Introduction

The global prevalence of allergic diseases has increased steadily for more than 50 years [1], initially due to increasing prevalence in the developed countries and more recently in developing countries [2]. The etiology of these conditions is believed to be complex [3, 4]. Exposure to indoor air pollution has been hypothesized to be particularly relevant to allergic disease pathogenesis as people spend much of their time indoors [5]. Recognized indoor air pollutants include tobacco smoke [6, 7], dust [8, 9], and pollutants generated from cooking and heating [10]. However, studies have failed to find consistent associations between indoor air pollutants and allergic disease [11, 12]. The reason for these inconsistent findings may be partly related to failure to account for potential gene-environment interactions.

One key group of genes proposed to modify the impact of air pollutants on asthma and allergies are the Glutathione S-transferases (GSTs) genes [13, 14]. The GSTs are major phase II detoxification enzymes found mainly in the cytosol and include the Mu (GSTM1), Theta (GSTT1), and Pi (GSTP1) classes [15]. These enzymes provide the first line of defense against oxidative stress through their ability to protect cells from reactive oxygen species (ROS) [16•, 17]. ROS are active participants in complex biological processes, including allergic sensitization and allergic asthma [18]. ROS can cause airway inflammation, which may result in airway dysfunction [19, 20]. The GSTM1 and GSTT1 genes have null alleles resulting in partial or complete loss of the enzyme activity [16•, 17, 21,22,23]. Despite the biological plausibility of a relationship between GST polymorphisms and allergies and poorer lung function, the supporting evidence is scarce and inconsistent. A 2010 systematic review investigated the role of GST genes in asthma, finding no direct association [24•]. The authors postulated that the null findings may be due to not including exposure data. Failure to take gene-environment interactions into account in this review may explain the absence of consistently reported observable effects of the GST genes. There have been no previous systematic syntheses of the literature focusing on the interaction between indoor environmental exposures and GST genes on outcomes of allergic disease and lung function.

Establishment of this potential gene-environment interaction may help to explain the inconsistencies found where these factors are investigated in isolation, and help identify an important subpopulation requiring intervention to prevent respiratory morbidity. We aimed to address this through a systematic review and meta-analysis of the existing literature investigating the influence of GST genes on the relationship between indoor air pollution and allergy/lung function.

Methods

We performed a systematic review of published studies that reported original data, according to PRISMA guidelines. The review protocol was registered with PROSPERO (CRD42017062275).

Literature Search and Study Selection

We searched PubMed and Embase databases from inception for journal articles written in English. The final search was performed on 25/03/2017. We used search terms in PubMed to identify articles having the following three categories in common: indoor air pollution (MeSH heading “smoke,” “air pollution, indoor”), and, allergy (MeSH heading “asthma,” “rhinitis,” “allergic,” “eczema,” “dermatitis;” “food allergy”) or lung function (MeSH heading “respiratory function test”), and GST (MeSH heading “glutathione s-transferase”). A similar search strategy was developed for Embase (see online supplement 1 and 2). The search was performed by two independent investigators (XD and GB).

Eligibility Criteria

To be included, a study had to meet the following criteria: (1) human population, (2) indoor air pollution exposure (smoke, cleaning products, heating, cooking, air quality and airborne exposure), (3) investigated interaction by GST genes, (4) assessed any allergic disease (asthma, rhinitis, eczema, dermatitis and food allergy) or lung function outcome, (5) cohort study, cross-sectional, or case-control study design. Non-English language papers, unpublished studies, and studies not reporting original data (reviews, editorials, and conference abstracts) were excluded. Two authors (XD and GB) independently evaluated all articles retrieved.

Data Extraction

Two researchers (XD and GB) independently reviewed titles and abstracts, and where necessary full texts, to determine if they met the inclusion criteria as described above. Any disagreements were settled by a third author (CL). For each included study, information regarding study design, target population, time of exposure measurement, outcome measures and age, interactions investigated, covariates included, comparison, and association/findings were extracted into a standard table and checked by two authors (XD and GB). Exposure and outcome definitions and effect estimates (odd ratios (OR), relative risks, prevalence ratios, lung function parameters) with 95% confidence interval (95% CI) were extracted where available for inclusion in meta-analyses.

