SEARCH

SEARCH BY CITATION

Keywords:

  • asthma;
  • environment correlation;
  • genetic correlation;
  • modelling;
  • multivariate

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Background:  Intermediate phenotypes are often measured as a proxy for asthma. It is largely unclear to what extent the same set of environmental or genetic factors regulate these traits.

Objective:  Estimate the environmental and genetic correlations between self-reported and clinical asthma traits.

Methods:  A total of 3073 subjects from 802 families were ascertained through a twin proband. Traits measured included self-reported asthma, airway histamine responsiveness (AHR), skin prick response to common allergens including house dust mite (Dermatophagoides pteronyssinus [D. pter]), baseline lung function, total serum immunoglobulin E (IgE) and eosinophilia. Bivariate and multivariate analyses of eight traits were performed with adjustment for ascertainment and significant covariates.

Results:  Overall 2716 participants completed an asthma questionnaire and 2087 were clinically tested, including 1289 self-reported asthmatics (92% previously diagnosed by a doctor). Asthma, AHR, markers of allergic sensitization and eosinophilia had significant environmental correlations with each other (range: 0.23–0.89). Baseline forced expiratory volume in 1 s (FEV1) showed low environmental correlations with most traits. Fewer genetic correlations were significantly different from zero. Phenotypes with greatest genetic similarity were asthma and atopy (0.46), IgE and eosinophilia (0.44), AHR and D. pter (0.43) and AHR and airway obstruction (−0.43). Traits with greatest genetic dissimilarity were FEV1 and atopy (0.05), airway obstruction and IgE (0.07) and FEV1 and D. pter (0.11).

Conclusion:  These results suggest that the same set of environmental factors regulates the variation of many asthma traits. In addition, although most traits are regulated to great extent by specific genetic factors, there is still some degree of genetic overlap that could be exploited by multivariate linkage approaches.

There has been a strong interest in searching for susceptibility genes for asthma, driven by the prospect of better disease prevention, diagnosis and treatment. A common approach used for this is to measure different asthma-related traits, such as atopy, airway responsiveness and eosinophilia and subsequently test each trait individually for linkage to genetic markers genotyped throughout the genome. Regions of significant linkage are then followed up with higher resolution association studies. However, the initial stage of univariate linkage analysis has a number of caveats, including the lack of efficient methods to correct for multiple trait testing and the loss in power arising from the disregard of the additional information provided by multivariate data sets. This may result in both false-positive and false-negative linkage results. One alternative is to perform multivariate linkage analysis of related traits, a concept that has long been proposed (1–3) but that has only recently been implemented for the dissection of complex traits (4–6). Multivariate analysis not only avoids an inflated type I error as a result of multiple trait testing, as it can also provide greater statistical power to detect quantitative trait loci (QTL) than univariate analysis (7, 8).

Multivariate linkage analysis is appropriate to analyse traits that have common QTL. However, it has been demonstrated that multivariate analysis of traits that have a high genetic correlation is unlikely to be more powerful than univariate analysis (8). Thus, in order to guide the choice of traits for multivariate linkage, it is important not only to identify traits that have a common genetic aetiology but also to estimate the degree of genetic overlap.

Although a number of studies have addressed the first issue (9–12), few have provided estimates of the genetic correlation between different asthma traits. Using a regression-based method, Palmer et al. (13) reported that the genetic overlap between asthma and baseline lung function was <10%, whereas the overlap between airway responsiveness and total serum immunoglobulin E (IgE) was <30% (14). Using bivariate variance components, Nystad et al. (15) estimated the genetic correlation between asthma and wheeze at 0.76.

In the present study, we describe a new sample of 802 Australian asthmatic families ascertained through a twin proband. The aim of this study was to use this large data set to estimate the genetic and environmental correlations between self-reported and clinical asthma traits through variance components methods.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Ascertainment of families

