Human Genetics

, Volume 113, Issue 5, pp 437–446

Sex differences in heritability of sensitization to Blomia tropicalis in asthma using regression of offspring on midparent (ROMP) methods

Authors

    • Division of Epidemiology and Clinical ApplicationsNational Heart, Lung,and Blood Institute
  • Kathleen C. Barnes
    • Department of MedicineJohns Hopkins University School of Medicine
  • Terri H. Beaty
    • Department of EpidemiologyJohns Hopkins University Bloomberg School of Public Health
  • Paul N. Levett
    • Division of Bacterial and Mycotic DiseasesCenters for Disease Control and Prevention
  • Raana P. Naidu
    • School of Clinical Medicine and ResearchUniversity of the West Indies
  • Alexander F. Wilson
    • Inherited Disease Research BranchNational Human Genome Research Institute
Original Investigation

DOI: 10.1007/s00439-003-1005-6

Cite this article as:
Manolio, T.A., Barnes, K.C., Beaty, T.H. et al. Hum Genet (2003) 113: 437. doi:10.1007/s00439-003-1005-6

Abstract

A genetic basis for asthma- and atopy-related quantitative traits, such as allergen-specific immunoglobulin E (IgE) levels, has been suggested by the observed familial aggregation of these traits in temperate climates. Less information is available for tropical climates, where different allergens may predominate. Sensitivity to the mite Blomia tropicalis is related to asthma in tropical climates, but heritability of B. tropicalis sensitivity and the impact of age, sex, and other environmental covariates on heritability have not been widely explored. Total and specific IgE levels were measured by immunochemiluminescent assay in 481 members of 29 Barbadian families (comprised of 340 parent-offspring trios or pairs) ascertained through two asthmatic siblings. Trait heritability was estimated using regression of offspring on mid-parent (ROMP) and pairwise correlation analysis of unadjusted IgE levels and on residual values after adjustment for covariates. Heritability of IgE levels to the major antigen of B. tropicalis (Blo t M) estimated by ROMP in 180 complete parent-offspring trios was 0.56. Heritability was consistently greater for male offspring than for female offspring. Similar sex-specific patterns were observed for specific IgE to Dermatophagoides pteronyssinus and total IgE levels and were relatively unaffected by adjustment for covariates. Pairwise correlational analyses of specific and total IgE levels showed similar results. Moderate heritability of Blo t M IgE levels was detected in these Barbadian families and was greater for sons than daughters. Adjustment for covariates had minimal impact. This suggests that future investigations of genetic determinants of IgE levels should include approaches that allow for potential sex differences in their expression.

Introduction

Atopic asthma is a chronic inflammatory disease of the lower airways caused by an immunoglobulin E (IgE)-mediated reaction to environmental allergens, especially those associated with house dust (National Asthma Education Program 1991; Sporik et al. 1992). A genetic basis for asthma- and atopy-related quantitative traits, such as allergen-specific IgE levels (particularly to the dust mite Dermatophagoides pteronyssinus), has been suggested by the demonstration of familial aggregation of these traits in temperate climates (Duffy 1997). Less information is available in tropical climates, however, where different allergens may predominate and degree of exposure to certain allergens may be greater (Puerta et al. 1996). Sensitivity to the mite Blomia tropicalis is related to asthma in tropical climates, but heritability of B. tropicalis sensitivity and the impact of age, sex, and environmental covariates on heritability have not been widely explored. Assessment of heritability is a necessary first step in examining potential genetic influences on a trait.

Environmental influences on specific IgE levels have been proposed to be greater than those influencing overall IgE levels (Marsh et al. 1995), with greater environmental variance demonstrated for response to dust mite allergens (Young et al. 1992). Specific IgE levels have been shown to be related to age, current smoking and occupational exposures (Omenaas et al. 1994). Blo t M IgE levels have been shown to be associated with asthma, younger age, cigarette smoking, and housing construction in these Barbadian families (Manolio et al. 2003). Heritability of specific IgE responses has been estimated in prior studies to be roughly 15–20%, much lower than the 50–80% reported for total IgE levels (Blumenthal 2000).

Barbados is a tropical country with a high prevalence of B. tropicalis sensitization (Manolio et al. 2003). Data from the Hopkins-Barbados Study on the Genetics of Asthma and Allergy were analyzed to: (1) determine the heritability of IgE level to the major antigen of this mite (Blo t M) in 481 members of 29 Barbadian families ascertained through two asthmatic siblings, and compare it with heritability of IgE levels to the major antigen of D. pteronyssinus (Der p 1) and to total IgE levels; (2) examine the impact of age, sex, and environmental factors on heritability estimates; and (3) assess sex-specific differences in heritability before and after adjustment for these factors.

