TWIN STUDIES OF ASTHMA AND ALLERGY

Earlier studies

Wuthrich et al [1981]:

Total and specific sIgE level, and skin atopy were studied in 50 pairs of twins (30 MZ pairs, 20 DZ pairs) in which one or both suffered from atopic disease [1981]. They were recruited via a survey of the Swiss Twins Association (30 pairs) and from the sample studied by Schnyder [1960, see above]. Skin prick and RAST results for eight common allergens were presented.

MZ concordance for atopy was higher than DZ concordance (see Table). Wuthrich also presents a summary table from the German language review of twin studies by Schnyder [1970] that shows these findings to be consistent with the literature, though there is considerable heterogeneity across the earlier studies, no doubt reflecting differences in ascertainment (for the Table, Mantel-Haenszel MZ v DZ common OR=2.7, 95%CI=1.9-4.0; Breslow-Day X2(6 df)=19.2, P=0.004). Intertwin differences in total sIgE did not differ significantly between the MZ and DZ twin pairs. Similarly, the concordance in type of specific skin and RAST reactivity among the MZ and DZ twins did not differ significantly (with the exception of timothy grass - MZ pairwise concordance, 25/30; versus DZ, 12/20).

Table 1. MZ and DZ concordances for atopy from Wuthrich et al [1981], along with a summary table cited from Schnyder [1970].
ReferenceMZDZ
ConDisConDis
Wuthrich et al (1981)1713416
Spaich-Ostertag (1936)66216
Harvald-Hauge (1956)1010440
Rajka (1960)3263
Schnyder (1960)66122
Teodori (1962)5337134159
Niermann (1967)50310
Total Number10074154266
(Percentage)57.5%42.5%36.6%63.3%

Milani-Comparetti:

This short paper reports an "anamnestic study" of allergic symptoms among 500 adult twin pairs from the twin register maintained by the Mendel Institute. Atopy was defined as a history of oculorhinitis, asthma, urticaria, allergic eczema, or "unspecified allergic disease". The MZ concordance for atopy (28/36) was significantly higher than the DZ concordance (30/40).

Falliers et al [1971]:

This paper describes two pairs of asthma discordant MZ pairs. Its interest lies in the detailed physiological measures performed on the twins. These included dermatoglyphic and serological zygosity typing, skin atopy testing, total sIgE level determination, pulmonary function testing including exercise and methacholine provocation testing, and a leucocyte membrane adenylate cyclase and ATPase activity assay.

In both cases, the unaffected cotwin gave a history consistent with atopy, and of at least one episode of RTI associated wheeze. Onset of asthma in the affected twin was associated in each case with a febrile RTI which did not affect the cotwin. One asthmatic was steroid dependent. The nonasthmatics did not exhibit any bronchial responsiveness. Leucocyte adenylate cyclase and ATPase activity was significantly higher in the asthmatic compared to the nonasthmatic subjects. This could be regarded as evidence supporting the previously discussed sympathetic overactivity associated with atopy, or to be due to therapy with sympathomimetic drugs.

Bazaral et al [1974]:

Bazaral, Orgel and Hamburger [1974] examined total sIgE levels in 54 pairs of MZ twins and 39 pairs of DZ twins. Zygosity was determined by serological typing. The twins were mainly military veterans (age 45-55). An additional 10 pairs of MZ twin and 13 pairs of DZ twin children were also examined. The within twin and between twin variances of the log transformed sIgE level (logIgE) were used in the statistical analysis, and Holzinger's H calculated. The test-retest correlation coefficient for logIgE was 0.948 (N=23 pairs, mean delay 17 months).

Despite the small sample size, genetic effects were detected in the children (F13,10=MWDZ/MWMZ=4.59, P<0.01), with Holzinger's H estimated at 79% (90% confidence interval 44%-93%). Reanalysis of this data using LISREL 7.16 confirmed the preferability of the genetic model over the shared environment model (CE v ACE X2(1 df)=5.45, P=0.02; AE v ACE X21=1.8, P=0.15; see Table for abbreviations). For the adult twins (see Table), total MZ variance did not differ significantly from total DZ variance (F77,107=1.47, P=0.05). Holzinger's heritability was estimated at 59%. Since Holzinger's measure does not correspond directly to the biometrical broad sense heritability, these values have been recalculated with LISREL 7.16 (Table), giving a genetic variance of 0.22 and thus a heritability of approximately 80%. This agrees well with the unweighted estimate of the genetic variance derived from the intrapair MZ-DZ variance difference (0.22, 95% confidence interval 0.05-0.39). No nonadditive component could be estimated, but as discussed elsewhere, the power to detect nonadditivity in small twin samples is low. For the children, the ML estimate of the heritability was 93% (88-99%).