Two rounds of ascertainment were used to select families registered on the volunteer Australian Twin Registry (ATR) with at least one twin who had reported ever wheezing in his/her lifetime in previous studies (Fig. 1). The first round of ascertainment (1990–1992) has been described previously (11). As part of this first round, 863 twins from 443 families completed a detailed asthma questionnaire, attended clinical testing and donated blood for DNA extraction. The second round of ascertainment (1996–1998) involved two different selection approaches: first, extension of 164 round 1 families to include additional affected sibs or affected children of twins. Secondly, selection of 690 new families from five other ATR-based studies with at least one dizygotic (DZ) twin proband and any additional affected sibs, or at least one monozygotic (MZ) twin proband and additional affected sibs. As with round 1, 3824 participants from these 854 families were invited to complete a similar asthma and allergy questionnaire, donate blood or a buccal swab for DNA extraction and, if under the age of 50, to attend full clinical testing. Questionnaires were sent to 2270 participants of round 2 and 2100 (93%) from 511 families returned it completed. The outstanding 1554 participants who were not sent questionnaires were untraceable, deceased, overseas or unable to attend clinical testing. Of the 2100 participants with a completed questionnaire, 1965 (94%) also donated blood for genetic and/or haematological analyses and 1435 (68%) were clinically tested. One family consisted of one father and no other relatives and, thus, was dropped from further analysis. When the samples from both rounds were combined, the overall sample consisted of 3073 participants from 802 families: 974 participated exclusively in round 1 (292 families), 1846 exclusively in round 2 (359 families) and 253 (151 families) in both rounds. Overall, 2716 participants completed one or both asthma questionnaires, 2852 donated blood and 2087 were clinically tested. Complete data were available for 1969 participants.

image

Figure 1. Ascertainment scheme used to select 802 families registered with the Australian Twin Registry (ATR) with at least one asthmatic twin proband. ATR families are representative of the Australian population for a number of traits, including asthma (28).

Download figure to PowerPoint

Evaluation of phenotypes

The following procedures were performed with the appropriate ethical approval of all institutions involved.

Questionnaire administration.  The questionnaire used in round 1 was designed to validate the diagnosis of asthma and to obtain data on respiratory symptoms, environmental exposures relevant to respiratory disease and family history of asthma (11). The round 2 questionnaire included the key items of the first questionnaire plus a section for the diagnosis of chronic bronchitis (16). There were two versions of this questionnaire: one for adults (aged >16 years) and one to be completed by parents of children (aged ≤16 years). A participant was considered affected for asthma if he or she answered ‘Yes’ or ‘Yes, told to me by a doctor’ to the question ‘Have you ever suffered from asthma or wheezing'? and ‘Yes’ to the question ‘Have you ever taken any medicine for asthma or wheezing'? A participant was considered affected for doctor asthma if he or she answered respectively ‘Yes, told to me by a doctor’ and ‘Yes’ to the previous two questions. Asthma and doctor asthma were highly correlated (tetrachoric correlation 0.99). For this reason, familial correlations, heritabilities and bivariate analyses are presented for asthma only.

Lung function test.  Forced vital capacity (FVC) and forced expiratory volume in 1 s (FEV1) were measured with the subject standing, using the vitalograph 5 wedge bellows spirometers (Vitalograph Ltd, Buckingham, UK) in round 1 and the Jaeger Master Scope spirometer (Viasys, Hoechberg, Germany) in round 2, both compliant with the 1995 ATS requirements (17). The measurements were repeated to a maximum of six times until a curve reproducible to within 200 ml was obtained. Both FEV1 and FEV1/FVC were normalized and adjusted for significant (P < 0.05) covariates, which included age, gender, height (FEV1) and ever smoking (FEV1/FVC); the adjustment of covariates was performed prior to estimation of familial correlations or as part of variance components analyses. Residual kurtosis was 1.5 for FEV1 and 0.43 for FEV1/FVC.

Airway histamine challenge test.  The test was performed in both rounds after baseline spirometry according to the protocol of Yan et al. (18) as modified by Woolcock (19). Each dose of saline or histamine was administered via a DeVilbiss No. 45 nebuliser with a hand pump (30% of round 2) or a mechanical (70% of round 2) dosimeter, calibrated at first use. The histamine provocation test was performed seated on all subjects with a FEV1 ≥ 60% (round 1) or >70% (round 2) as predicted by the reference values of Quanjer et al. (20) and Polgar and Promadhat (21), respectively. Each consecutive dose was given at approximately 2-min intervals until either a cumulative dose of 7.8 μmol (round 1) or 2.605 μmol (round 2) of histamine was reached or a fall in FEV1 of ≥ 20% of the postsaline inhalation FEV1 was observed. Participants with a baseline FEV1 <60% (round 1) or ≤70% (round 2) of the predicted value underwent a bronchodilator test, which consisted of 400 μg of salbutamol administered over 60 s via a large spacer. An airway histamine responsiveness (AHR) test was considered positive if the participant experienced a drop in FEV1 of ≥ 20% (hand-held pump) or ≥ 25% (mechanical dosimeter) of the postsaline inhalation FEV1 after the last dose of histamine given. This allowed for the observed difference in histamine output between both methods. A participant with baseline airway obstruction was considered AHR-positive if the bronchodilator test induced an increase in FEV1 of ≥ 15%.