Materials and methods

Ascertainment of families

Asthmatic probands were identified from patients with a documented history of asthma seen in public and private clinics and in the Accident and Emergency Department of Queen Elizabeth Hospital in Barbados (Barnes et al. 1996). Patients with a positive family history of asthma or one or more asthmatic siblings were referred for study by physicians cooperating with on-site study investigators at the University of the West Indies. After giving informed consent, available first-degree relatives and other family members were recruited into the study between April 1993 and December 1996. The study protocol was approved by the Johns Hopkins Medical Institutions' Joint Committee on Clinical Investigations (JCCI) and the Ministry of Health in Barbados. Data were available on 481 subjects in 29 extended families, including 45 asthmatic probands, 94 asthmatic relatives and 342 non-asthmatic relatives. All subjects but one were Afro-Caribbean (race self-reported as 'Black').

Data collection

Family structure and family history information were obtained by self-report from a designated family member, generally the proband or proband's parent, and refined and verified by interview of other family members. Interviewer-administered questionnaires (Barnes et al. 1999; Blumenthal et al. 1995) were used to collect information on tobacco use, housing construction, and other covariates from adults or through proxy report by a responsible adult for children under age 12. Smoking was reported as current, past, or never and housing construction as wood, mixed wood/concrete, or concrete. Information on environmental covariates was missing for 11% of subjects. Asthma was defined as previously described (Barnes et al. 1996) and included a history of asthma and current asthma symptoms reported on a standardized interviewer-administered questionnaire, and confirmation of the diagnosis by a co-investigator who was one of the primary diagnosticians of asthma in Barbados.

Phlebotomy was performed under aseptic conditions in a non-fasting state. Blood was drawn into vacutainers and centrifuged to separate plasma and serum. Samples were shipped on ice to the Johns Hopkins Asthma and Allergy Center for IgE measurements and extraction of DNA. Allergen-specific and total serum IgE levels were measured in duplicate using the immunochemiluminometric Magic Lite assay (Magic Lite Total IgE Extended Range; CIBA-Corning, Medfield, Mass.) (Barnes et al. 1996). The assay had a minimum detectable concentration of 1.43 standardized units (U) per ml and correlated with radioimmunoassay methods with a correlation coefficient of 0.82 and a slope and intercept of 5.33 and 0.17 U/ml, respectively (Ciba-Corning 1990). Intra-assay coefficients of variation (CV) ranged from 6.2 to 8.9% across concentrations from <4 to 800 U/ml, with CVs being greater at the lower end of concentration. Total CVs ranged from 8.0 to 16.0%, again being greatest at low concentrations (Ciba-Corning 1990). Despite the known 43% sequence homology between Blo t 5 and Der p 5 (a minor antigen of D. pteronyssinus), cross-reactivity of Blo t 5 and Der p 1 is minimal due to the lack of similarity between the major antigens of the two mites (Chew et al. 1999). Sensitization to Blo t M and Der p 1 was defined as any detectable IgE to these antigens.

Statistical methods

Total and allergen-specific IgE levels were log10-transformed and analyzed as continuous variables. Heritability was estimated in nuclear families (defined as unique spousal pairs with one or more offspring identified within the 29 extended pedigrees) as the slope of the regression of offspring on midparental value (Fisher 1918) using the regression of offspring on mid-parent (ROMP), as implemented by Wilson and co-workers (Pugh et al. 2001). If phenotypic values for one parent were missing, the regression was performed on the remaining parent and the resulting slope was doubled. Comparability of this approach was assessed in offspring with data on both parents by regressing on the value of one randomly selected parent and doubling the resulting slope. Heritability attributable to maternal and paternal effects was estimated separately using offspring-father and offspring-mother regressions. Sex-specific differences were assessed by stratifying regressions by sex of offspring, with and without stratification by sex of parent. Significance of differences in heritability by sex was tested by including an interaction term between sex and parental IgE levels.

Associations of covariates with log-transformed IgE levels were assessed using t-tests for discrete variables, analysis of variance for categorical variables, and linear regression and correlation for continuous variables. Age, current smoking, and mixed wood/concrete or concrete housing construction were associated (all P<0.05) with log-transformed IgE levels in multiple linear regression analysis. Interaction terms of age with sex and with current smoking were used to test differences in age coefficients by these covariates, but were not significant. Adjustment for covariates was accomplished by using residuals of the linear regression of IgE on age alone; age and sex; and age, sex, and environmental covariates (smoking and housing construction) as the trait for analysis.

The change in heritability after inclusion of covariates was assessed by comparing heritabilities of IgE levels adjusted as described above to heritabilities estimated from analysis of unadjusted values. Although the age relationship with IgE levels was not strictly linear and quadratic and cubic terms were significant, their inclusion did not appreciably change point estimates or significance of heritability. For simplicity, only linear models were reported here. Analyses were performed using the Statistical Analysis System (SAS) version 6, Cary, N.C. (SAS Institute 1990).