Table 2. Mean squares and intraclass correlations for logIgE in adult MZ and DZ twins [from Bazaral et al 1974].
SourcedfMSr
MZ between pairs530.4029
MZ within pairs540.07670.68
DZ between pairs380.5188
DZ within pairs390.18720.47
Reliability (average delay between tests=17 mo)
between occasions220.2550
within occasions230.00920.95

Table 3. LISREL modelling using reported mean squares for adult twins [from Bazaral et al 1974]. A=additive genetic component; C=shared environment; E=unique environment.
Model testedh2 (95% CI)X2dfP
ACE73%3.6210.16
AE79% (68-90%)3.6720.30
CE0%13.0320.01

Konig & Godfrey [1974]:

These authors [1974] examined 8 pairs of MZ twins and 7 pairs of DZ twins selected because at least one twin was asthmatic, or a first degree relative was asthmatic. Zygosity typing was performed, except in one unambiguous case. All twins performed an exercise (bronchial) provocation protocol (6 minutes on a treadmill running at a pace to maintain the heart rate at 160-170 beats per minute) with peak expiratory flow rate being recorded using a Wright peak flow meter. A second free running exercise test where previously diagnosed asthmatics failed to demonstrate significant bronchial hyperresponsiveness (defined as a >22% lability index eg (maximum PEF-minimum PEF)/(resting or expected PEF)*100). Skin prick testing for 24 common allergens was performed.

While 3 of the MZ and 1 of the DZ twin pairs were concordant for diagnosed asthma, 6/8 MZ twin pairs were concordant for BHR and 1/7 DZ twin pairs. Similar figures (6/8 and 1/7 pairs) were noted for presence of skin atopy.

Al-Agidi [1980]:

In this Iraqi study (1980), 64 MZ and 43 DZ pairs of twins "in good health" donated serum for immunoglobulin A,E,G and M level determination. Heritability of all four classes was high.

Edfors-Lub [1971]

This is the second largest twin study of allergic disease [1971]. Nearly 7000 complete same sex twin pairs (a 74.5% pairwise response rate) from the Swedish Twin Registry (a community based twin registry with near complete ascertainment - 95%) born between 1886 and 1925 took part in the study. All subjects completed a questionnaire which included 12 items on allergic disease, as well as items for zygosity diagnosis [Cederlof 1961]. The questionnaire was piloted on 1400 twin pairs of whom 39 pairs were also examined by an allergologist. The correlation between clinical diagnosis and self reported diagnosis in the twins was good (see Table). The author notes that the twins as a group were older than the general population (the study being performed in 1966-7), so that active atopic disease would be relatively uncommon, and that medical diagnosis would largely be anamnestic anyway. Test- retest reliability from the initial 1961 questionnaire (2 items on allergy), Edfors-Lubs' pilot and the allergy questionnaire was good for "have you ever had asthma?" where only 5% of subjects had given inconsistent answers. In two questionnaires "have you ever had hayfever?" was asked; 12.5% of answers were inconsistent. There were 2434 pairs of MZ twins and 4302 pairs of DZ twins.

The lifetime prevalence of asthma was 3.8% overall (males 3.5%, females 4.0%), representing 521 asthmatic individuals. Hayfever was present in 14.8% of the sample (males 15.5%, females 14.2%), while 2% reported a history of childhood eczema (no sex difference), 2.5% eczema at any time in their life and 2.4% a history of allergic dermatitis.

Table 4. Association between physician diagnosis and questionnaire diagnosis [from Edfors-Lub 1971].
Clinical DiagnosisQuestionnaire diagnosis
AsthmaEczemaHayfever
YesNoYesNoYesNo
Yes714065
No130134127
Tetrachoric r (SE)0.98 (0.03)0.97 (0.04)0.86 (0.12)
The atopic diseases tended to occur within the same individuals and within the same twin pairs (see Table). No differences in prevalence in city and rural dwellers were noted, with the exception of contact dermatitis, which was increased in city dwellers. Contact dermatitis was not associated with the atopic conditions, and is not generally recognised as part of the allergic diathesis.

These results lead to heritability estimates for the individual traits (using the WLS option in LISREL 7.16) of 63% for asthma (95% confidence interval 54-73%); for hayfever 43% (37-50%); and for eczema 62% (51-75%). There was also significant concordance for number of allergic syndromes reported in each twin; this was greater for MZ than DZ twins.