Quality Assessment

Two authors (XD and GB) independently assessed the quality and bias of included studies using the Newcastle-Ottawa quality assessment scale (NOS). Each study was scored using a star (*) method to report the quality based on selection of sample, comparability and the ascertainment of the exposure or outcome measures. All papers were classified into high (8–9*), medium (6–7*), low (4–5*), and unsatisfactory quality (0–3*) based on the Newcastle-Ottawa quality score (supplemental table 1). We resolved any disagreement through consensus with a third author (CL). We aimed to perform forest plots to address potential publication bias if applicable.

Results

Search Results

From a total of 576 citations identified initially, 71 were selected for full-text review and 22 met the inclusion criteria. Of these, 16 papers evaluated allergy-related outcomes, and 8 measured lung function outcomes (2 papers measured both asthma and lung function (Fig. 1). Supplemental table 2 outlines reasons for exclusion at the full-text review stage. We were unable to perform meta-analyses due to high heterogeneity, with specific gene-pollutant-outcome combinations lacking numbers.

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Characteristics of the Included Studies

Of the 16 studies with allergic disease outcomes (Table 1), 14 investigated outcomes of asthma/wheeze [5, 25,26,27,28,29,30,31,32,33,34•, 35••, 36, 37••], and the remainder investigated outcomes of rhinitis [35••, 38] and dermatitis [35••, 39]. The 8 studies [5, 26, 40,41,42,43,44,45] investigating lung function outcomes (Table 2) used a variety of measurements including maximal flow at functional residual capacity (VFRCmax), forced expiratory volume in 1 s (FEV1)%, forced vital capacity (FVC)%, peak expiratory flow rate (PEFR)%, FEV1, FVC, FEV1/FVC, FEV1/height2, 5-year change in FEV1%, mean annual change in FEV1, FVC, and high/low lung function categories. The ages at which outcomes were measured varied from 1 month [42] to 69 years [5]. No studies included an elderly population. Sample sizes ranged from 140 to 3738. The numbers of cohort, cross-sectional, and case-control studies were 9, 9, and 4, respectively.

Table 1 Characteristics and methods of the 16 studies with allergic disease outcomes
Full size table
Table 2 Characteristics and methods of the 8 studies with lung function outcomes
Full size table

Studies were conducted at a variety of locations across Europe, North American, Mexico, Australia, Taiwan, and South Korea. All papers reported the geographic location of their studies, 8 further provided information on ethnicity. Those 8 studies were mainly conducted on Non-Hispanic white/white/Caucasian.

Measurement of Exposures and Outcomes

Genetic Polymorphisms

Genotyping methods were reported for all studies. Six of the 22 papers described confirmation of genotyping results by analyzing a duplicated random sample (5%–15%) and all were concordant with the initial results [27, 31, 33, 39, 42, 45].

Environmental Exposures

Nineteen of the 22 studies evaluated indoor air pollutants by self-report questionnaires. The remaining 3 studies used more objective measures to quantitate exposure, including blood cotinine level [39], urinary cotinine level [30], and visible mold assessed by a home visit [33]. No studies measured indoor air pollution directly through air sampling (e.g., air quality monitors).

Outcome Assessment

Asthma and allergic outcomes were evaluated by self-report questionnaires for most studies, apart from 1 that extracted asthma diagnosis from hospital records [31], and another that undertook independent clinical examination by doctors [26]. All lung function studies performed spirometry, except 1 study in infants that measured VmaxFRC using the rapid thoracoabdominal compression technique during tidal breathing [42].

Assessment of Quality of Included Studies

Based on NOS assessments, 6 studies were low, 11 studies medium, and 5 studies high quality. The main issue identified was failure to adjust for a minimum set of confounders, including age (not applicable for birth cohort studies), gender, socio-economic status (SES), and family history of allergic disease (Fig. 2). Only 5 studies fully adjusted for these variables [27, 28, 36, 38, 39], while 2 did not describe any confounders [30, 43].