Heritability was also estimated using pairwise correlations generated from the familial correlations (FCOR) program of S.A.G.E. (1997), including six maintype correlations (marital, parent-offspring, sibling, half-sibling, grandparental, avuncular, cousin) and seven sex-specific subtype correlations (mother-daughter, mother-son, father-daughter, father-son, sister-sister, sister-brother, and brother-brother). Correlations were estimated by giving equal weight to pairs. Adjustment for covariates was performed by regressing IgE levels on these factors and using the residuals as the phenotype. Heritability was estimated as twice the parent-offspring correlation (Fisher 1918).

Results

Two hundred and sixty-one participants (54%) were sensitized to Blo t M (i.e., had detectable Blo t M IgE levels) as reported previously (Manolio et al. 2003). Mean levels of Blo t M IgE estimated from log-transformed values were 8.2 U/ml (range 0–3,655 U/ml). Sensitization to Der p 1 was detectable in 129 (27%) subjects with mean levels of 0.32 U/ml (range 0–120 U/ml). All but three subjects had measurable total IgE levels with mean values of 364 U/ml (range 0–17,565 U/ml). Smoking was uncommon, but was more frequent in men than women; total IgE levels were also higher in men (Table 1). Other covariates and IgE levels did not differ by sex.
Table 1.

Major covariates and IgE levels by sex in 481 subjects

Variable

Women

Men

P Value

n=257

n=224

Age (years)

28.0

27.2

0.6

Asthma (n, %)

73 (28.4)

66 (29.5)

0.8

Concrete/mixed housing (n, %)a

168 (70.9)

149 (78.0)

0.10

Current Smoking (n, %)a

3 (1.3)

14 (7.3)

0.001

Total IgEb (units)

326.1

426.3

0.09

B. tropicalis IgEb (U/ml)

7.4

9.5

0.3

D. pteronyssinus IgEb (U/ml)

0.33

0.32

0.6

aInformation available in 89% of subjects

bGeometric mean levels shown for IgE levels, comparison of means performed with log-transformed values

A total of 340 parent-offspring trios or pairs from 169 nuclear families had Blo t M IgE levels measured and were included in this analysis. Of these 169 nuclear families more than half (88) had a single offspring, 30 had two, 25 had three, 16 had four, nine had five, and one had six offspring. A little more than half the offspring (185 of 340, or 55%) were daughters, while over 90% (142 of 152) of parents in single parent-offspring pairs were mothers. The mean and median number of nuclear families included in an extended pedigree were 6.6 and 6, respectively, with three pedigrees yielding one nuclear family each and six pedigrees yielding ten families or more.

Of the 340 parent-offspring trios or pairs, 236 or 70% had complete information on age, sex, housing construction, and smoking for offspring and at least one parent. Offspring from pairs or trios with complete data did not differ by sex, housing construction or smoking but were younger (19 vs 24 years, P< 0.002) and had higher geometric mean Blo t M IgE levels (15.1 vs 6.9 U/ml) as well as higher levels of Der p 1 and total IgE compared with offspring without complete data (P<0.02 for all differences).

Heritability using ROMP

The estimated heritability (slope of the linear regression of offspring on mid-parental Blo t M IgE value) for 180 complete parent-offspring trios was 0.56 (P=0.0002). The estimates of heritability in these trios were greater for male (0.97) than for female (0.18) offspring (Table 2). Levels of IgE in sons increased more steeply with increasing parental levels than did IgE in daughters (Fig. 1). The same was true of Der p 1 IgE and total IgE. Similar patterns obtained when one parent was selected at random from complete trios (data not shown). Regressions of offspring on single parent values were performed and doubled to estimate heritability for incomplete trios (Table 3). Patterns were less consistent, with heritability being less for sons than daughters for Blo t M, greater for sons than daughters for Der p 1, and equal for total IgE.
Table 2.

Regression of offspring log-transformed IgE level on mid-parental value for 180 complete parent-offspring trios

Independent variable

n

Heritability

P value

Blo t M IgE

  All Offspring

180

0.56

0.0002

  Male offspring

83

0.97

0.0001

  Female offspring

96

0.18

0.4

Der p 1 IgE

  All Offspring

180

0.33

0.05

  Male offspring

83

0.51

0.02

  Female offspring

96

0.11

0.7

Total IgE

  All offspring

180

0.35

0.0002

  Male offspring

83

0.52

0.0001

  Female offspring

96

0.18

0.3

Fig. 1.

