Glutathione-S-transferase P1, early exposure to mould in relation to respiratory and allergic health outcomes in children from six birth cohorts. A meta-analysis


  • Edited by: Stephan Weidinger


Christina G. Tischer, Helmholtz Centre Munich, German Research Centre for Environmental Health, Institute of Epidemiology I, Neuherberg, Germany.

Tel.: +49-89-3187-4498

Fax: +49-89-3187-3380




There are conflicting study results regarding the association of exposure to visible mould and fungal components in house dust with respiratory and allergic diseases in children. It has been suggested that functional polymorphisms of the GSTP1 gene may influence the risk for allergic disorders through an impaired defence against oxidant injury.


We examined in six birth cohorts of over 14 000 children whether the association between early exposure to reported mould at home in relation to respiratory and allergic diseases is modified by a single nucleotide polymorphism of the GSTP1 gene.


We observed a positive association of mould exposure with nasal symptoms (2–10 year) aOR: 1.19 (1.02–11.38). Further, there was a borderline significant increased risk of rhinoconjunctivitis (6–8 year) in children homozygous for the minor allele Val/Val, aOR: 1.25 (0.98–1.60). In stratified analyses, subjects homozygous for the minor allele and exposed to mould at home were at increased risk for early wheezing aOR: 1.34 (1.03–1.75), whereas the major allele may confer susceptibility for later nasal outcomes, (6–8 year) aOR: 1.20 (1.00–1.45) and (2–10 year) aOR: 1.30 (1.04–1.61), respectively. For none of the health outcomes studied, we found gene by environment interactions.


A genetic influence of the GSTP1 gene cannot be ruled out, but the magnitude of the effect is a matter of further research. In conclusion, the interplay between gene and environments is complex and remains subject of further study.

Exposure to mould indoors is considered a risk factor for childhood respiratory and allergic disorders [1, 2]. On the other hand, higher levels of fungal and bacterial components in house dust or a diverse composition of the microbial milieu may have protective effects [3-5]. However, a protective effect of a rural lifestyle could not be confirmed in all studies [6, 7]. Moreover, a large worldwide study in children (ISAAC) recently reported that exposure to farm animals was associated with an increased risk of asthma, rhinoconjunctivitis and eczema in children living in nonaffluent countries [8]. The reason for the discrepancies between the effects of various kinds of mould exposure on allergic health outcomes remains largely unknown. Gene–environment interactions may modify the relationship between exposure to indoor fungal load and (allergic) respiratory health outcomes. Glutathione-S-transferase (GST) genes, which are involved in defences against oxidant injury, have been suggested to modulate the effects of oxidant injuries through environmental exposure on asthma and allergy [9, 10]. The GSTP1 gene encodes the GSTP1 enzyme that is strongly expressed in the airways, and a functional polymorphism may confound the association between environmental exposure and clinical disease manifestation [11, 12]. In the context of air pollution from traffic, it has been observed that carriers of the minor allele of the GSTP1 gene were at higher risk to develop allergic sensitization [9], persistent wheeze [13] and incident asthma [14]. With regard to air pollution indoors, Lee et al. [15] reported that Taiwanese school children exposed to environmental tobacco smoke (ETS) at home and carrying any GSTP1 variant were at higher risk for ever and current wheezing.

Mould consists of a complex, partly unknown mixture of allergens, irritants and components that are able to induce inflammatory reactions of the airway epithelia [16, 17], and exposure to mould may be mediated through oxidant pathways. To our knowledge, there is only one comparable small study in 570 children from the Cincinnati Childhood Allergy and Air Pollution Study cohort (CCAAPS) which observed in stratified analyses that children carrying at least one variant of the GSTP1 gene were at increased risk for wheezing up to 2 years of age with early exposure to mould [13].

The current comprehensive meta-analysis was designed to answer whether early exposure to mould modifies respiratory and allergic health outcomes in children, especially among those carrying any variant of the GSTP1 gene (rs1695). This research question was studied in a population of more than 14 000 children from six birth cohorts at different time points between birth and 10 years of age.