Fig. 2
figure 2

Casual diagram for GST gene interaction on the association between indoor air pollution and allergic disease and lung function

Full size image

Narrative Synthesis of All Identified Studies

Of the 22 included studies, 15 supported the presence of a gene-indoor air pollution interaction on the outcome of allergic disease, lung function, or both [25,26,27,28,29, 32, 34•, 35••, 36, 39,40,41,42,43, 45]. Interestingly, of those 7 papers failing to detect gene-air pollution interactions for allergies or lung function, 2 found interactions instead for outcomes of high total IgE [38]or atopy (positive skin prick test), which are recognized markers of allergy [31, 38]. Seventy-nine percent (n = 11) of childhood papers found at least one interaction [25,26,27,28,29, 32, 34•, 35••, 36, 39, 42] compared with only half (n = 3) of the papers in adulthood [40, 41, 45]. One paper that spanned both age groups (young children to young adults aged 3–21 years) also found an interaction [43]. All five Taiwanese studies identified genetic susceptibility for environmental factors, while 75% of the European studies and only half of the North American studies found an interaction.

Asthma and Wheeze

Fifteen papers reported results on asthma or wheeze, and nine identified an interaction with one or more of the GST polymorphisms. Four out of eight papers investigating the potentially modifying effect of GSTM1 on the association between indoor air pollution and allergies found that carriers of the null genotype had an increased risk of asthma/wheeze when school-aged children were exposed to markers of increased indoor air pollution [25, 26, 28, 35••]. However, one Taiwanese study found weak conflicting evidence: decreased risk of asthma in children carrying the null GSTM1 genotype and exposed to ETS (p = 0.0573) [29]. Another oxidative stress gene Quinone oxidoreductase 1 (NQO1) may be part of a three-way interaction with children carrying the GSTM1 null and NQO1 ser genotype at increased risk of asthma in the exposed group, but not in the unexposed group [29]. Of the three papers reporting on GSTT1, two cross-sectional studies investigating asthma in school-aged children found evidence of an interaction [25, 36]. They found that environmental pollutants including in-utero smoking and incense burning smoke were associated with increased risk of asthma/wheeze if children had GSTT1 null genotypes. The other GSTT1 paper failed to replicate this interaction in adulthood [5]. Of eight papers investigating GSTP1, three found that the GSTP1 val105 allele was associated with an increased risk of wheezing/asthma with in-utero smoking exposure and/or non-exposed children carrying GSTP1 ile105 alleles were at increased risk [27, 32, 34•]. One paper further investigated GSTP1 by using a haplotype-based analysis [34•]. GSTP1 genes were categorized by four haplotype variation tagging SNPs. Consistent with a risk effect of the GSTP1 val 105 variant, the effect of in-utero exposure to maternal smoking on wheezing was largest in children within the h1011 haplotype (105val with no other variants) compared with other haplotypes [34•]. In contrast, there was one study suggesting that GSTP1 ile105 increased the risk of atopy (but not asthma severity) when exposed to ETS [31]. Although a cross-sectional study in Korea did not observe an interaction between ETS, GSTP1, and asthma, when they further stratified by vitamin A intake (low vs high), they found that school children with GSTP1 homozygous ili genotypes exposed to ETS and low vitamin A had an increased risk of asthma [37••]. The remaining three studies showed no evidence of a GSTP1 interaction for all investigated current environmental exposures [30, 31, 33].

Other Allergic Disease

There were two studies investigating rhinitis [35••, 38], and two investigating atopic dermatitis (AD) [35••, 39]. No study investigated the outcome of food allergy. No significant associations were found between allergic rhinitis and GST genotypes in different smoking categories in adulthood [35••, 38]. However, children with GSTM1 null genotypes were found to be more susceptible to AD if they had been exposed to in-utero smoking, while GSTP1 ile105 homogeneity increased the risk of AD in the in-utero unexposed group, but no association was seen in the exposed group [39].

Lung Function

Six of the eight studies that investigated the potential interaction between smoking and GST genes found positive results. All three studies investigating GSTM1 found that carriers of null variants exposed to smoking, had an increased risk of impaired lung function in terms of FEV1% predicted [41, 43], PEFR % predicted [43], FVC% predicted, and FEV1/FVC ratio [26]. These associations were not seen in the non-smoke-exposed group. Two studies formally tested and found statistical evidence for an interaction term [41, 43]. For GSTT1, one study investigated infant lung function, finding that among infants exposed to in-utero smoking, those with null variants had reduced VmaxFRC in the first year compared with non-null infants. No significant associations were seen in the group not exposed to in-utero smoke [42]. Another two adult studies identified significant interactions between GSTT1 genotype and pack-years on lung function in smokers [40, 45]. Two studies which support the presence of interaction by GSTP1 provided evidence that in those exposed to tobacco smoke, having the GSTP1 val105 allele increased the risk of reduced lung function. One found that GSTP1 val105 homozygote individuals demonstrated a reduction in PEFR% predicted in the older ETS exposed group (13–21 years vs 3–12 years) [43]. The second study found that in persistent smokers, those homozygous for GSTP1 val105 had a faster rate of lung function decline in FVC during 11 years’ follow-up compared to heterozygotes. However, irrespective of smoking status, no independent effect of GSTP1 was found [45].