Offspring Blo t M IgE level plotted against mid-parental value for all offspring (solid line), sons alone (dashed line), and daughters alone (dotted line)

Table 3.

Regression of offspring log-transformed IgE level on single available parent for 155 incomplete trios

Independent variable

n

Heritability

P value

Blo t M IgE

  All offspring

155

0.13

0.7

  Male offspring

69

−0.17

0.8

  Female offspring

85

0.34

0.4

Der p 1 IgE

  All offspring

155

0.72

0.02

  Male offspring

69

1.00a

0.002

  Female offspring

85

0.21

0.7

Total IgE

  All offspring

155

0.44

0.03

  Male offspring

69

0.54

0.10

  Female offspring

85

0.41

0.09

aPoint estimates of heritability exceeding 1.0 were truncated at 1.0.

To determine whether heritability demonstrated sex-specific differences, regressions were stratified by sex of offspring, parent, or both (Table 4). Heritability was consistently greater in male than female offspring, regardless of sex of the parent or type of IgE assessed. None of the heritability estimates in female offspring differed significantly from zero, while all but one did in male offspring. Testing of interactions with sex in models with all offspring showed only Blo t M IgE heritability to differ significantly by sex (P<0.007).
Table 4.

Regression of offspring log-transformed IgE level on paternal and maternal IgE levels

Independent variable in regression

n

Heritability

P value

Paternal Blo t M IgE

  All offspring

196

0.58

0.01

  Male offspring

91

0.73

0.02

  Female offspring

105

0.32

0.4

Maternal Blo t M IgE

  All offspring

322

0.56

0.0005

  Male offspring

147

0.87

0.0003

  Female offspring

175

0.28

0.19

Paternal Der p 1 IgE

  All offspring

196

0.15

0.6

  Male offspring

91

0.40

0.3

  Female offspring

105

−0.10

0.8

Maternal Der p 1 IgE

  All offspring

322

0.57

0.002

  Male offspring

147

0.75

0.0008

  Female offspring

175

0.34

0.3

Paternal total IgE

  All offspring

196

0.34

0.03

  Male offspring

91

0.47

0.02

  Female offspring

105

0.16

0.5

Maternal total IgE

  All offspring

322

0.83

0.0002

  Male offspring

147

0.59

0.0002

  Female offspring

175

0.25

0.10

Recognizing the strong associations of IgE levels with age, and the substantial age differences inherent between parents and offspring, these regressions were repeated after adjustment for age. Substitution of age-adjusted residual Blo t M IgE as the trait for analysis reduced the heritability estimate based on mid-parental values minimally, from 0.56 to 0.52 (Table 5, column 1). Age adjustment also had minimal impact on heritability estimates based on paternal and maternal Blo t M IgE, regardless of the sex of the offspring. Age adjustment of Der p 1 IgE and total IgE regressions had similarly minimal impact on heritability estimates, regardless of sex of the parent or offspring. Significance of heritability estimates using age-adjusted residuals was nearly identical to that of unadjusted values for all three IgE levels examined; no estimate gained or lost significance following age adjustment. Further adjustment for sex and additional adjustment for housing construction and smoking also had little and inconsistent impact on estimates or significance of heritability (Table 5, columns 2, 3).
Table 5.

ROMP heritability estimates adjusted for age; age and sex; age, sex and environmental correlates

Independent variable

Age

+ Sex

+ Environmental correlates

h2

(P value)

h2

(P value)

h2

(P value)

Mid-parental Blo t M IgE

  All offspring

0.52

(0.0002)

0.51

(0.0003)

0.43

(0.001)

  Male offspring

0.82

(0.0001)

0.82

(0.0001)

0.67

(0.0005)

  Female offspring

0.21

(0.3)

0.21

(0.3)

0.23

(0.3)

Paternal Blo t M IgE

  All Offspring

0.58

(0.007)

0.57

(0.008)

0.44

(0.06)

  Male Offspring

0.70

(0.02)

0.70

(0.02)

0.55

(0.10)

  Female Offspring

0.39

(0.3)

0.39

(0.3)

0.30

(0.4)

Maternal Blo t M IgE

  All offspring

0.42

(0.007)

0.43

(0.006)

0.46

(0.007)

  Male offspring

0.73

(0.002)

0.73

(0.002)

0.80

(0.002)

  Female offspring

0.17

(0.5)

0.17

(0.5)

0.23

(0.4)

Mid-parental Der p 1 IgE

  All offspring

0.31

(0.056)

0.32

(0.056)

0.26

(0.11)

  Male offspring

0.44

(0.04)

0.44

(0.04)

0.37

(0.08)

  Female offspring

0.15

(0.6)

0.15

(0.6)

0.08

(0.8)

Paternal Der p 1 IgE

  All offspring

0.20

(0.4)