Materials and methods

For this investigation, six birth cohort studies with relevant information on mould and/or dampness exposure, respiratory and allergic outcomes and GSTP1 (rs1695) genotype were included. The cohorts recruited subjects between 1991 and 1997 with a sample size between 351 (CAPPS) and 8639 children (ALSPAC). All birth cohorts have been described in detail elsewhere [LISAplus [18], GINIplus [19], BAMSE [20], PIAMA [21], CAPPS [22] and ALSPAC [23]] and obtained ethical approval from their local review boards. For twins, only one child per pair was included in the present analysis. Exposure was defined as parent-reported mould and/or dampness in any room of the home during the first 2 years of life. We focused on seven respiratory and allergic health outcomes based on the comparability across the birth cohort studies: early wheezing (0–2 years), defined as ‘wheezing or whistling in the chest’, early asthma symptom complex (0–3 year) and school-age asthma symptom complex (6–8 year). The asthma definition was based on the ISAAC-related questions [24] defined by the presence of two of the following three conditions: physician-diagnosed asthma ever, parental-reported wheezing (last 12 months) and asthma medication (last 12 months). We further defined two different symptoms for nasal disorders, ‘nasal symptoms’ (6–8 and 2–10 year) and ‘rhinoconjunctivitis’ (6–8 year). In the five European cohorts, ‘nasal symptoms’ were defined identically as ‘sneezing attacks, runny, blocked and itchy nose without having a cold’ according to the ISAAC definition [24]. Within the Canadian CAPPS cohort, there was no identical questionnaire information available, and therefore, ‘nasal symptoms’ were defined as ‘physician-diagnosed rhinitis’. ‘Rhinoconjunctivitis’ at early school age (6–8 year) was defined as having ‘runny, itchy eyes’ in addition to ‘nasal symptoms’. Sensitization to aero-allergens (6–8 years) was defined as specific Immunoglobulin E (IgE) of more than 0.35 kU/l to at least one of the measured aero-allergens (pollen, mites, cat, dog, mould). For ALSPAC and CAPPS, allergic sensitization was defined by skin prick test (SPT) with a positive reaction of having a weal diameter of ≥2 mm to at least one of the measured aero-allergens. Genotyping of the GSTP1 polymorphism rs1695 was performed within the individual cohorts and is published elsewhere in detail [9, 25-28].

Logistic regression analyses were applied to calculate adjusted odds ratios (aOR) for the association of allergic disorders with the GSTP1 polymorphism and early exposure to visible mould and/or dampness within the individual cohorts. For the meta-analyses, main effect models of early exposure to mould and GSTP1 genotype on health were performed. Second, the effect of early mould exposure was assessed by genotype of the GSTP1 in stratified analyses. An interaction term between early mould exposure and GST genotype (reference: Ile/Ile wild type) was included in the regression. All effect estimates were adjusted for sex, parental allergy and smoking during pregnancy. For the CAPPS birth cohort, there was no information on smoking during pregnancy available and the cohort-specific effect estimates are therefore based on sex and information on parental allergy. To account for the heterogeneity in effect estimates between the cohorts, meta-analyses with random-effects models were applied. Statistical analyses were performed using the statistical software R, version 2.14.1 (


Our study population comprised 14 595 individuals from six birth cohorts from Europe and Canada with information on the GSTP1 (rs1695) genotype.

During the first 2 years of life, exposure to residential mould and/or dampness ranged from 27% in Germany (GINIplus) to 68% in Bristol, UK (ALSPAC), as shown in Table 1. The genotype frequencies of the GSTP1 polymorphism were similar across the birth cohorts and ranged from 40% to 48% for the Ile/Val and from 11% to 12% for the Val/Val genotype.

Table 1. Overview of the study populations from six birth cohorts
Study subjects (N)N = 8639N = 956N = 1778N = 351N = 1875N = 996
  1. SPT: ALSPAC (8 year) – grass, mite, cat; CAPPS (7 year) – Alternaria, Cladosporium, Penicillium, tree, grass, weed, ragweed, cat, dog, feather, der p 1, der f 1, roach.

  2. IgE: BAMSE (8 year) – mite, cat, dog, horse, timothy, birch, mugwort, mould; PIAMA (8 year) – Der p 1, cat, dog, birch, cocksfoot, Alternaria alternata; GINIplus (6 year) – Der p1, cat, dog, rye, birch, mugwort, timothy, Cladosporium herbarum; LISAplus (6 year) – inhalation mixture sx1.