Two adult studies did not identify any interaction. The Health 2006 study in Copenhagen found no evidence of GST interaction for a wide range of indoor exposures (active smoking, passive smoking, and use of woodstove or candles during wintertime) and the risk of FEV1% predicted (interaction p varies from 0.1–0.6) [5]. Another study nested in the Lung Health Study in North America found no significant GST interactions between smoking history and rapid decline in lung function over 5 years, either using single genes or a combination [44].

Discussion

This is the first systematic review to investigate the interaction between indoor air pollution and GST genes on allergic disease and lung function. Our narrative synthesis of 22 studies found evidence that GST polymorphisms interact with indoor air pollution exposures to adversely affect lung health. Carriers of GSTM1 null, GSTT1 null, and GSTP 105val alleles were at greater risk of developing asthma and reduced lung function when exposed to tobacco smoke. Three of the seven papers that did not find an interaction between indoor air pollution, GST, and allergic disease reported interactions on other allergy-related outcomes, in term of IgE [38] or atopy status [31], or after consideration of antioxidant nutrient intake [37••]. Two previous systematic reviews on oxidative stress genes, outdoor air pollution, and lung function outcomes, both suggested the presence of an interaction [16•, 46]. We reviewed GST genes and indoor air pollution, providing further and more current evidence for the presence of an interaction.

Biologic Plausibility

Our findings are supported by a strong biologic plausibility. Oxidative stress is a key component of inflammatory disorders, and host antioxidant systems are activated in response to an external “oxidant attack” caused by exposure to environmental pollution [47]. ETS has been shown to have detrimental effects on the lung, mostly related to increased oxidative stress [48], as well as suppression of the immune system through modulation of T cell function [49]. Several studies have demonstrated that environmental exposures can influence GST gene activity. A study in rats found that exposure to cigarette smoke five times/week led to an increase in GST activity in both the brain and lungs [50]. Similarly, a human study found that the degree of antioxidant gene expression is directly related to the amount of cigarette smoke exposure [51], suggesting upregulation in response to environmental stimuli. Individuals who lack the protection of certain antioxidant genes may have a lower capacity for antioxidant defense, and be more susceptible to environmental toxins. This may lead to increased risk of asthma or impaired lung function when faced with oxidative exposures. Fourteen studies supported that cigarette smoke-exposed carriers of the GSTM1/T1 null genotypes may have increased risk of allergic disease and impaired lung function.

Evidence for GSTP1 Risk Allele Is Inconsistent

We found that carrying the GSTP1 val105 allele increased the risk of asthma and lung function deficits in individuals exposed to indoor air pollution. One study on asthmatic children found that carriers of GSTP1 val/val genotypes had higher levels of oxidative stress (measured as plasma levels of malondiadehyde) and less antioxidant defense capacity (measured as glutathione levels), compared with other genotypes [47]. Therefore, if ile alleles are replaced by val alleles, the resultant-encoded enzyme can alter the level of antioxidative capacity in the individual and may further contribute to airway inflammation when exposed to oxidative threats [34•, 52]. Conversely, another study indicated that carriers of homozygous GSTP1 ile alleles may have increased risk of atopic dermatitis in those unexposed to in-utero smoking but not in the exposed group [39]. The reason for this contradictory outcome is unclear. There is evidence showing that enzymes with GSTP1 val105 have a greater catalytic effect for diol epoxides of polycyclic aromatic hydrocarbons but less effect with 1-chloro-2,4-dinitrobenzene compared to ile105 [53]. Therefore, we propose that the effect of GSTP1 gene polymorphisms on allergy risk and lung function is likely to vary for different pollutants or in different scenarios.

