INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX: THE EGEA COOPERATIVE... REFERENCES

Keywords: asthma; severity; genetics; familial resemblance

Asthma is a disease of very variable severity and asthma severity in an individual depends upon a combination of inflammatory and structural changes in the airways, environmental exposures, and psychological factors (1). This has important consequences on morbidity, mortality, and health care utilization. The definition of asthma severity based on several dimensions is still a matter of debate. Items used in the clinical and epidemiological literature cover clinical, functional, health care utilization, and therapeutic dimensions, which concern both asthma control and severity (2).

Genetic predisposition for asthma is well known (3, 4), and many candidate genes or genetic regions have already been reported by linkage or association studies as being linked or associated with the susceptibility to asthma (5). However, research focused on genetic predisposition for severe asthma is scanty. A study conducted among 51 children with severe wheeze in the first 2 yr of life showed that a family history of asthma was related to the asthma attack severity score and the interval between symptoms, but the severity of asthma in the parents was not known (6). Results of a twin study showed significant correlation in asthma severity in monozygotic pairs with asthma, but not in dizygotic pairs, supporting the hypothesis of genetic factors involved in asthma severity (7).

Genetic studies have reported many associations between markers of asthma severity and some polymorphisms of candidate genes, including the ₂-adrenergic receptor gene (8), the interleukin-4 (IL-4) receptor gene (12, 13), and the IL-4 gene promoter (14). Other candidate genes, such as those coding for tumor necrosis factor- (TNF-), high-affinity immunoglobulin E (IgE) receptor, and angiotensin converting enzyme (ACE), because of their function, have also been proposed as contributing to asthma severity. However, association between polymorphisms of these genes and asthma severity has not been demonstrated until now (15, 16).

The aim of the present study was to analyze the familial resemblance of asthma severity using the data from EGEA (Epidemiological study on the Genetics and Environment of Asthma, bronchial hyperresponsiveness and atopy), a case- control and a family study. Specific objectives were to study whether the risks of asthma and of severe asthma in first- degree relatives of subjects with asthma were related to the severity of the index case.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX: THE EGEA COOPERATIVE... REFERENCES

The EGEA study combines a case-control study and a family study (17). Probands (subjects with asthma and control subjects), either children aged 7 to 15 yr or adults aged 16 to 64 yr old, fulfilled inclusion criteria including familial structure (one sibling older than 7 yr and two parents younger then 70 yr or a spouse younger than 70 yr and an offspring older than 7 yr). Control subjects were recruited through electoral rolls, surgery departments, and from a check-up center. Cases with asthma were recruited in chest clinics and defined through a self-completed questionnaire with positive answers to four standardized questions: (1) Have you ever had attacks of breathlessness at rest with wheezing? (2) Have you ever had asthma attacks? (3) Was this diagnosis confirmed by a physician? (4) Have you had asthma attacks in the last 12 mo? In relatives, asthma status was based on a positive answer to the first or the second question.

The EGEA population comprises 348 nuclear families totaling 1,847 subjects: 348 cases, 1,083 relatives with asthma (including 123 spouse), and 416 control subjects. Among the 1,116 family members (parents, siblings, and offspring) invited to participate, 475 relatives of adult probands and 485 relatives of pediatric probands were examined resulting in participation rates of 77.0% and 97.2%, respectively. There was no participation bias in first-degree relatives of adult cases according to their reported asthma status (participation rate 77.8% in subjects without asthma versus 76.7% in subjects without asthma, p = 0.82) and according to the asthma severity of the case (80.3% and 78.0% in relatives of subjects with mild and severe asthma, respectively, p = 0.51). The population studied in the present analysis included 944 subjects: 348 cases with asthma (213 adults, 135 children), 357 control subjects without asthma (266 adults, 91 children), and 239 relatives with asthma: 105 relatives of adult cases (73.3% were adults) and 134 relatives of pediatric cases (60.4% were adults). All subjects answered a questionnaire about their own respiratory status including detailed questions on asthma if applicable and performed a pulmonary function test to measure FEV₁. Probands were interviewed on the medical and smoking history of all their first-degree relatives (alive or not), allowing a record of asthma (ever) and smoking status (never, ex, and current smokers) of 1,668 relatives of cases with asthma and 1,799 relatives of control subjects without asthma.