Skin prick test.  The skin prick test (SPT) was performed using a Hollister-Steier prick lancetter on the volar forearm. The longest and the perpendicular diameters of the response wheal were measured after 10 (round 1) or 15 min (round 2) to the nearest millimetre and the mean wheal diameter estimated. Eleven common allergens were tested in each round: cockroach mixture, European house dust mite Dermatophagoides pteronyssinus (D. pter), American house dust mite D. farinae (round 2), house dust (round 1), cat dander, dog dander, Canary grass (round 1), Timothy grass (round 1), southern grasses mixture, Rye grass, Aspergillus spp. mixture, Alternaria spp. mixture, cows milk (round 2) and egg white (round 2). Histamine-buffered saline solution (3.125 mg/ml) was used as the positive control and glycerine as the negative control. A participant was defined as atopic if at least one of the 11 allergens tested elicited a mean wheal diameter at least 3 mm greater than the negative control wheal. Similarly, a SPT to D. pter was considered positive if the mean wheal diameter was 3 mm greater than the negative control wheal.

Measurement of serum total immunoglobulin E (IgE) levels and eosinophil counts.  Total IgE was estimated in both rounds by standard enzyme immunoassay; all samples were assayed in a single batch at the conclusion of each round. Serum eosinophil counts were performed by flow cytometry for round 2 samples only. IgE was log-transformed and eosinophils normalised using an inverse normal transformulation; both were adjusted for age and gender. Residual kurtosis was −0.36 for IgE and −0.07 for eosinophils.

Genotyping and pedigree error checking

DNA was extracted from blood or buccal swabs according to standard procedures (22). DNA from 2760 consenting participants from 735 families was genotyped at 108–440 microsatellite markers at Sequana Inc. as described by Hall and Nanthakumar (23). This genotyping data were used to verify familial relationships using GRR (24): 68 of 735 (9%) families were found to have pedigree errors, mostly due to sample duplications (4.4%), incorrect self-reported paternity (2.1%) or incorrect self-reported zygosity (1.6%). These errors were corrected prior to analyses.

Familial correlations and heritability estimates

Pearson (continuous traits) and tetrachoric (affection traits) correlations were calculated with version 8.02 of sas. Continuous traits were adjusted for significant covariates (defined above) prior to the estimation of familial correlations. Heritabilities for both quantitative and affection traits were estimated by maximum likelihood univariate variance components analysis using solar 2.1.4 (25). Briefly, this analysis decomposed the expected phenotypic covariance between relatives into a component due to polygenic additive genetic variance (inline image), dominance genetic variance (inline image), environmental variance due to environmental factors shared between siblings (inline image) and a component of variance because of environmental factors not shared between relatives (inline image). The significance of the parameters in the model was then tested by the likelihood ratio chi-square test 2[ ln (LH0)− ln (LH1)] with n-df, where LH0 is the likelihood of the data under a saturated model H0, LH1 the likelihood under a nested submodel H1 where the parameters being tested were fixed to zero, and n the difference in the number of parameters estimated by H0 and H1. Significant covariates (defined above) were retained in the means model and proband ascertainment correction implemented. A proband was defined as a twin who reported or was reported by the parents to have a history of asthma or wheeze in either the round 1 or the round 2 questionnaires.

Bivariate analysis of asthma traits

Maximum likelihood variance components analysis was performed to estimate cross-trait correlations and to partition these into a component due to both traits sharing the same set of latent additive genetic factors and to a component due to both traits sharing the same set of latent-specific environmental factors. For this purpose, we estimated the genetic and environmental correlations for any two traits using the multivariate modelling algorithm implemented in a prerelease version of solar 3.0.3 and outlined in Ref. (26). The genetic correlation (ρA) is defined here as the correlation between the latent additive genetic factors, whereas the environmental correlation (ρE) is defined as the correlation between the latent environmental factors not shared between relatives. For eight traits, there are 28 possible two-trait combinations and, therefore, 28 bivariate analyses were performed. Each bivariate analysis estimated six parameters of interest: inline image and inline image for each trait and ρA and ρE between the traits. In addition, the significance of ρA and ρE was tested as described above. To correct for multiple testing, the significance threshold for these parameters was set at P = 0.05/28 = 0.0018. As with the univariate approach, the effects of significant covariates and proband ascertainment correction were included in the analyses. The overall cross-trait correlation for any two traits ti and tj was then calculated as r(ti,tj) =inline image. In this formula, inline image corresponds to the component of the cross-trait correlation that results from traits ti and tj sharing the same set of latent additive genetic factors and inline imageinline image corresponds to the component of the correlation that results from both traits sharing the same set of latent-specific environmental factors. For most traits inline image and inline image did not significantly influence trait variance in the univariate analyses. For this reason, these parameters were fixed at zero in bivariate analyses. At present, solar cannot test the significance of ρA or ρE when analysing more than two traits. In addition, the simultaneous analysis of continuous and affection traits restricted the number of phenotypes solar could analyse and still achieve convergence to only four. Nonetheless, multivariate analyses of up to four traits were performed with parameter estimates that correlated >0.99 with those obtained with the corresponding bivariate analyses (data not shown). For this reason, only results from the bivariate analyses are presented.