0.21

(0.4)

0.08

(0.7)

  Male offspring

0.39

(0.18)

0.39

(0.18)

0.29

(0.4)

  Female offspring

0.01

(0.9)

0.01

(0.9)

‑0.14

(0.7)

Maternal Der p 1 IgE

  All offspring

0.49

(0.007)

0.49

(0.006)

0.52

(0.006)

  Male offspring

0.67

(0.003)

0.67

(0.003)

0.72

(0.005)

  Female offspring

0.25

(0.4)

0.25

(0.4)

0.28

(0.4)

Mid-parental total IgE

  All offspring

0.30

(0.002)

0.33

(0.0004)

0.38

(0.0001)

  Male offspring

0.49

(0.0001)

0.49

(0.0001)

0.44

(0.0004)

  Female offspring

0.20

(0.16)

0.20

(0.16)

0.31

(0.05)

Paternal total IgE

  All offspring

0.32

(0.04)

0.32

(0.04)

0.27

(0.18)

  Male offspring

0.45

(0.02)

0.45

(0.02)

0.41

(0.10)

  Female offspring

0.18

(0.5)

0.18

(0.5)

0.08

(0.8)

Maternal total IgE

  All offspring

0.39

(0.0003)

0.38

(0.0003)

0.40

(0.0007)

  Male offspring

0.57

(0.0003)

0.57

(0.0003)

0.62

(0.0002)

  Female offspring

0.24

(0.11)

0.24

(0.11)

0.24

(0.16)

Heritability using pair-wise correlation methods

Correlations of unadjusted log Blo t M IgE levels were calculated for 71 spouse pairs, 518 parent-offspring pairs, 326 sibling pairs and 200–400 other types of relative pairs (Table 6, column 1). Spouse correlations for all three IgE levels were <0.06, suggesting minimal correlation due to shared environment. The interclass correlation of log Blo t M IgE levels in 518 parent-offspring pairs was 0.19, yielding a heritability estimate of 0.38. The intraclass correlation among siblings was similar at 0.16; among half-siblings, 0.10; and among cousins, 0.16. Interclass correlations of log Blo t M IgE among second-degree relatives (grandparent-grandchild and avuncular pairs) were near zero.
Table 6.

Familial correlations of log IgE levels, unadjusted and adjusted for age; age and sex; and age, sex, and environmental correlates. Numbers of pairs: spouse (71), parent-offspring (518), sibling (326), half-sibling (213), cousin (338), grandparent-grandchild (223), avuncular (436), mother-daughter (175), mother-son (147), father-daughter (105), father-son (91), sister-sister (99), sister-brother (172), brother-brother (55)

