhood and less improvement in FEV₁ over time was associated with more severe BR in adulthood. This stresses the need for studies on early intervention to optimize lung function and for further research to assess which factors influence airways and lung growth. Our data suggest that the healthy smoker hypothesis is valid. The relevance of high IgE level for progression of BR is not fully elucidated and clearly needs further investigation. Finally, we have shown that not only the prevalence but also the severity decrease from childhood to early adulthood and thereafter increase again.