To assess the validity of the report of asthma in relatives, we compared the information on asthma status of the parents obtained indirectly by the declaration of the case and directly by parents themselves when they were interviewed. Based on data available for 65 mothers and 67 fathers, kappas were 0.87 and 0.86 for the presence of asthma in fathers and mothers, respectively. Asthma status of 37 mothers and 37 fathers was also assessed by questioning one sibling without asthma, and kappas were 0.89 and 1 for the presence of asthma in fathers and mothers, respectively. Therefore there was no apparent bias for the report of parental asthma according to the presence of asthma of the respondent.

The choice of items used to assess asthma severity was based on international guidelines (18), on their availability in the study, and on their applicability to the whole population of children and adults with asthma of the study. It allowed investigation of three dimensions of severity: (1) a clinical composite score (three items) ranging from 0 to 7 based on asthma attack frequency in the past 12 mo (from 0 for less than once a month to 3 for at least once a day), persisting symptoms between attacks (from 0 for none to 3 for limiting activities), and hospitalization for asthma in the past 12 mo (0 for none, 1 for more than 0); (2) a measure of respiratory function: FEV₁ as % predicted; (3) the report of inhaled corticosteroid use in the past 12 mo. Because risk factors for asthma severity may be different in children and adults, the analyses were done separately in families of children and adult cases, but it should be noted that family members (siblings, parents, and offspring) comprised children and adults.

Statistical Analysis

Three questions were addressed in the present investigation: (1) Was there a familial resemblance of asthma? (2) Was the risk of asthma in relatives of cases associated with asthma severity of the case? (3) Was there a familial resemblance of asthma severity?

Analyses were performed with the statistical software package SAS. Two statistical methods were used to assess the familial resemblance. The first one is based on regression models where the status (asthma or asthma severity) of the relatives is the response variable and the status of the probands is the explanatory variable. For the analysis of the risk of asthma in relatives according to asthma and asthma severity in probands, odds ratios (OR) with their 95% confidence intervals (CI) were estimated taking into account familial dependence between observations with generalized estimated equations (GEE) using the GENMOD procedure (19). For the analysis of familial resemblance of asthma severity, the same regression models were used with the MIXED procedure for continuous variables and the GENMOD procedure for binary variables, with results expressed with estimated regression parameters.

The second method is based on calculation of the intraclass (intrafamilial) correlation coefficient (ICC) among all family members (cases and relatives) (20). These analyses were limited to families with at least one relative with asthma. Fisher ICC and Fisher tests derived from ANOVA procedure were given.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX: THE EGEA COOPERATIVE... REFERENCES

The characteristics of the population studied are described in Table 1. Familial resemblance of asthma was studied separately in relatives of adult probands and in relatives of pediatric probands (Table 2). The prevalences of asthma in relatives of adult and pediatric cases were significantly higher than in relatives of control subjects without asthma, with OR of 3.4 (95% CI 2.5- 4.7) and 4.5 (2.6-8.1), respectively. Analysis according to the familial status of relatives (parents, siblings, and offspring) showed similar results in each group (Table 2). Mothers did not confer an increased risk of asthma compared with fathers (data not shown). ICC values for all family members were 0.076 (p < 0.0001) and 0.129 (p < 0.0001) in families of adult and pediatric probands, respectively. Tobacco smoking in relatives was studied as a potential confounding factor, as a familial tendency in smoking was observed in both control subjects and cases. In relatives of adult probands, tobacco smoking was not a confounder. In relatives of pediatric probands, relatives with asthma stopped smoking more often than relatives without asthma (68.1% versus 40.4%, p = 0.001) and relatives of cases stopped smoking more often than relatives of control subjects (50.3% versus 38.0%, p = 0.04). Taking into account the smoking status of the relatives in the analysis, the increased risk of asthma in families recruited with a pediatric proband was slightly decreased to OR 3.7 (95% CI 2.3-6.1).