Relative pair

Adjusted for

Unadjusted

Age

+ Sex

+ Environmental correlates

Blo t M IgE

  Spouse

0.05

0.07

0.06

0.09

  Parent-offspring

0.19

0.17

0.17

0.17

  Sibling

0.16

0.10

0.10

0.07

Half-sibling

0.10

0.09

0.09

0.05

Cousin

0.16

0.15

0.15

0.05

Grandparent-grandchild

−0.01

−0.01

−0.01

0.06

Avuncular

−0.01

−0.03

−0.03

0.06

Mother-daughter

0.10

0.06

0.06

0.09

Mother-son

0.29

0.26

0.26

0.29

Father-daughter

0.10

0.12

0.12

0.10

Father-son

0.25

0.26

0.26

0.21

Sister-sister

0.04

−0.02

−0.02

−0.09

Sister-brother

0.23

0.16

0.16

0.13

Brother-brother

0.17

0.12

0.12

0.11

Der p 1 IgE

Spouse

−0.06

−0.06

−0.06

−0.13

Parent-offspring

0.13

0.12

0.12

0.12

Sibling

0.20

0.18

0.18

0.18

Half-sibling

−0.00

−0.01

−0.01

−0.06

Cousin

−0.01

−0.01

−0.01

0.01

Grandparent-grandchild

0.08

0.10

0.10

0.07

Avuncular

0.03

0.04

0.04

0.02

Mother-daughter

0.09

0.07

0.07

0.09

Mother-son

0.27

0.24

0.24

0.26

Father-daughter

−0.03

0.00

0.00

−0.05

Father-son

0.13

0.14

0.14

0.12

Sister-sister

0.14

0.13

0.13

0.10

Sister-brother

0.26

0.24

0.24

0.26

Brother-brother

0.06

0.07

0.07

0.04

Total IgE

Spouse

0.03

0.05

0.05

−0.03

Parent-offspring

0.18

0.18

0.18

0.17

Sibling

0.25

0.22

0.22

0.19

Half-sibling

0.05

0.05

0.05

0.04

Cousin

0.13

0.14

0.14

0.00

Grandparent-grandchild

−0.01

0.01

0.01

0.10

Avuncular

0.07

0.06

0.06

0.08

Mother-daughter

0.13

0.12

0.12

0.12

Mother-son

0.30

0.30

0.30

0.28

Father-daughter

0.07

0.08

0.08

0.03

Father-son

0.26

0.25

0.25

0.22

Sister-sister

0.19

0.12

0.12

0.01

Sister-brother

0.27

0.25

0.25

0.25

Brother-brother

0.29

0.27

0.27

0.23

Sex-specific correlations of Blo t M IgE levels again demonstrated stronger correlations between parents and their sons (>0.25) than between parents and their daughters (both=0.10), regardless of the sex of the parent. The correlation between sisters was near zero, while sister-brother and brother-brother correlations were 0.17–0.23.

Correlations for Der p 1 IgE and total IgE were more variable, though sibling correlations tended to be stronger than for Blo t M IgE. Sex-specific parent-offspring correlations showed the same pattern of stronger parent-son than parent-daughter correlations regardless of the sex of the parent. Sex-specific sibling correlations showed no consistent pattern, with Der p 1 correlations being weakest in brother-brother pairs and total IgE correlations of approximately equal magnitude in all three types of sib pairs.

Use of age-adjusted residual values in familial correlation analyses reduced the sibling correlation from 0.16 to 0.10 for Blo t M IgE but had little impact on other familial correlations (Table 6, column 2). Age adjustment had little impact on correlations among second-degree relatives or on sex-specific correlations. Greater correlations remained in sons than daughters but correlation of sex-specific sib pairs, particularly those including brothers, was reduced for Blo t M IgE levels after age adjustment. Further adjustment for sex increased overall sibling correlations and correlations for pairs involving brothers for Blo t M IgE levels but otherwise had little effect (Table 6, column 3).

Further adjustment for environmental covariates tended to reduce correlations somewhat but the pattern of stronger correlations in relative pairs that included sons and brothers remained (Table 6, column 4). Adjustment had minimal impact on Der p 1 IgE and total IgE correlations except for eliminating the sister-sister correlation for total IgE. Sex-specific patterns were otherwise similar to those for unadjusted levels.

Discussion

Summary of findings

Heritability of Blo t M IgE levels using ROMP was estimated at 0.56, with greater heritability demonstrated for male offspring than for female offspring, regardless of the sex of the parent. Similar sex-specific patterns were observed for Der p 1 IgE, as well as total IgE levels, and were relatively unaffected by adjustment for age, sex, or environmental correlates. As expected, pairwise correlational analysis gave similar estimates of heritability, similar sex-specific patterns, and similar minimal impact of adjustment on these patterns.

Heritability of total and specific IgE levels

Previous reports of heritability of IgE levels have focused on total IgE and were derived almost exclusively from twin studies and variance components analysis (Blumenthal 2000; Duffy 1997; Holloway et al. 1999; Tovey et al. 1998). Total IgE levels have been shown to be heritable (Duffy 1997), with estimates of heritability ranging from 50–80% (Blumenthal 2000; Burrows et al. 1995; Duffy 1997; Hasstedt et al. 1983; Holloway et al. 1999; Jacobsen et al. 2001; Meyers et al. 1987; Tovey et al. 1998). Recent analyses from a founder population, the Hutterites, have reported heritabilities of 63% (Abney et al. 2001), with age and sex as important covariates (Ober et al. 2001), though these estimates might be expected to be upwardly biased due to the Hutterites' high degree of environmental homogeneity (Ober et al. 2001). Palmer and co-workers estimated heritability of total IgE at 47% and demonstrated significant age and sex effects using 232 Australian Caucasian nuclear families. In that analysis, additive genetic effects were independent of age, sex, smoking, and height (Palmer et al. 2000a, 2000b).

Heritability of specific IgE levels has been less extensively studied but has been suggested to be lower than that of total IgE (Blumenthal 2000; Tovey et al. 1998) and to be explained partly, but not entirely, by overall IgE hyperresponsiveness (Palmer et al. 2000a; Tovey et al. 1998). Heritability of specific IgE, assessed as a combination of response to whole house dust mite (D. pteronyssinus) and Timothy grass (Phleum pratense), was estimated at 34%, also with significant age and sex effects (Palmer et al. 2000b). Strachan and co-workers reported a total IgE heritability of 60% among adult British twins but only reported specific IgE responses as concordance rates for dichotomized phenotypes (Strachan et al. 2001). Although concordance for specific responses was consistently higher for monozygotic than dizygotic twins in that study, the proportion of co-twins sensitized to the same allergen was higher in monozygotic twins only for grass pollen, suggesting only weak support for genetic influences on allergen-specific sensitization. Heritability of Blo t M IgE levels, and its relationship to heritability of Der p 1 and total IgE levels in the same sample, do not appear to have been reported.