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Participants

Overall, 802 families ascertained via at least one twin proband participated in this study. These families were of arbitrary size, ranging from 1 to 10 relatives (mean 4) and from 1 to 3 generations (mean 2). Over 75% of the families consisted of two twins (21%), two twins and one or two parents (36%) or two twins, two parents and one sib (19%). There were 817 sibships available for analysis, of which 533 (65%) were of size 2, 211 (26%) of size 3, 63 (8%) of sizes 4 and 10 (1%) of size 5 or larger. The families included 490 (61%) DZ twins and 312 (39%) MZ twins. Ancestry was reported by 511 families: 72% reported being of European ancestry, 3% of native Australian or New Zealander ancestry, 1% American or Asian, 9% of mixed ancestry and 15% reported ambiguous or unknown ancestry. Data were collected from 3073 relatives from these 802 families, including 1289 self-reported asthmatics (Table 1). Of these, 35% used bronchodilators once a week or more often in the last 12 months and 30% inhaled steroids; 23% reported having previously been admitted to hospital as a result of an asthma attack. These figures were similar for the 919 probands. Males reported a mean age at onset of asthma of 8.1 and females of 11.9 (P < 0.001).

Table 1.  Characteristics of study participants in the full sample and for asthmatics and probands only
 Full sampleSubsamples
ParentsTwinsSibsChildrenSpousesTotalProbandsAsthmatics
  1. *Sample size for ever smoker, current smoker, asthma and doctor asthma corresponds to the number of subjects with questionnaire.

  2. †Sample size for AHR, atopy, D. pter, FEV1 and FEV1/FVC corresponds to the number of subjects that were clinically tested.

  3. AHR, airway histamine responsiveness; FEV1, D. pter, Dermatophagoides pteronyssinus; forced expiratory volume in 1 s; FVC, forced vital capacity; SPT, skin prick test; IgE, immunoglobulin E.

Subjects total10111568376942430739191289
 With questionnaire (%)6510010010010088100100
 Clinically tested (%)898989954689887
 Donated blood (%)9094968988939594
 Full testing (%)892948954649382
Female sex (%)5658504917565957
Age, mean (range)53 (34–83)29 (7–76)24 (7–50)22 (7–43)51 (38–67)36 (7–83)28 (7–76)30 (7–79)
Ever smoker* (%)4832222958343131
Current smoker (%)1217101113141613
Asthma (%)28477783264780100
Doctor asthma (%)2643738125447292
Positive AHR test† (%)154339420415450
Atopy (%)6068757738697575
Positive SPT to D. pter (%)395463668566263
Mean FEV1 [1]3.03.13.23.23.53.13.03.0
Mean FEV1/FVC0.800.830.830.810.790.820.820.82
FEV1 ≤ 0.7 predicted (%)167790789
IgE [IU/ml; geometric mean (N)]55 (440)81 (1344)131 (309)168 (69)52 (17)82 (2179)101 (785)109 (1224)
Eosinophil [×109/l; geometric mean (N)]0.19 (542)0.25 (831)0.28 (347)0.35 (74)0.18 (16)0.24 (1810)0.27 (519)0.28 (1042)

Familial correlations and heritabilities

Most phenotypes had a high reliability, with a test–retest correlation >0.8 (Table 2). Typical familial correlations, uncorrected for ascertainment, ranged from 0.15 to 0.39 for parent–offspring pairs, 0.26 to 0.44 for sib pairs, 0.17 to 0.64 for DZ twin pairs and 0.63 to 0.90 for MZ pairs. The exception was asthma, which, as expected, showed low familial correlations as a result of the selective ascertainment procedure used. Heritabilities corrected for ascertainment ranged from 0.28 for eosinophils to 0.71 for FEV1. The ascertainment correction had a negligible effect in variance components modelling for all traits other than asthma.

Table 2.  Test–retest and familial correlations for asthma and related traits
 Trait correlations* (N pairs)Variance components†
Test–retest‡Parent–offspring pairsSib pairsDZ twinsMZ twinsinline imageinline imageinline imageinline image
  1. *Tetrachoric (affection traits) and Pearson (continuous traits) correlations, uncorrected for ascertainment. Continuous traits were normalized and, with the exception of test–retest, adjusted for significant covariates (see Methods).