Characteristics of asthma severity in cases are described in Table 3. According to the asthma severity criteria used, children had milder asthma than adults both in cases and relatives. Moreover relatives with asthma had a milder asthma than cases (Table 3). Dividing the clinical score into two classes ( 1 versus > 1), the mean age in relatives of cases with mild asthma was similar to relatives of cases with severe asthma (27.4 versus 25.7 and 26.9 versus 30.3 in families of adult and pediatric cases, respectively).

We also studied whether familial risk of asthma was associated with asthma severity of the case. Both in relatives of adult and pediatric cases, no significant associations were found between asthma prevalence in relatives and asthma severity of the case (Table 4). Adjustment for smoking (smoking status of the relative or environmental tobacco smoke exposure from parents or spouse) did not change the results.

Familial resemblance of asthma severity is presented in Table 5. There was a significant familial aggregation of the clinical severity score and of FEV₁ in families of adult and pediatric probands, either analyzed by regression models or by calculation of ICC. Relatives with asthma tended to be treated more often with inhaled corticosteroids if cases were treated (OR = 2.15 [0.56-8.27] and 2.48 [0.77-8.02] for relatives of adults and children), but the intrafamilial correlation coefficients did not show any tendency for a familial resemblance of the use of inhaled corticosteroids. Smoking (smoking status of the relative or exposure to environmental tobacco smoke from parents or spouse) was not a confounder in these analyses.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX: THE EGEA COOPERATIVE... REFERENCES

In the present study asthma occurrence in the families of cases with asthma was independent of the severity of the case, but results are in favor of a familial resemblance of asthma severity when asthma occurs.

There were potential biases in the study. We checked that there were no biases associated with the report of asthma in relatives. The participation of first-degree relatives of cases was high and not influenced by their own asthma status or the severity of the case. It was not possible to assess whether the participation of relatives with asthma might be biased by their own asthma severity. Nevertheless if there was such an association, it could change only the power but not the magnitude of the relationship between the severity of asthma in cases and relatives.

Results from the two statistical methods performed (regression model and intrafamilial correlation coefficient) are consistent, except for a discordance regarding the familial resemblance of inhaled corticosteroid treatment. The low ICC for inhaled corticosteroid treatment may be the result of the study design. As cases were recruited in chest clinics a much larger proportion of cases with asthma than of relatives was treated with inhaled corticosteroids resulting in a greater aggregation among cases (between families) than within families. Because the magnitude of the ICC and not of the OR is dependent on the prevalence of the character studied, regression analysis may be more appropriate than the ICC to assess the familial resemblance in this situation. This may also explain why, inversely to the OR, the magnitude of the ICC calculated for the resemblance of asthma was low compared with the ICC for the resemblance of asthma severity.

As expected, we found a familial resemblance of asthma and the magnitude of the increased risk was consistent with the results of previous studies (3). However, we did not find an increased familial risk of asthma according to asthma severity. Few studies have looked at this association. The severity but not the frequency of asthma attacks has been significantly related to a family history of asthma (6), a result at variance with our observations, but that study was based on infant cases, for whom different risk factors may be involved. In a study of young adults with asthma, risks for intubation, although increased in relation to a parental history of asthma and allergy, did not reach significance and were of a smaller magnitude than other significant covariates (21). Even if our analysis has a good statistical power, more studies are needed to reach definite conclusions.

As regards clinical severity scores and FEV₁, we showed a familial resemblance of asthma severity and the results are consistent using different statistical analyses. Other clinical studies are in agreement with a familial aggregation in asthma severity based in particular on twin studies (7). Our results obtained among subjects with asthma confirmed the familial resemblance in FEV₁ in the general population (22, 23). Whereas general segregation analyses suggest a polygenic inheritance, analyses conducted in families with asthma have suggested the possible effect of a recessive major gene (24). Factors explaining familial resemblance in FEV₁ among families with asthma may therefore provide clues in understanding the severity of asthma. The level of airway hyperresponsiveness is related to asthma severity and Gray and coworkers reported a 2-fold increase in the risk for a child to have airway hyperresponsiveness if either parent had the same condition. In addition, 26% of the variation in the provocative dose of an agonist causing a 20% fall in FEV%₁ (PD₂₀) in the child was explained by the PD₂₀ value in the parent (25).