ROMP is based on a linear model that is a hybrid of the traditional regression of offspring on mid-parent model used to estimate heritability and the analysis of variance (ANOVA) model used to test for association. ROMP is completely non-parametric with respect to the underlying genetic model; it produces a moment-based, rather than a likelihood-based, statistic; it is fast and computationally simple; and it requires no specialized genetic analysis software. The method is not limited to the regression of offspring on parent values but can be modified for other relatives as well. There is some decrease in power due to non-independence of offspring observations, but the decrease is fairly small.

The methodology for estimating trait heritability using twins (ANOVA) is quite similar to ROMP, although it is limited to twins. Similarly, path analysis could be used to estimate heritability in other relatives and the ROMP framework for testing for a covariate that is in fact the candidate locus could easily be applied. Both of these methods, like ROMP, are moment-based. Likelihood-based variance component methods produce an estimate of trait heritability that is essentially identical to that produced in ROMP, but the estimates of locus-specific heritability — i.e., the variance attributable to a specific genetic effect — are overestimated by as much as 256% with the likelihood-based variance components method (Pugh et al. 2001).

Relationship of age and sex to IgE levels

Studies of asthma and atopy in the 1980s showed total IgE levels to be related to age, smoking, and skin-test reactivity to allergens (Burrows et al. 1981, 1989). Total IgE levels increase from infancy through roughly age ten and then decline in adolescence (Barbee et al. 1981; Grundbacher et al. 1985). Serum levels are inversely correlated with age in adulthood (Annesi et al. 1992; Omenaas et al. 1994), though these relationships may vary by smoking status, with less of an age-related decline for smokers than non-smokers (Omenaas et al. 1994; Sherrill et al. 1994). Sex differences in total IgE levels have been reported in some (Freidhoff et al. 1984; Jarvis et al. 1995; Omenaas et al. 1994; Oryszczyn et al. 2000), but not all (Orren et al. 1975; Villar et al. 1994; Warren et al. 1982), studies, with men typically having higher levels than women, especially at young ages. Sex differences remain poorly explained but have been attributed to higher prevalence of smoking (Sherrill et al. 1994) or occupational exposures (Omenaas et al. 1994) among men, both of which are related to increased IgE levels and allergic sensitization (Burrows et al. 1981, 1989; Shirakawa et al. 1992, 1993). As previously reported, IgE levels in these Barbadian families were related to age, sex, current smoking and housing construction (Manolio et al. 2003). Mixed wood/concrete and concrete housing in Barbados are typically associated with indoor plumbing, which may in turn produce higher humidity levels and greater mite proliferation (R. Naidu, personal communication). Higher IgE levels in black vs white subjects have been reported repeatedly (Grundbacher et al. 1985; Orren and Dowdle 1975; Orren et al. 1975), and were demonstrated previously in the Barbadian sample used here when compared with a similar family study of white subjects (Barnes et al. 1996). Less information is available on determinants of specific IgE levels, but they appear similar to those reported for total IgE (Jarvis et al. 1995; Omenaas et al. 1994; Plaschke et al. 1996).

Sex-specific heritability of total and specific IgE levels

Sex-specific heritability estimates in the present study demonstrated intriguing differences by sex, with estimates for male offspring and siblings being greater than corresponding estimates for female offspring and siblings. This is not compatible with sex-linked inheritance, as a trait locus on the X chromosome would be expected to produce greater correlations between fathers and daughters than between fathers and sons, since the latter do not share an X chromosome. Conversely, a trait locus on the Y chromosome is unlikely and would not produce correlations between mothers and sons (and sisters and brothers) as they do not share a Y chromosome.

Sex-specific heritability estimates may also differ due to presence of two possible alleles at a given X-linked locus in females but only one in males, or by variations in gene expression depending on the sexual background on which they occur ('sex-limited' inheritance) (Lynch and Walsh 1999). The observed heritability patterns are unlikely to reflect gametic imprinting, in which genes are expressed differently depending on parent of origin (Solter 1988), as this should produce consistently stronger correlations between fathers (or mothers) with offspring of both sexes than between mothers (or fathers) and offspring. The possibility of gametic imprinting on alleles controlling IgE levels has been debated in relationship to a locus on chromosome 11q, with some data suggesting transmission limited to the maternal line (Cookson et al. 1992; Deichmann et al. 1999; Shirakawa et al. 1994), and others not (Dold et al. 1992; Doull 1996).