  2. †Estimated from univariate variance components analyses with ascertainment correction. All traits were adjusted for significant covariates (see Methods).

  3. Variance components estimates are reported according to the best fitting model.

  4. ‡253 participants were tested in both phases (mean age 36, range: 19–61). The mean timespan between test and retest was 4.8 years (range: 4.2–6.2).

  5. AHR, airway histamine responsiveness; D. pter, Dermatophagoides pteronyssinus; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity; IgE, immunoglobulin E; DZ, dizygotic; MZ, monozygotic.

  6. inline image, additive polygenic variance; inline image, dominance genetic variance; inline image, variance due to environmental factors shared between siblings; inline image, variance due to environmental factors not shared between relatives.

Asthma0.99 (225)−0.01 (1795)0.14 (1251)−0.16 (430)0.34 (263)0.380.62
AHR0.77 (110)0.39 (185)0.32 (675)0.32 (246)0.63 (169)0.580.42
Atopy0.95 (210)0.25 (324)0.44 (1232)0.38 (439)0.78 (281)0.490.160.35
D. pter0.93 (210)0.05 (312)0.33 (1225)0.20 (436)0.76 (280)0.300.320.38
FEV10.87 (209)0.15 (303)0.37 (1199)0.64 (435)0.90 (278)0.710.29
FEV1/FVC0.67 (208)0.25 (305)0.28 (1196)0.37 (434)0.65 (278)0.510.49
IgE0.87 (179)0.24 (1125)0.26 (964)0.18 (355)0.67 (243)0.520.48
Eosinophils0.17 (1362)0.32 (898)0.17 (230)0.59 (113)0.280.140.58

Bivariate analyses of asthma traits

All traits were found to be regulated by additive genetic factors but also by environmental factors not shared between relatives. We then investigated the extent to which different traits shared the same set of additive genetic factors or the same set of specific environmental factors (Table 3). First, large environmental correlations were observed between AHR and atopy (0.67), AHR and asthma (0.63) and between AHR and eosinophils (0.60). By contrast, the two lung function measurements showed low environmental correlations with most traits, particularly with atopy, D. pter and IgE. Secondly, when compared with the environmental effects, fewer genetic correlations were significantly different from zero. As expected, atopy, D. pter and IgE showed a considerable degree of genetic overlap but the three genetic correlations were significantly different from 1 (P < 0.0018). This excluded complete pleiotropy between these traits. As observed with the environmental factors, the genetic correlations between FEV1 and atopy, D. pter and IgE were close to zero. Finally, eosinophils had a significant genetic correlation with IgE (0.44) but with no other trait. Multivariate analyses of up to four traits were also performed and parameter estimates were equivalent to those obtained with the bivariate analyses (not shown). The genetic and environmental correlations were then combined with the trait heritabilities to estimate the overall cross-trait correlations (Fig. 2). As expected from the low genetic and environmental correlations, FEV1 showed low cross-trait correlations with all traits other then FEV1/FVC. By contrast, AHR showed high and comparable cross-trait correlations with most traits, with equivalent contributions from genetic and environmental effects shared between traits. The exception was eosinophils. The correlation observed between this trait and AHR, or between this trait and other traits other than IgE was mostly due to environmental factors shared between traits.

Table 3.  Environmental (above main diagonal) and genetic (below main diagonal) correlations between asthma traits
 Environmental and genetic correlations*
AsthmaAHRAtopyD. pterFEV1FEV1/FVCIgEEosinophil
  1. *Estimated from bivariate variance components analyses with ascertainment correction.

  2. Correlations significantly different from zero are highlighted in bold (P < 0.0018). A negative correlation indicates that environmental or genetic factors induced a negative phenotypic correlation between the two traits. Affection traits were coded as 0 (unaffected) and 1 (affected).

  3. All traits were adjusted for significant covariates (see Methods).

  4. AHR, airway histamine responsiveness; D. pter, Dermatophagoides pteronyssinus; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity; IgE, immunoglobulin E.