The familial resemblance of a trait can be due to shared environment as well as shared genes. A genetic contribution to the severity of asthma is strongly suggested by previous genetic studies reporting association with candidate genes. There seems to be associations between asthma severity and polymorphisms in the ₂-adrenoceptor expression (₂-AR). The polymorphism of Glu-27 ₂-AR has been shown to be associated with increased airway hyperresponsiveness (9), and some aspects of asthma severity (amount of daily inhaled corticosteroids or baseline FEV₁) but not with the risk of fatal/near-fatal asthma (11). Another mutation, the Gly-16 polymorphism of the ₂-AR, was associated with more severe asthma as indicated by the use of corticosteroids and immunization therapy (8), nocturnal asthma (10), and bronchodilator desensitization (26). More recently, this polymorphism has also been shown to be specifically associated with asthma severity defined by hospital admission and use of corticosteroids (27). Genetic variants in the interleukin-4 (IL-4) intron 2 region have been related to asthma severity characterized by symptoms and FEV₁ while variants in the IL-4 gene promoter have been shown to be associated with lower FEV₁ in asthma and life-threatening asthma (14, 15). Polymorphisms of the R576 IL-4 receptor  were also reported to be associated with lower FEV₁ in asthma (12, 15). Finally, a mutation of the platelet-activating factor (PAF) acetylhydrolase gene was shown to be associated with asthma and the severity of asthma in a Japanese population (28), but another study did not confirm such an association (29). It should be noted that heterogeneous phenotypes of asthma severity may explain discordant results.

Familial resemblance of asthma severity may also be the result of exposures to shared environmental factors associated with asthma severity. In the EGEA study, active smoking was a risk factor for asthma severity (30). Environmental tobacco smoke increases asthma severity in children (31). Nevertheless, tobacco smoking was studied as a potential confounding factor for all analyses and was taken into account in estimated OR when needed. Therefore tobacco smoking is unlikely to explain the observed results.

Exposure to indoor allergens may be another shared environmental factor that could explain familial resemblance of asthma severity. The level of house dust mite exposure has been shown to be associated with asthma severity as assessed by treatment use (32), wide peak flow variability, and asthma symptoms (33). Risk of hospitalization for asthma and numbers of days of wheezing were increased in children who were both allergic to cockroach allergen and exposed to high levels of this allergen (34). However, studies in a population of Danish twin pairs showed that apart from a major influence of genetic factors, part of the liability to asthma was due to specific and not shared environmental factors (35).

Our results are in favor of specific familial risk factors being involved in the severity of asthma independently of asthma occurrence. Further analyses using information on both genetic factors, as candidate gene polymorphisms, and environmental factors would be helpful to disentangle the underlying mechanisms of asthma severity.

	APPENDIX: THE EGEA COOPERATIVE GROUP

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION APPENDIX: THE EGEA COOPERATIVE... REFERENCES

Respiratory epidemiology: I. Annesi-Maesano, F. Kauffmann (coordinator), M. P. Oryszczyn (INSERM U472, Villejuif); F. Neukirch, M. Korobaeff (INSERM U408, Paris).

Genetics: M. H. Dizier (INSERM U155, Paris, then U535, Kremlin-Bicêtre), J. Feingold (INSERM U155, Paris), F. Demenais (INSERM U358, Paris, then INSERM EMI 00-06, Evry), M. Lathrop (INSERM U358, then WTCHG, Oxford, now at the CNG, Evry).

Clinical centers: Grenoble: I. Pin, C. Pison; Lyon: D. Ecochard (deceased), F. Gormand, Y. Pacheco; Marseille: D. Charpin, D. Vervloet; Montpellier: J. Bousquet; Paris Cochin: A. Lockhart, R. Matran (now in Lille); Paris Necker: E. Paty, P. Scheinmann; Paris-Trousseau: A. Grimfeld.

Data management: J. Hochez (INSERM U 155), N. Le Moual (INSERM U472).