Monogenic hypotheses were fitted to the data using traditional methods. All modes of Mendelian inheritance fitted equally well, and Edfors-Lubs concluded by suggesting a polygenic model might best explain the genetic correlation noted between the number of atopic symptoms reported by each twin. As noted earlier in describing the study of Borecki et al [1985], a pleiotropic gene may act in a dominant fashion for one syndrome, and in a codominant or recessive manner for another manifestation. Therefore the author's conclusion is not strictly justified, and an increasing number of concurrent symptoms in heterozygotes and homozygote affecteds could be invoked as an explanation for this observation.

Table 5. Approximate intraindividual and intrapair tetrachoric correlations (and standard errors) of reported atopic symptoms (with equivalent odds ratios in bold print in the upper triangles). Calculated from Tables 3.1 and 3.2.1 in Edfors-Lub [1971].
Asthma in index twinHayfever in index twinEczema in index twin
Asthma in index twin1.006.1 (5.1-7.3)4.0 (2.9-5.5)
Hayfever in index twin0.47 (0.02)1.002.3 (1.8-3.0)
Eczema in index twin0.29 (0.04)0.21 (0.03)1.00
Asthma in MZ cotwin0.67 (0.05)2.6 (1.7-4.2)4.1 (1.9-8.9)
Hayfever in MZ cotwin0.25 (0.06)0.42 (0.04)1.9 (1.1-3.5)
Eczema in MZ cotwin0.30 (0.09)0.15 (0.08)0.64 (0.06)
Asthma in DZ cotwin0.20 (0.07)1.8 (1.2-2.6)1.5 (0.6-3.5)
Hayfever in DZ cotwin0.15 (0.05)0.23 (0.03)1.4 (0.9-2.3)
Eczema in DZ cotwin0.08 (0.09)0.08 (0.06)0.26 (0.08)

I have also fitted univariate LISREL genetic models, which revealed a large dominance component for asthma, as is suggested by the ratio of the MZ to DZ correlations. The proportion of total variance due to dominance was estimated at 0.54 (X2(1 df)=3.4, P=0.07; 95% CI 0.11- 1.00). Hayfever and eczema were adequately described by purely additive models (X2(1 df)=0.31 and 0.40 respectively).

Binary correlation models for asthma concluded that there was strong nonadditivity present. These were fitted across two age bands (born after 1912, or born before 1913 - using raw data made available by the Swedish Twin Registry). There were significant cohort effects on prevalence, but not concordance. The additive genetic component was estimated as zero. No sex differences were detected.

Finnish Twin Registry studies

Baseline survey [1978]:

This is the largest twin study of asthma to date. The Finnish Twin Cohort contains 13888 pairs of same-sex twins born 1905-1957 and was assembled using centralised birth and death information [Kaprio et al 1978]. The baseline questionnaire mailed in 1975-6 was wide ranging, and included items for the diagnosis of zygosity, and disease items prefaced by the question "Have you ever been told by a doctor that you have or have had" for asthma, hayfever, and eczema [Kaprio et al 1979]. The lifetime prevalence of asthma varied from 1% in those aged under 40 years, up to 4% of those aged over 60 years, with no significant sex differences. MZ twins were significantly more concordant for asthma.

Table 6. Prevalence and probandwise concordance for doctor- diagnosed asthma among members of the Finnish Twin Registry in 1976. Table 7.
MalesFemales
Age (years)18-3940-5960-18-3940-5960-
Prevalence (%)0.942.114.160.812.623.85
MZ concordance0.630.510.530.870.320.65
DZ concordance0.150.110.000.100.180.24

Nieminen et al [1990]:

This paper reports refinement of the diagnosis of asthma in the Registry. The Twin Registry database was correlated with a number of other registers such as hospital admissions and medications covering the period 1970-1987. The cumulative incidence of asthma was approximately 2% in this cohort using this definition.

The choice of a relatively "hard" endpoint (hospital admission or enrolled with the Social Insurance Institution as requiring chronic anti- asthma medication) in a country with a relatively low prevalence of asthma meant that only a small number of concordant affected pairs were recruited into the study (22 out of 13888). As a result, the relative risks and tetrachoric correlations had broad confidence intervals (Table), especially in the men.

Table 8. Asthma in the Finnish Twin Cohort [Nieminen et al 1990] - concordances for chronic asthma.
GroupAsthma in pairPhi Tetrachoric rExact 95% Confidence Limits
+++---
MZ Males36719800.070.3140.033-0.558
DZ Males714947490.070.3350.156-0.499
MZ Females77121790.150.4990.291-0.6705
DZ Females519444770.030.160-0.029-0.342

I fitted genetic multifactorial threshold models using LISREL and MX. MX was used to fit threshold models directly to the combined ages concordances for male and female twins (Table). This showed that there was no evidence for sex heterogeneity on X2 test (homogeneity X2(2 df)=4.65-2.10-0.26=2.29, P=0.32). The pooled estimate for heritability was then 0.36 (LISREL WLS 90%CI=0.02-0.56) for the full (ACE) model, and 0.46 for the more parsimonious additive genetic (AE) model. Similar conclusions were drawn on fitting log-linear models to the same data.