Asthma 0.630.440.490.310.280.230.30
AHR0.24 0.670.60−0.320.380.570.60
Atopy0.460.35 0.890.02−0.220.710.55
D. pter0.380.430.76 −0.060.270.580.52
FEV1−0.24−0.20−0.05−0.11 0.420.040.11
FEV1/FVC−0.240.430.22−0.220.33 0.25−0.09
IgE0.410.330.610.58−0.18−0.07 0.38
Eosinophils0.240.130.270.19−0.21−0.160.44 
image

Figure 2. Maximum likelihood cross-trait correlations for eight asthma phenotypes, corrected for ascertainment. The correlation between any two traits was decomposed into a component due to both traits sharing the same set of latent additive genetic factors (A) and to a component due to both traits sharing the same set of latent-specific environmental factors (E). The cross-trait correlations estimated by this maximum likelihood approach were highly correlated (0.98) with the corresponding Pearson (continuous–continuous traits), tetrachoric (affection–affection traits) or biserial correlations (continuous–affection traits) estimated with sas (not shown). *Indicates traits with environmental or genetic correlations significantly different from zero. AHR, airway histamine responsiveness; D. pter, Dermatophagoides pteronyssinus; FEV, forced expiratory volume in 1 s; FVC, forced vital capacity; IgE, Immunoglobulin E.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Asthma has long been recognized to be heritable (27). Large population-based twin studies have estimated the heritability of asthma to be between 60% and 73% (28–30). Similarly, most asthma intermediate phenotypes, such as atopy and total serum IgE, have been shown to be moderately to highly heritable in population-based samples (14, 31). The present study involved selective ascertainment of families through twins who had reported asthma or wheezing in previous screening questionnaires. Approximately 70% of subjects were found to be atopic, 44% had previously been diagnosed with asthma by a doctor and 40% had a positive AHR test. As a result of this selection procedure, self-reported asthma displayed very low familial correlations. On the contrary, the bias introduced by the ascertainment did not extend to any of the other asthma-related phenotypes tested. This has been shown previously (32, 33) and it is likely to reflect the low to moderate cross-trait correlations observed between asthma and related phenotypes (range: −0.25 to 0.42). Also, the ascertainment bias observed for asthma was largely adjusted by the correction used when fitting variance components models. Heritabilities corrected for ascertainment ranged from 28% (eosinophils) to 71% (FEV1).

We performed bivariate variance components analyses to estimate the genetic and environmental correlation between asthma traits. Similar analyses have been performed previously to estimate the genetic and environmental overlap between different allergic diseases (28, 34) and between different allergic diseases and their symptoms (15), but not between different asthma intermediate phenotypes. Our analyses showed that many asthma traits had significant environmental correlations with each other, often >0.5. This is consistent with the hypothesis that the same set of environmental factors regulates the variation of many asthma traits. This hypothesis is also supported by the observation that MZ twin pairs discordant for asthma are also likely to be discordant for other traits and that the case for the different traits is often the same twin (11, 12). The main exception in our study was FEV1, which showed very low environmental correlations with atopy, D. pter and IgE. This strongly suggests that the environmental factors not shared between relatives that regulate variation in FEV1 are entirely distinct from the environmental factors that regulate variation in allergy status.

The bivariate analyses also showed that only a comparatively smaller number of traits had genetic correlations significantly different from zero. As expected, the most significant of these were between atopy, D. pter and IgE, although there was no evidence for complete pleiotropy. This implies that although a large proportion of genetic factors were shared between these traits, there were still specific genetic pathways that regulated the expression of each trait individually. Palmer et al. (14) reported similar results when comparing the genetic overlap between total and specific IgE with regression models. They also found that the genetic determinants of total IgE and airway responsiveness exhibited <30% sharing, which is consistent with our estimate of the genetic correlation between these two traits (0.33). In addition, in the present study, both asthma and AHR showed some degree of genetic overlap with atopy, thus supporting the concept that these three traits share a group of genetic risk factors (10). Interestingly, Eosinophils and IgE were also found to have a significant genetic correlation, which could explain the recent report of genetic linkage of eosinophilia to 2q33 (35), a region previously linked to atopy (36). Finally, there was very limited or no genetic overlap between FEV1 and any other trait other than FEV1/FVC. Thus, to a great extent, baseline FEV1 was genetically distinct from Asthma, AHR and Atopy, which is also consistent with previous findings (13).

How can these results be used to guide the choice of traits for multivariate linkage analysis? The joint analysis of multiple correlated traits has been demonstrated to increase the power of linkage analysis, both with real (4–6) and simulated (7, 37) data sets. However, not all correlated traits provide the same improvement in power when analysed with multivariate methods. Traits with strong genetic correlations provide increased power only when the pleiotropic QTL and the residual sources of variation induce cross-trait correlations in opposite directions (7, 8, 38). As this may be uncommon, multivariate analysis of traits such as atopy and D. pter is unlikely to be justified. Similarly, traits with minimal genetic correlations, such as FEV1 and Atopy, are unlikely to have common QTL and so are also unlikely to benefit from being analysed with multivariate methods. Therefore, the traits most likely to provide increased power for multivariate linkage are traits with low to moderate genetic correlations, such as Asthma, AHR, Atopy, IgE and FEV1/FVC. The observed genetic correlations between these imply that although each trait is regulated to great extent by specific genetic factors, there is still some degree of genetic overlap that could be exploited by multivariate approaches. Nonetheless, we note that the moderate to high environmental correlations observed between some of these traits may prove counterproductive.