Exposure (%)
Visible mould/dampness (0–2 year)4876/7167 (68)235/817 (29)478/1756 (27)218/351 (62)772/1875 (41)380/977 (39)
Health end points (%)
Early wheezing (0–2 year)2401/7250 (33)420/956 (44)348/1673 (21)122/347 (35)636/1849 (34)353/960 (37)
Early asthma symptom complex (0–3 year)n.a.272/954 (29) 2 year33/1691 (2) 3 year36/279 (13) 2 year94/1584 (6) 3 year111/981 (11) 2 year
Asthma at school age (6–8 year)710/5795 (12) 7 year174/896 (19) 8 year69/1769 (4) 6 year56/319 (18) 6 year97/1646 (6) 6 year26/991 (3) 6 year
Ever nasal symptoms (2–10 year)1488/4999 (30)238/897 (27)702/1483 (47)148/349 (42)942/1759 (54)340/771 (44)
Nasal symptoms at school age (6–8 year)776/6131 (13) 7 year116/897 (19) 8 year356/1771 (20) 6 year129/351 (37) 7 year386/1810 (21) 6 year157/989 (16) 6 year
Rhinoconjunctivitis at school age (6–8 year)311/5738 (5) 7 year94/893 (11) 8 year169/1770 (10) 6 year60/130 (46) 7 year96/1807 (5) 6 year69/989 (7) 6 year
Early school-age sensitization to aero-allergens (6–8 year) IgE/SPT

SPT 8 year

1080/5428 (20)

IgE 8 year

220/752 (29)

IgE 6 year

538/1773 (30)

SPT 7 year

154/339 (45)

IgE 8 year

457/1457 (31)

IgE 6 year

256/993 (26)

Risk factors (%)
GSTP1 (rs1695)
Ile/Ile3671/8639 (42)434/956 (45)745/1778 (42)170/351 (48)745/1875 (40)432/996 (43)
Ile/Val3908/8639 (45)414/956 (43)822/1778 (46)141/351 (40)897/1875 (48)451/996 (45)
Val/Val1060/8639 (12)108/956 (11)211/1778 (12)40/351 (11)224/1875 (12)113/996 (11)
Sex (female) (%)4178/8639 (48)450/956 (47)869/1778 (49)160/351 (46)915/1875 (49)456/996 (46)
Maternal smoking during pregnancy (%)2048/7955 (26)141/956 (15)239/1761 (14)n.a.198/1857 (11)132/962 (14)
Parental allergy (%)4813/6585 (73)317/946 (34)1137/1728 (66)345/351 (98)1048/1875 (56)641/953 (67)
Study regionUKSwedenGermany (multicenter)Canadathe NetherlandsGermany (multicenter)
Recruitment year (starting)19911994199519951996/971997

The mutually adjusted main effects for early exposure to mould and GSTP1 genotype on health are presented in Table 2a. Mould exposure within the first 2 years of life was significantly associated with an increased risk for early wheezing (0–2 year) aOR: 1.27 (1.09–1.47); however, there was heterogeneity between the cohorts indicated (P = 0.048). No association was observed in relation to the asthma symptom complex at any age group. Exposure to mould conferred a significant risk increase for childhood nasal symptoms (2–10 year) aOR: 1.19 (1.02–1.38), however, with concurrent heterogeneity between the cohorts (P = 0.048). Similarly, a nonsignificant association with mould exposure was seen for nasal symptoms at 6–8 years [aOR: 1.10 (0.98–1.24)], but not for rhinoconjunctivitis and sensitization to aero-allergens at early school age. For the main effect models, no significant genotype effect was identified except for a borderline significant risk increase of rhinoconjunctivitis at early school age (6–8 year) in the group of children, homozygous for the Val/Val genotype [aOR: 1.25 (0.98–1.60)].