Table 9. Fitting of threshold models to Finnish twin data [Nieminen et al 1990].
Model tested*Proportion of variation due toTest of model goodness-of-fit
AC/DELR X2dfP
Male Twins
ACE0.0 00.3 30.6 70.2630.97
AE0.4 2---0.5 82.8840.58
CE---0.3 30.6 70.2640.99
Female Twins
ADE0.1 20.3 90.4 91.2230.75
AE0.4 7---0.5 32.1040.72
CE---0.2 90.7 17.6140.11
Both Sexes
ACE0.3 60.0 70.5 74.6580.79
AE0.4 4---0.5 64.8490.85
CE---0.3 10.6 97.6290.57

* A,C, and E refer to additive genetic, shared (or family) environmental and unshared (or unique) environmental sources of variation included in the model. D refers to nonadditive genetic variation.

Finally, I compared the concordances presented by Nieminen et al with those from the similar population based study of Edfors-Lub (Table). This again suggested that one model could adequately describe both studies (homogeneity X2(8 df)=13.56-3.62-2.10-0.26=7.59, P=0.11).

Table 10. Heterogeneity testing of Finnish and Swedish data using threshold models fitted with MX.
Twin GroupX2 (d.f.) fit of ACE* model to each groupHeritability from separate modelsX2 (d.f) fit of ACE* model to combined dataHeritability from combined groups
Finnish Males0.26 (3)0.00
Finnish Females2.10(3)0.47 4.65 (8)0.34
Swedish twins**3.62(3)0.6313.56 (13)0.55

* Model comprising additive genetic (A), shared (C) and unshared (E) environmental sources of variation.

** Using contingency tables reconstructed from Tables 3.1 and 3.2.1 in Edfors-Lub [1971].

Asthma in the Norwegian Twin Registry

Harris and coworkers [1997] describe another Scandinavian population sample of twins. All 5078 pairs of twins born 1967-1974 were ascertained, of which only 4231 pairs contained both members surviving past age 3 years! Questionnaires were sent in 1992 to the 3996 pairs contactable.

There was a 73% individual response rate (2570 complete pairs and 724 responses from one twin only). A positive response was received from 332 individuals (5.7%) to the asthma item on a question "Have you now, or have you ever had any of the following illnesses or health problems?". Males had an earlier age at onset, but there were no significant effect on cumulative incidence of sex, zygosity, or year of birth.

MZ twins were significantly more concordant than DZ twins with crude or survival analysis giving very similar point estimates (Table). The MFT heritability was estimated at 71% under the best fitting AE model.

Table 11. Asthma in Norwegian twins [Harris et al 1997] - concordances for asthma.
GroupAsthma in pairTetrachoric rExact 95% Confidence Limits
+++---
MZ Males13243750.8060.637-0.910
DZ Males3403410.230-0.118-0.534
MZ Females10334840.6830.472-0.830
DZ Females2503900.051-0.298-0.383
DZ Opp Sex6727160.2890.039-0.509

Table 12. Fitting of threshold models to Norwegian twin data [Harris et al 1997].
Model tested*Proportion of variation due to Test of model goodness-of-fit
AC/DELR X2dfP
ADE0.100.650.251.5230.68
AE0.71---0.294.8340.31
CE---0.470.5326.4340.00

* A,C, and E refer to additive genetic, shared (or family) environmental and unshared (or unique) environmental sources of variation included in the model. D refers to nonadditive genetic variation.

Asthma in the Danish Twin Registry

The 1953-1982 birth cohort of the Danish Twin Registry was sent a seven page questionnaire in 1994 that contained one item on asthma in a checklist: "Do you have, or have you ever had asthma?" [Skadhauge et al 1999]. Hayfever and atopic dermatitis were also asked after. Of the 34076 individuals mailed, there were 29183 (86%) who returned it completed (up to one reminder letter was sent).

The cumulative prevalence of asthma was 6.2%. This decreased with age in males but not females. There was significant sex heterogeneity in the concordances, the evidence for which comes especially from the older panel, and the MZ/DZ difference in tetrachoric correlations was consistent with dominance.