Finally, a number of limitations need to be born in mind when examining the conclusions of the present study. First, the clinical protocol differed in a few aspects between the two rounds of testing. However, these differences were minor and are unavoidable when pooling data from different samples or from different studies. Secondly, the definition of asthma used here may not fully represent the clinical entity of asthma. Nonetheless, it was our intention to contrast a self-reported asthma phenotype with more objective clinical phenotypes. Thirdly, all subjects that were tested twice in this study were over the age of 18. As the majority of our sample (75%) was within this age group, this is unlikely to have greatly overestimated the reliability of the asthma traits measured. Lastly, the Bonferroni correction applied when testing the significance of genetic and environmental correlations may have been conservative. The parameter estimates were nonetheless unaffected by this correction and should be assessed in conjunction with the reported significance.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Authors thank all families and those who helped to undertake this study. Especially, Dr Claire Wainwright (Royal Brisbane Hospital), Barbara Haddon and Dixie Statham (QIMR, Brisbane), Wendy Oddy (Institute for Child Health Research, Perth), Sunalene Devadason and Dr Nigel Dore (Princess Margaret Hospital, Perth) and Marita Dalton (Royal Children's Hospital, Melbourne). Also thank two anonymous referees for comments that helped to improve the original manuscript.

Data collection was funded by Sequana Therapeutics Inc. and later by AxyS Pharmaceutical Inc. and by a National Health and Medical Research Council grant to DLD. MARF was supported by a doctorate grant from the Fundação para a Ciência e Tecnologia, Portugal.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References
  • 1
    Schork NJ. Extended multipoint identity-by-descent analysis of human quantitative traits: efficiency, power, and modeling considerations. Am J Hum Genet 1993;53: 13061319.
  • 2
    Amos CI. Robust variance-components approach for assessing genetic linkage in pedigrees. Am J Hum Genet 1994;54: 535543.
  • 3
    Blangero J, Almasy L. Multipoint oligogenic linkage analysis of quantitative traits. Genet Epidemiol 1997;14: 959964.
  • 4
    Marlow AJ, Fisher SE, Francks C, MacPhie IL, Cherny SS, Richardson AJ et al. Use of multivariate linkage analysis for dissection of a complex cognitive trait. Am J Hum Genet 2003;72: 561570.
  • 5
    Olswold C, de Andrade M. Localization of genes involved in the metabolic syndrome using multivariate linkage analysis. BMC Genet 2003;4(Suppl. 1):S57.
  • 6
    Turner ST, Kardia SL, Boerwinkle E, De Andrade M. Multivariate linkage analysis of blood pressure and body mass index. Genet Epidemiol 2004;27: 6473.
  • 7
    Allison DB, Thiel B, St Jean P, Elston RC, Infante MC, Schork NJ. Multiple phenotype modeling in gene-mapping studies of quantitative traits: power advantages. Am J Hum Genet 1998;63: 11901201.
  • 8
    Amos C, de Andrade M, Zhu D. Comparison of multivariate tests for genetic linkage. Hum Hered 2001;51: 133144.
  • 9
    Edfors-Lubs ML. Allergy in 7000 twin pairs. Acta Allergol 1971;26: 249285.
  • 10
    Clarke JR, Jenkins MA, Hopper JL, Carlin JB, Mayne C, Clayton DG et al. Evidence for genetic associations between asthma, atopy, and bronchial hyperresponsiveness: a study of 8- to 18-yr-old twins. Am J Respir Crit Care Med 2000;162: 21882193.
  • 11
    Duffy DL, Mitchell CA, Martin NG. Genetic and environmental risk factors for asthma: a cotwin-control study. Am J Respir Crit Care Med 1998;157: 840845.
  • 12
    Strachan DP, Wong HJ, Spector TD. Concordance and interrelationship of atopic diseases and markers of allergic sensitization among adult female twins. J Allergy Clin Immunol 2001;108: 901907.
  • 13
    Palmer LJ, Knuiman MW, Divitini ML, Burton PR, James AL, Bartholomew HC et al. Familial aggregation and heritability of adult lung function: results from the Busselton Health Study. Eur Respir J 2001;17: 696702.
  • 14
    Palmer LJ, Burton PR, Faux JA, James AL, Musk AW, Cookson WO. Independent inheritance of serum immunoglobulin E concentrations and airway responsiveness. Am J Respir Crit Care Med 2000;161: 18361843.