Table 2. Adjusted odds ratios (aORs) and 95% confidence intervals (95% CI) for the association between symptoms of asthma, nasal symptoms and sensitization to aero-allergens with early exposure to mould and/or dampness (0–2 years) and GSTP1 (reference group Ile/Ile) (a) model with main effects (b) stratified by genotype and (c) interaction model
 Summary effect N Test for heterogeneity (P) P < 0.05
Early wheezing (0–2 year)
Mould 1.27 (1.09–1.47) ** 10 915 0.048
Ile/Val1.07 (0.91–1.25)0.440
Val/Val0.93 (0.81–1.06) 0.033
Early asthma symptom complex (0–2/3 year)
Mould1.24 (0.86–1.80)5174 0.024
Ile/Val1.04 (0.77–1.41)0.130
Val/Val0.85 (0.61–1.19)0.548
School-age asthma symptom complex (6/8 year)
Mould1.04 (0.90–1.21)97520.853
Ile/Val1.14 (0.96–1.36)0.338
Val/Val1.18 (0.86–1.62)0.218
School-age nasal symptoms (6/8 year)
Mould1.10 (0.98–1.24)10 2220.387
Ile/Val1.03 (0.92–1.15)0.440
Val/Val1.09 (0.81–1.45) 0.033
Childhood nasal symptoms ever (2–10 year)
Mould 1.19 (1.02–1.38) ** 9038 0.048
Ile/Val1.07 (0.98–1.18)0.396
Val/Val1.07 (0.92–1.23)0.444
School-age rhinoconjunctivitis (6/8 year)
Mould1.07 (0.81–1.40)9676 0.032
Ile/Val1.06 (0.86–1.30)0.235
Val/Val1.25 (0.98–1.60)0.410
School-age sensitization to aero-allergens (6/8 year)
Mould1.02 (0.92–1.13)90460.537
Ile/Val0.99 (0.87–1.12)0.229
Val/Val0.98 (0.80–1.19)0.221
Early wheezing (0–2 year)
Ile/Ile1.28 (0.99–1.65)10 915 0.019
Ile/Val 1.24 (1.09–1.41) ** 0.650
Val/Val 1.34 (1.03–1.75) ** 0.970
Early asthma symptom complex (0–2/3 year)
Ile/Ile1.24 (0.81–1.90)51740.168
Ile/Val1.27 (0.86–1.87)0.219
Val/Val1.19 (0.53–2.67)0.305
School-age asthma symptom complex (6/8 year)
Ile/Ile1.03 (0.81–1.30)97520.938
Ile/Val1.08 (0.89–1.30)0.745
Val/Val0.85 (0.56–1.30)0.660
School-age nasal symptoms (6/8 year)
Ile/Ile 1.20 (1.00–1.45) * 10 2220.347
Ile/Val1.04 (0.84–1.29)0.183
Val/Val1.03 (0.74–1.44)0.549
Childhood nasal symptoms ever (2–10 year)
Ile/Ile 1.30 (1.04–1.61) ** 90380.089
Ile/Val1.13 (0.85–1.51) 0.013
Val/Val0.94 (0.71–1.24)0.681
School-age rhinoconjunctivitis (6/8 year)
Ile/Ile1.22 (0.88–1.71)96760.169
Ile/Val0.94 (0.63–1.38)0.052
Val/Val0.79 (0.49–1.27)0.868
School-age sensitization to aero-allergens (6/8 year)
Ile/Ile1.03 (0.83–1.28)90460.130
Ile/Val1.02 (0.85–1.23)0.265
Val/Val0.99 (0.73–1.34)0.891
  N Mould P Ile/Val P Val/Val P Mould* Ile/Val P Mould* Val/Val P
  1. *P < 0.1, **P < 0.05, ***P < 0.001.