Table 13. Asthma in Danish twins [Skadhauge et al 1999] - concordances for asthma.
GroupAsthma in pairTetrachoric rExact 95% Confidence Limits
+++---
1953-67
MZ Males 21406800.810.68-0.89
DZ Males 8 8210350.370.16-0.56
MZ Females21707850.650.50-0.77
DZ Females611811100.15-0.06-0.365
DZ Opp Sex821718170.05-0.12-0.23
1968-82
MZ Males 29638440.760.65-0.85
DZ Males 121018800.360.17-0.53
MZ Females298110270.710.59-0.80
DZ Females181038730.470.30-0.61
DZ Opp Sex1819413970.250.10-0.40

Asthma in the Virginia "30000"

The next data set was kindly made available by Dr. Lindon Eaves, and has yet to be analysed by that group. The Virginia "30000" comprises a population sample (N=1915 pairs) of twins born 1915-71 in Virginia ascertained via birth records of the Commonwealth of Virginia and located using state government records, especially those of the Department of Motor Vehicles. A further 3347 pairs are volunteers recruited through an article in the newsletter of the American Association of Retired Persons. This group were resident throughout the US. In addition, relatives and spouses of the twins designated by the twins were also invited to take part in the study. Data was collected using a 10 page questionnaire which covered a number of different subject areas. The twin pairwise response rate was 70%. Complete responses were received from 18320 relatives (45% of those mailed a questionnaire). The data I have analysed however is a subset dating from before completion of data entry and comprises 23200 individuals and 4310 complete twin pairs.

In the item on health conditions, subjects were asked to check a variety of conditions if they had "ever [had] any of these problems diagnosed or treated by a physician. Please check any that apply only if you know the problem was DIAGNOSED OR TREATED BY A PHYSICIAN". One of these 16 health problems was "asthma". In addition, subjects reported the age at onset of the conditions ("as accurately as you can, the [...] AGE when the problem was first DIAGNOSED OR TREATED BY A PHYSICIAN").

The following analyses have been limited to twinships where a response from both twins has been received. These twins have been divided into three groups for the purposes of genetic analysis: those pairs who have no self-report information on parental asthma - Group A; those with data from only one parent (usually the mother) - Group B; and those where data from both the parents is available - Group C.

Within the 22993 subjects who had a recorded age, the prevalence of physician diagnosed asthma was 5.5%. There were no differences between the sexes. The prevalence of asthma increased with age to 70 years old, and then declined (Figure 9.1). However, this trend was not discernible when the data was divided into the three subgroups.

Examining the reported age-at-onset of asthma using standard life table methods found a significant increase in the incidence of asthma between different age cohorts. Within each cohort, the age-specific incidence curve was U-shaped, falling to a minimum around age 20-30, and rising monotonically thereafter. No sex differences in incidence were detectable.

Prior to path analysis using the multifactorial threshold model, I fitted log-linear models separately to the three groups using GLIM and PROC CATMOD in SAS (Table). For Group A, no effects of birth cohort was detected on prevalence of asthma, or on concordance between twins. MZ twins were significantly more concordant than DZ twins. No birth cohort effects were detected in Groups B and C. However, there was no significant influence of zygosity on twin concordance in Group B, and only a marginal effect in Group C.

Table 14. Age-adjusted (Mantel-Haenszel) Relative Risks for asthma in MZ and DZ twins, and P value for significance of zygosity (MZ=1, DZ=0) term in log-linear models (that included birth cohort, parental status, zygosity, and twin status).
GROUPRR-MZRR-DZInteraction P
Group A9.23.90.00
Group B5.44.30.89
Group C10.62.50.29

The path analyses were performed using LISREL 7.16 and MX, the latter allowing a multigroup model (Table 7.13). A significant genetic component of variation in liability to asthma was detected. Shared environmental and nonadditive genetic components did not contribute significantly to this.

Table 15. Parameter estimates and WLS-adjusted goodness-of-fit X2 for threshold models fitted to the three groups simultaneously and separately. The homogeneity X2 for the AE model was not significant (X2(2 df)=1.04, P=0.6).
MODELh2c2d2e2X2df
ACDE 3 groups0.530.100.010.3614.217
ACE 3 groups0.540.10---0.3614.318
AE 3 groups0.65------0.3515.619
CE 3 groups---0.51---0.4943.019
AE Group A0.67------0.330.021
AE Group B0.46------0.541.345
AE Group C0.57------0.4313.211

Hopp et al

This study is smaller than the previous three, but contains data on bronchial responsiveness. The authors examined 107 pairs of twins, recruited from a number of sources including via media advertising. There were 61 MZ pairs and 46 DZ pairs. The mean age of the subjects was 14 years (SD=6.7 y). Zygosity was determined by questionnaire, "16 morphological markers" and in ambiguous cases by serological tests. Subjects filled out questionnaires, and underwent skin prick/intradermal testing, a total sIgE determination and a methacholine challenge.