  • 15
    Nystad W, Roysamb E, Magnus P, Tambs K, Harris JR. A comparison of genetic and environmental variance structures for asthma, hay fever and eczema with symptoms of the same diseases: a study of Norwegian twins. Int J Epidemiol 2005; DOI: 10.1093/ije/dyi061.
  • 16
    Medical Research Council Committee on the Aetiology of Chronic Bronchitis. Standardized questionnaire on respiratory symptoms. Br Med J 1960;2: 1665.
  • 17
    American Thoracic Society. Standardization of spirometry, 1994 update. American Thoracic Society. Am J Respir Crit Care Med 1995;152: 11071136.
  • 18
    Yan K, Salome C, Woolcock AJ. Rapid method for measurement of bronchial responsiveness. Thorax 1983;38: 760765.
  • 19
    Woolcock AJ. A protocol for population studies of respiratory disease in adults and children. University of Sydney, NSW, Australia: Institute of Respiratory Medicine, 1996.
  • 20
    Quanjer PH, Tammeling GJ, Cotes JE, Pedersen OF, Peslin R, Yernault JC. Lung volumes and forced ventilatory flows. Report Working Party Standardization of Lung Function Tests, European Community for Steel and Coal. Official Statement of the European Respiratory Society. Eur Respir J Suppl 1993;16: 540.
  • 21
    Polgar G, Promadhat V. Pulmonary function testing in children: techniques and standards. Philadelphia, USA: W. B. Saunders, 1971.
  • 22
    Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res 1988;16: 1215.
  • 23
    Hall J, Nanthakumar E. Automated fluorescent genotyping. In: BoyleAL, editor. Current protocols in human genetics. New York, USA: Wiley & Sons, 1997.
  • 24
    Abecasis GR, Cherny SS, Cookson WO, Cardon LR. GRR: graphical representation of relationship errors. Bioinformatics 2001;17: 742743.
  • 25
    Almasy L, Blangero J. Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet 1998;62: 11981211.
  • 26
    Williams JT, Van Eerdewegh P, Almasy L, Blangero J. Joint multipoint linkage analysis of multivariate qualitative and quantitative traits: I. Likelihood formulation and simulation results. Am J Hum Genet 1999;65: 11341147.
  • 27
    Salter HH. On asthma: its pathology and treatment. London, UK: John Churchill, 1860.
  • 28
    Duffy DL, Martin NG, Battistutta D, Hopper JL, Mathews JD. Genetics of asthma and hay fever in Australian twins. Am Rev Respir Dis 1990;142: 13511358.
  • 29
    Harris JR, Magnus P, Samuelsen SO, Tambs K. No evidence for effects of family environment on asthma. A retrospective study of Norwegian twins. Am J Respir Crit Care Med 1997;156: 4349.
  • 30
    Skadhauge LR, Christensen K, Kyvik KO, Sigsgaard T. Genetic and environmental influence on asthma: a population-based study of 11,688 Danish twin pairs. Eur Respir J 1999;13: 814.
  • 31
    Lawrence S, Beasley R, Doull I, Begishvili B, Lampe F, Holgate ST et al. Genetic analysis of atopy and asthma as quantitative traits and ordered polychotomies. Ann Hum Genet 1994;58(Pt 4):359368.
  • 32
    Felsenfeld S, Kirk KM, Zhu G, Statham DJ, Neale MC, Martin NG. A study of the genetic and environmental etiology of stuttering in a selected twin sample. Behav Genet 2000;30: 359366.
  • 33
    Wade T, Martin NG, Neale MC, Tiggemann M, Treloar SA, Bucholz KK et al. The structure of genetic and environmental risk factors for three measures of disordered eating. Psychol Med 1999;29: 925934.
  • 34
    Lichtenstein P, Svartengren M. Genes, environments, and sex: factors of importance in atopic diseases in 7–9-year-old Swedish twins. Allergy 1997;52: 10791086.
  • 35
    Evans DM, Zhu G, Duffy DL, Montgomery GW, Frazer IH, Martin NG. Major quantitative trait locus for eosinophil count is located on chromosome 2q. J Allergy Clin Immunol 2004;114: 826830.
  • 36
    Ober C, Tsalenko A, Parry R, Cox NJ. A second-generation genomewide screen for asthma-susceptibility alleles in a founder population. Am J Hum Genet 2000;67: 11541162.
  • 37
    Amos CI, Laing AE. A comparison of univariate and multivariate tests for genetic linkage. Genet Epidemiol 1993;10: 671676.
  • 38
    Evans DM, Duffy DL. A simulation study concerning the effect of varying the residual phenotypic correlation on the power of bivariate quantitative trait loci linkage analysis. Behav Genet 2004;34: 135141.