Early wheezing (0–2 year)10 9151.28 (0.99–1.65) 0.019 1.10 (0.91–1.31)0.1080.89 (0.73–1.09)0.9350.99 (0.80–1.22)0.3351.06 (0.79–1.43)0.865
Early asthma symptom complex (0–2/3 year)51741.23 (0.81–1.87)0.1681.07 (0.81–1.40)0.6420.87 (0.55–1.36)0.9480.96 (0.65–1.42)0.6800.92 (0.39–2.21)0.307
School-age asthma symptom complex (6/8 year)97521.03 (0.82–1.30)0.9311.07 (0.85–1.34)0.6221.25 (0.90–1.75)0.4671.08 (0.79–1.47)0.8880.81 (0.50–1.31)0.569
School-age nasal symptoms (6/8 year)10 222 1.21 0.05(1.00–1.47) ** 0.3301.13 (0.96–1.33)0.5381.19 (0.77–1.84) 0.016 0.85 (0.65–1.09)0.3450.85 (0.56–1.30)0.310
Childhood nasal symptoms ever (2–10 year)90381.30 (1.051.62)**0.0881.12 (0.96–1.31)0.2711.24 0.04(1.011.52)**0.6910.90 (0.69–1.18)0.1750.68 (0.43–1.08)0.136
School-age rhinoconjunctivitis (6/8 year)96761.21 (0.87–1.69)0.1831.21 (0.91–1.62)0.2161.47 (1.022.13)**0.3270.75 (0.50–1.14)0.2670.85 (0.51–1.43)0.454
School-age sensitization to aero-allergens (6/8 year)90461.03 (0.83–1.27)0.1400.98 (0.85–1.14)0.5711.01 (0.81–1.25)0.4690.96 (0.68–1.36)0.0590.98 (0.69–1.37)0.976

We further investigated the results from the adjusted logistic regression models, stratified by genotype (Table 2b). With regard to wheezing up to 2 years, early exposure to mould was observed to be a risk factor regardless of the genotype, however, borderline significant for the major allele: GSTP1 Ile/Ile: aOR: 1.28 (0.99–1.65), Ile/Val: aOR: 1.24 (1.09–1.41) and Val/Val: aOR: 1.34 (1.03–1.75). We did not observe associations between exposure to mould, GSTP1 genotype and asthma symptom complex at early life (0–3 year) and early school age (6–8 year). For school-age (6–8 year) and childhood (2–10 year) nasal symptoms, visible mould exposure significantly increased the risk for children homozygous for the major allele: aOR: 1.20 (1.00–1.45) and aOR: 1.30 (1.04–1.61), respectively. There was a nonsignificantly decreased risk of school-age rhinoconjunctivitis (6–8 year) in children exposed to mould early in life carrying at least one Val variant of the GSTP1 genotype. For school-age sensitization to aero-allergens (6–8 year), no significant association was observed.

In addition, we tested the statistical significance of the gene–environment interaction effects of early exposure to mould and/or dampness at home in models with interaction terms. For none of the outcomes studied, the interaction was statistically significant (Table 2c). The results of the meta-analyses presented as forest plots are provided online (Fig. S1). Moreover, to increase statistical power, all carriers of the minor allele were grouped together, but the results did not change with the recessive coding (Table S3).


We observed a significant effect of mould exposure on wheezing in the first 2 years of life, and later in childhood, exposure to mould was also significantly associated with nasal symptoms between 2 and 10 years. There was evidence of a borderline main genetic effect of GSTP1 Val/Val on early school-age rhinoconjunctivitis (6–8 year). The stratified analyses revealed that in children carrying at least one Val allele, exposure to mould early in life significantly increased the risk for early wheezing. In contrast, the major allele may confer susceptibility for later nasal outcomes in childhood. No significant statistical gene by environment interaction effects were found for the outcomes assessed.

To our knowledge, this is the first collaborative investigation that looked at gene–environment associations with respect to mould exposure and different respiratory and allergic health outcomes using individual participant data from birth cohorts. Little is known about how or if mould or sources of mould are processed through oxidant pathways. Jussila et al. [29] observed that exposure to specific mould spores is related to an augmentation of intracellular levels of reactive oxygen species (ROS) in mice. In a small study in humans, it was reported that being exposed to mould at home after birth was a risk factor for persistent wheezing up to 2 years for infants, carrying any variant of the GSTP1 genotype [13], which could be confirmed by the data of our comprehensive meta-analyses. However, the clear distinction of a specific wheezing phenotype is challenging, especially at early ages [30, 31], and we found no indication that mould or a specific genotype of the GSTP1 was related to asthma outcomes later in childhood.