Other explanations might include gene-gene interaction of a trait locus with a sex-linked locus expressed only in males, or gene-environment interaction of an environmental factor that affects males more than females, but again the observed correlations, of males with their mothers and sisters, is difficult to explain. Adjustment for fixed effects of sex should strengthen parent-offspring and sibling correlations if true sex differences were present, and this was suggested for Blo t M IgE levels but not for other traits.  Adjustment for age and environmental covariates did not eliminate correlations among first-degree relatives, suggesting a genetic component of familial aggregation independent of age and measured environmental factors. Although formal testing of interaction terms showed only Blo t M IgE heritability to differ significantly by sex, power was limited for these formal comparisons and sex-specific differences were clear and consistent.

As the estimate of heritability is the slope of the regression of offspring on mid-parental values, the slope could be biased upwards if the ascertainment scheme preferentially included offspring with both higher and lower trait values, assuming mid-parental values were unbiased (as in extremely discordant sib pairs). The ascertainment of single probands with asthma and a history of one or more asthmatic siblings could upwardly bias the estimate of heritability, but the extent of the bias will depend largely on the proportion of individuals in the sampling frame relative to those not in the sampling frame. As the proportion of individuals in the sampling frame increases, the bias of the estimate will increase relative to random sampling, although the extent of this bias has not been determined and regression techniques tend to be fairly robust. We can anticipate no reason, however, that these families would be more likely to show the sex-specific differences found here compared with randomly ascertained families.

There appears to be only one prior report of heritability of IgE levels that examined sex-specific differences (Kacprazak-Bergman 1994). This small study of twins showed higher heritability of IgA, IgG, and IgM in male than female twins, though heritability of IgE levels was not reported. Sex differences in heritability of other traits have been reported from twin studies, including height (Silventoinen et al. 2000), low HDL-cholesterol and diastolic blood pressure (Poulsen et al. 2001), systolic blood pressure (McIlhany et al. 1975; Rice et al. 1989), and adiposity (Clarke et al. 1980). Sex differences in heritability observed in twin studies have been attributed to known sex differences in mean levels of these traits, such as blood pressure or lipids (Perusse et al. 1997; Poulsen et al. 2001); to effects of sex steroids (Kacprazak-Bergman 1994); or have remained unexplained (Boomsma et al. 1998; Clarke et al. 1980; Hong et al. 2001; McIlhany et al. 1975; Silventoinen et al. 2000).

The HERITAGE study has examined sex-specific differences in heritability using nuclear families, with heritability greater in males than females for testosterone and in females than males for dihydrotestosterone and lipoprotein lipase activity (Hong et al. 2001; Perusse et al. 1997). Here again, sex differences in heritability have been attributed to known sex differences in mean levels of these traits, but it is not clear how greater or lesser genetic contribution would necessarily lead to differences in mean levels among unrelated men and women. These reports and the twin studies cited above differ from the current report, however, in that correlations tended to be consistently highest among pairs of the same sex; for example, mother-daughter and sister-sister correlations for LPL activity higher than father-son and brother-brother correlations, with opposite sex correlations intermediate (Perusse et al. 1997), rather than the pattern of significant correlations with sons (or daughters) only as seen in the current study.

This pattern of higher heritability in males could be seen in a trait controlled by genes expressed only in males, as might be expected with sex-limited quantitative traits such as prostate size, or by a reduced environmental variance in males. Based on these findings, we propose that at least some portion of the underlying genetic influence on IgE levels is expressed only in males, possibly due to the same factors that cause greater susceptibility to allergic diseases and higher IgE levels in males (Freidhoff et al. 1984; Jarvis et al. 1995; Omenaas et al. 1994; Oryszczyn et al. 2000). If this is true, genetic determinants of IgE levels should be explored using approaches that allow for potential sex differences in expression.

Acknowledgements

We wish to thank the families in Barbados for their generous participation in this study; the physicians and staff of the Black Rock, Sir Winston Scott, Randal Phillips, Maurice Byer, Six Cross Roads, and St. John polyclinics and the Accident and Emergency Department, Queen Elizabeth Hospital; the staff of the Leptospira Laboratory, Barbados, for their technical support; and the Ministry of Health, Barbados, for permission to conduct this study. We are especially grateful to Maria Stockton for assistance with the fieldwork in Barbados, and for technical support with the Barbados data by Rasika A. Mathias, Daniel R. Breazeale, Xielun Xue, Liu Xin, John Neely, Dana Mulkern and Suzen Moeller. Dr. Barnes received support in this work from NIH Grant 1 R01 AI20059 and the Asthma and Allergy Foundation of America. Some of the results of this paper were obtained using the program package S.A.G.E., which is supported by a US Public Health Service Resource Grant (1 P41 RR03655) from the National Center for Research Resources.

Copyright information

© Springer-Verlag 2003