There was no difference between MZ and DZ twins in casewise concordance for asthma or atopy - not surprising in view of the methods of recruitment, and small numbers. The total variance for total sIgE, methacholine area, and summed skin atopy score did not differ significantly between the MZ and DZ pairs (Table). The intraclass correlations were significantly greater for MZ twins than DZ twins for these three measures (using a modified Fisher Z transformation). This gives heritability estimates for these three measures of 60-70%, with 95% confidence intervals approximately 40-80%, agreeing with the results reviewed earlier. Since the use of intraclass correlations instead of the weighted variance ratios in genetic analyses is inefficient, these analyses have been repeated using LISREL 7.16 (Table). It can be seen that in each case a role for shared environment is not detected (in keeping with the power associated with the sample size).

Table 16. Twin concordance for reported asthma, total sIgE, Area-under-curve for 35% drop in FEV1 on methacholine provocation, and aggregate skin atopy score [from Hopp et al 1984].
Zygosity GroupReported AsthmaIntraclass rNo. of Pairs
+++---sIgEArea35Atopy
MZ99430.820.670.8261
DZ48340.520.380.4246

Table 17. Reanalysis of mean squares presented by Hopp et al [1984]. A=additive genetic component; C=shared environment; E=unique environment.
Model testedh2 (95% CI)X2dfp
Log transformed sIgE
ACE55% (15-95%)0.2710.58
AE*82% (74-100%)1.4720.36
CE0%11.5720.003
Methacholine Area35
ACE58% (4-100%)0.2710.58
AE*66% (52-80%)0.6020.75
CE0%5.6720.06
Skin atopy score (ISTS)
ACE70% (27-100%)0.0210.89
AE*82% (74-90%)0.2820.87
CE0%15.022<0.001

* Preferred model on LR X2 testing compared to "saturated" ACE model.

Minnesota group

This group has reported a study of MZ twins reared apart (MZA) which incorporates a number of physiological and psychological variables. In Hozouri et al [1990], results of skin atopy testing and immunoglobulin determinations were presented for 24 MZA pairs and 1 set of MZA triplets separated soon after birth, 8 pairs of DZ twins reared apart, and 17 pairs of MZ twins and 5 pairs of DZ twins reared together (MZT and DZT). The MZA twins demonstrated significantly higher concordance than the DZ twins for total sIgG and IgM levels, specific antibody titres to pneumococcal carbohydrate and tetanus toxoid. Similar findings were noted for RAST results, but the actual results were not presented. In another study [Hankins et al 1982], the concordance for pulmonary function in 15 pairs of MZA twins and the MZA triplets was shown to be highly significant. The lung function of the smoking twin in smoking discordant pairs differed significantly only in FEF25-75 and instantaneous flow at 75% FVC.

Hanson et al [1991] extend this work, presenting data from 53 MZA twin pairs, 21 DZA pairs, 110 MZT and 111 DZT twins from both Minnesotan and Finnish twin registries. Skin prick testing, total sIgE determination, and RAST to 3 allergens were performed. Concordance for asthma was greater in the MZA group (4 concordant, 1 discordant) than in the other groups. The heritability of logIgE level was 56% (46- 66%), with no shared environmental component of variation being detected. A (nonsignificant) tendency for MZ twins to be more concordant than DZ twins was also seen for skin atopy, but no evidence for concordance for sensitisation to particular allergens was seen.

Clarke et al [2000]:

This Australian study involved a questionnaire survey of 1262 Victorian twin pairs registered with the Australian Twin Registry in 1994. All the twins were born 1974-1986. Useable questionnaires were returned for 1049 pairs (83% pairwise response).

A total of 381 twin pairs went on to undergo an interview by a respiratory physician, a hypertonic saline challenge, skin prick testing to 8 allergens, and zygosity determination via 8 polymorphic STRP markers. The analysis of these phenotypes was conditioned on the responses to the screening questionnaire, allowing for overascertainment of concordant pairs into the clinical phase of the study.

The MZ concordances for physican-diagnosed asthma, atopy and BHR were all significantly higher than those for DZ twins. The MZ:DZ cross-trait correlation differences (expressed as interaction ORs) were also significantly greater than unity. For example, the conditional probability of a diagnosis of asthma was 0.71 if the index person exhibited BHR, 0.64 if the MZ cotwin exhibited BHR, and 0.34 if the DZ cotwin exhibited BHR (prior probability of asthma in this selected sample 0.26).