As observed in our previous meta-analysis [2], mould exposure in the first 2 years of life seems to be a risk factor for later nasal symptoms in childhood. In the current investigation, we found out that this concerns especially those children homozygous for the major allele, and a similar, nonsignificant tendency has been also observed for rhinoconjunctivitis at early school age. No comparable studies of mould exposure and GSTP1 are known at present. One study in adults looked at indoor air pollution from tobacco smoke, GSTP1 variants and allergic rhinitis; however, no association was reported [10]. Gilliland et al. [32] observed in a randomized, placebo-controlled crossover study of 19 adult subjects that patients homozygous for the GSTP1 Ile105 wild type were reported to show enhanced increase in allergen-specific IgE and histamine release when exposed to diesel exhaust particles. Gerbase et al. [10] reported that being homozygous for the minor allele of the GSTP1 gene was protective for high total and specific IgE in subjects persistently exposed to second-hand smoke. In contrast, a study in 982 4-year-old children from the BAMSE birth cohort found that children with the Ile/Val or Val/Val genotypes were at higher risk to be sensitized to common allergens, in the presence of high levels of NOx during the first year of life [9].

With regard to mould exposure, it is generally difficult to confine ‘mould’ to specific agents, and the impact on health is further dependent on personal behaviour and synergistic or repressive effects of the overall microbial pollution indoors [16]. Therefore, it is unclear whether the tested aero-allergens in relation to allergic sensitization, especially for mould, are the causal agents. Further, the assessment of sensitization to aero-allergens was not identical in the participating cohorts, and two of six cohorts performed skin prick tests compared to IgE levels in blood. The technical differences in the test procedures of the two assessment methods or different test substances used might have an unknown impact on the results which cannot be determined.

This study has important strengths such as the longitudinal study design, the considerable sample size and a broadly similar assessment of phenotypes. However, a few limitations should be noted. The prevalence for the homozygous Val/Val genotype was modest, with approximately 12% in each birth cohort which restricted analysis in terms of statistical power. Therefore, one cannot rule out completely the possibility of a gene–environment interaction between early exposure to mould at home in relation to the studied health outcomes. However, when grouping all carriers of the minor allele in analyses with dominant coding, the effect estimates did not change.

A further limitation could be considering only one gene within a highly complex system of detoxifying enzymes while many other genes might be mutually involved. Apart from that, further candidates might be chitinases, which are enzymes involved in cleaving chitin, a polysaccharide that is present in fungal cells [33]. Wu et al. [34] observed in subjects of the Childhood Asthma Management Program (CAMP) that high levels of mould indoors may modify the association between specific SNPs in CHIT1 gene and asthma exacerbations.

The heterogeneity between the cohorts was apparent for some of the models. The Canadian CAPPS birth cohort is designed as a high-risk birth cohort, and we therefore additionally performed a sensitivity analysis, excluding CAPPS. Although the heterogeneity decreased, the results did not change (data not shown). Further, there might be differences in gene–environment interactions at different levels of exposure to mould as it was observed for bacterial endotoxin and the CD14 promoter: it was reported that the same polymorphism was protective at some levels and the risk allele at others [35].

The current meta-analysis is one of the first investigating the effect of environmental exposure from biological origin in relation to oxidative stress and respiratory and allergic diseases. We cannot exclude residual confounding by unmeasured or inadequately measured risk factors given the various range of microbial and unknown xenogeneic exposure encountered at home. A genetic influence cannot be ruled out, but the magnitude of the effect and the responsible genes are a matter of further research. The impact of different levels of exposure to mould on health and the possibility of crossover interaction with different effects at different alleles should be considered in future investigations. In conclusion, the interplay between genes regulating the defence against oxidative stress in interaction with environmental exposure is complex and remains a subject of further study.

Authors' contribution

Christina G. Tischer, first author, involved in statistical analysis, manuscript preparation and manuscript revision. Joachim Heinrich involved in study design and provision of data, performed data analysis and made comments to the draft. Anna Gref, Mario Bauer, Anna Bergström, Mike Brauer, Christopher Carlsten, Ulrike Gehring, Raquel Granell, John Henderson, Marjan Kerkhof, Meaghan MacNutt, Erik Melén, Marie Standl, Magnus Wickmann involved in provision of data, made comments to the draft and critically revised the manuscript.

Funding support

This project was co-funded by AllerGen NCE, in support of the TAG project, HITEA project (Health Effects of Indoor Pollutants), Helmholtz Center Munich and UFZ Leipzig.

Conflict of interest

The authors declare no conflict of interest regarding this manuscript.