Table 18.Tetrachoric correlations for twins from Clarke et al [2000]
Within Individual
Asthma BHR Atopy
Asthma 1
BHR 0.79 1
Atopy 0.56 0.78 1
MZ Twins (N=183 pairs)
Asthma 0.86
BHR 0.70 0.77
Atopy 0.45 0.65 0.79
DZ Twins (N=198 pairs)
Asthma 0.24
BHR 0.20 0.50
Atopy 0.19 0.35 0.33

Strachan et al [2001]:

This British study used the St Thomas's UK Adult Twin Registry. A screening questionnaire was completed by 340 MZ female and 533 DZ females (it seems the male participation rate in this volunteer registry was not high enough).

Table 19. Twin concordance for reported asthma, total sIgE [from Strachan et al 2001].
Zygosity GroupReported AsthmaTotal IgE
+++---Tetrachoric rPearson rN Pairs
MZF16622480.430.59282
DZF121173800.080.29270

The heritability of total IgE level is close to that of most studies. Concordance for specific IgE was increased in MZ twins for mixed grasses house dust mite, and cat, but was not significant in the latter group.

Koeppen-Schomerus et al [2001]:

The Twins Early Development Study (TEDS) recruited a cohort of twins born in 1994-95. A questionnaire was mailed at the fourth birthday. A single item, "Have either of your twins been prescribed any medication to control asthma?" measured asthma.

Table 20. Twin concordances for asthma therapy in Koeppen-Schomerus et al [2001]
Zygosity+++---Probandwise concordance
MZF991046750.66
MZM1001135670.65
DZF461805780.34
DZM711756010.45
DZX10239111080.34

A threshold model estimated the heritability at 68% (95%CI=50-85%), and the shared environmental component at 13% (0-29%).

Recent minor studies

Svartengren et al on tracheobronchial clearance:

This small study [1989] examined tracheobronchial clearance in 6 MZ pairs discordant for asthma, and 3 nonasthmatic MZ pairs. Tracheobronchial clearance has been noted to be decreased in some studies of asthmatics compared to controls, but considerable within group variation has been noted. Conveniently, MZ concordance in normal MZ twins for tracheobronchial clearance is high in previous studies by the same group. Clearance of radiolabelled Teflon particles from the lungs was measured twice, as was resting Raw and FEV1 and a methacholine provocation test.

The two hour particle retention value (Ret2) for the asthmatic twin was greater than the nonasthmatic cotwin's in 3 out of the 6 pairs, and the overall differences were not significant. Clearance was slowed following methacholine challenge in the nonasthmatics but not in the asthmatics. In addition, Ret2 was positively correlated with PD20 in the nonasthmatics, but not the asthmatics.

Sistonen et al [1980]:

Total sIgE was measured in a random sample of 76 MZ and 81 DZ twins from the Finnish Twin Register [1980]. Twin concordance for high levels (>150 IU/ml) were higher in the MZ than the DZ group, but not significantly so (X2(2 df)=3.65). The female MZ intrapair variance for log sIgE was significantly smaller than the DZ intrapair variance, but no significant difference was exhibited by the males.

Ericsson et al [1991]:

Bronchial reactivity in 8 pairs of MZ twins discordant for allergic rhinitis was studied in and out of pollen season [1991]. The allergic twin, unsurprisingly, had higher reactivity on average than their cotwin, and more interestingly, tended to have a lower birth weight.

Bonini et al [1983]:

Skin prick tests, total and specific sIgE, sIgG, BHR and complement in 55 MZ and 20 DZ pairs ascertained for atopy (from undescribed sources) were examined in this study [1983]. For SPTs, the casewise (identification of this fact based on the concordance values quoted) concordance for MZ twins was 0.43, and DZ twins, 0.19 (X2(1 df)=2.8, P=0.09). No significant differences for BHR concordance were detected.

Bahna et al:

In 10 MZ and 22 DZ pairs of twin children (mean age 5 years) ascertained for clinical atopy, with MZ casewise concordance 0.89, and DZ, 0.74. No concordance for sensitisation to particular allergens was detected.

Marone et al [1986]:

These authors briefly describe [1986] a twin study involving 14 pairs of MZ and 12 pairs of DZ twins. It is not stated if any of the twins suffered from clinical allergic disease. Maximum anti- IgE-induced basophil histamine release was significantly correlated in the MZ twins (r=0.84), but not the DZ twins (r=0.16). The heritability under an additive genetic model (X2(1 df)=1.0) was 0.88. A similar picture was seen with the Ca ionophore A23187. There was no correlation between maximum histamine release and total sIgE level in the twins, although in a separate group of singletons aged less than twenty years, the correlation was 0.81. Deuterium oxide and f-met peptide stimulated histamine release were "influenced mainly by environmental factors".

Miles et al [1992]:

In an abstract for the European Thoracic Society 1992 meeting [1992], these authors describe a Birmingham questionnaire study of 190 pairs of twins. MZ odds ratios are significantly higher than DZ (Table). Model fitting in MX found evidence for heterogeneity in prevalences at the different ages (homogeneity X2(3 df)=19.88-7.55=12.33) but estimating different thresholds, the same AE model fitted both age groups (homogeneity X2(3 df)=7.55-7.55) giving a heritability of 85%.

Table 21. Twin concordances for asthma in Miles et al [1992]
Age/Zygosity++-+--Tetrachoric r (se)OR
<15 y.o.
MZ99800.86 (0.08)35.5
DZ3 18710.30 (0.23)2.6
>14 y.o.
MZ1212680.84 (0.08)22.7
DZ1122350.41 (0.18)3.2

Sicherer et al [2000]:

This study ascertained twins suffering from peanut allergy through the Food Allergy Network (a patient support organisation). Of 75 pairs, 58 contained at least one member with a "convincing" history of peanut allergy (52/70 probands reported a positive RAST or skin prick test). MZ twins were significantly more concordant than DZ twins (P=0.00003), and threshold models gave understandably high heritability estimates.

Table 22. Twin concordances for peanut allergy in Sicherer et al [2000]
Zygosity+++-Probandwise concordance
MZ950.79 (0.52--0.93)
DZ3410.13 (0.03--0.31)

Atopic dermatitis

Two studies of atopic dermatitis have been performed using data from the Danish Twin Registry. Larsen et al [1986] surveyed 97% of all like-sex twins born 1960-74 in one county - 592 pairs. The initial screening questionnaire contained an item "Have you or have you had any of the following diseases?" - "infantile eczema", asthma and hayfever. This was returned by 578 pairs (!). All twins who responded "Yes" or "Don't know" to the infantile eczema item were invited for clinical interview and examination, plus 48 pairs of age-sex-zygosity matched item-negative pairs.

Of the 76 individuals who reported eczema, 61 were considered confirmed cases, and another four were possible cases (Table). Of those who were uncertain, four out of twelve were diagnosed as definitely affected, and a further four as possibly affected. None of the item-negative twins were found to suffer atopic dermatitis. This gives a sensitivity of 94% and specificity of 87% for the screening item similar to that used in the present study. All but two cases had an onset before seven years of age.

Table 23. Validation of questionnaire diagnosis of atopic dermatitis in Danish twins [Larsen et al 1986].
Questionnaire DiagnosisClinical Diagnosis
YesPossibleNo
Yes61411
Unsure444
No0096

The results gave quite high heritabilities according to the tabulations of Smith [1974] for the threshold model. The high MZ/DZ concordance ratio of 5.0 (95% CI 2.4-12.6) means that it is likely epistasis is present (that is that the MFT model may be particularly appropriate). Reanalysing this data under the MFT using the program MX under the assumption that all true cases have been detected using this design (Table), I obtain similar heritability estimates (h2 at upper bound, d2=0.01, e2=0.02). The binomial heritability was 86% (epistasis was not required - X2(1 df)=0.11 for the dominance model).

Table 24. Full fourfold tables constructed from Larsen et al [1986]*.
ZygosityCooccurrence in TwinsPhiTetrachoric r
+++---
MZ1753930.870.99
DZ4223250.230.56

* Twins of uncertain zygosity have been allocated equally into the MZ and DZ groups.

Larsen [1993] reports an extension of this study to twins born up until 1979. This relied solely on the questionnaire response to an item on "infantile eczema", using the results cited above to validate this decision. The response (after three mailings) was 92%. The casewise concordance for questionnaire diagnosed atopic dermatitis was 26/36 (0.72) for definite MZ twins, and 16/70 (0.23) for definite DZ twins. The full fourfold table (Table) for those of definite zygosity gives essentially the same results as before. The binomial heritability is 80%.

Table 25. Full fourfold tables constructed from Larsen et al [1993].
ZygosityCooccurrence in TwinsPhiTetrachoric r
+++---
MZ26101750.810.97
DZ16543600.300.57

Conclusions

The family, twin and population studies reviewed show conclusively that asthma, hayfever and atopic dermatitis are genetically correlated, quite aside from the similarities in pathogenesis. Furthermore, the heritability of these atopic diseases is quite high. A role for a major gene in their aetiology is also highly likely, but genetic heterogeneity and/or a role for polygenes is supported by the complex segregation, molecular and biochemical genetic studies. A number of the biochemical genetic results have not yet been replicated, while others are weak and inconsistent. Atopic and nonatopic asthma, in some important studies, have been seen as genetically similar, if not identical. The entity of late onset asthma, with its tendency to greater severity, usual female preponderance, and supposed lesser association with family history of atopic disease has not really been properly addressed, and the possibility that this syndrome is genetically distinct from childhood onset disease needs further examination.