Reasons for Other Inconsistent Findings

Inconsistent findings from specific exposures may be caused by differing environmental conditions or study methodology. For instance, all 6 papers exploring in-utero smoking suggested that adverse effects were more likely to occur in individuals with a GST risk polymorphism. This suggests a potential critical time window, during which exposure may lead to long-term effects on fetal lung development [6]. Moreover, asthma and other allergic diseases are heterogeneous diseases with complex etiologies. A few studies in our review suggested the presence of complex interactions, including NQO1 [29], other GST genotypes [44, 45], bronchial hyperresponsiveness to methacholine chloride [45], antioxidative vitamins [37••], age [43, 45], and sex [35••, 45]. Asthma and reduced lung function may result from effects of multiple genes and their interactions with environmental and host factors. It is unlikely that a single gene polymorphism is the only mechanism responsible for management of oxidative stress within the lung. It is likely that multiple points of biological redundancy have been developed to protect the airways and prevent adverse respiratory outcomes. The influence of other factors may increase the magnitude of a proposed gene-environment interaction. The Cincinnati Childhood Allergy and Air Pollution Study suggested that GSTP1 genes could modify the negative effect of diesel exhaust particle (DEP) exposure. However, the protective effects of GSTP1 genotypes may be overridden when children are exposed to multiple environmental stresses in early life. In these infants, despite carrying genotypes considered protective for lesser levels and varieties of exposure, adverse outcomes were observed with multiple environmental loads [33]. Interactions were found more commonly in studies on children compared to adults, suggesting that the effect of gene-pollution interaction on respiratory health varies by age. Children’s lungs are still developing and therefore may be more susceptible to noxious exposures. Furthermore, immune responses are relatively naïve in children when compared to adults. For both reasons, antioxidant defenses may play a crucial role in overcoming the adverse effects of environmental exposures on childhood respiratory health [54, 55].

Evidence from the Past 5 years

In terms of the most recent evidence, since 2013, there have only been two original research articles published on GST genes and indoor air pollutants [35••, 37••]. One provided evidence that low vitamin A intake in genetically susceptible children (GSTP1 1695 AA) exposed to ETS was associated with an increased risk of asthma [37••]. The authors suggested that low vitamin A diets may increase the level of oxidative stress making genetically susceptible children more likely to be affected by environmental exposures like ETS. This finding suggests that supplemental vitamin A may be a preventive measure for those at high risk. The other recent study provides new evidence on gender differences and risk alleles for the interactive effect of GSTM1 with in-utero smoking on childhood asthma [35••]. This study demonstrated that the GSTM1 null genotype could have a bipolar effect on childhood asthma at 6 years depending on whether there is in-utero smoke exposure. The GSTM1 null genotype is a protective factor for asthma at age 6 years in girls without in-utero smoke exposure but becomes a risk factor for asthma with in-utero smoke exposure. The associations were not seen for boys. It has been suggested that the relative lack in antioxidant defenses resulting from the GSTM1 null genotype may be compensated for by increased antioxidant production by other members of the GST superfamily [56] or by other antioxidant enzymes. In support of this theory, two studies demonstrated that a protective polymorphism (Ser187) of a second antioxidant enzyme, quinone oxidoreductase 1 (NQO1), provided a protective effect among GSTM1 null subjects [57, 58]. These findings indicate the complexity of the body’s buffering system for antioxidant stress and suggest a direction for future studies. Comprehensive studies in this area should consider both a series of antioxidant genes and dietary “antioxidant” intake, when investigating the relationship between environmental exposures such as ETS and lung health outcomes.

Limitations

We were unable to meta-analyze the data due to high heterogeneity. Specific gene-pollutant outcome combinations lacked numbers for meta-analysis. However, the reporting of significant interaction terms from 15 of the 22 studies provides evidence that it is likely this interaction exists. There is potential for negative studies not to be published, which may have led to a bias towards greater magnitude of interaction effects being identified. Some of the estimates were based on unadjusted measures or unexplained adjustment which may increase the differences between study estimates. Finally, most studies were conducted in Asians or Caucasians and thus, our results may only be applicable to these ethnic groups.

Future Directions

Further studies investigating these gene-environment interactions are needed. The ability of individual studies to measure true effects can be improved by larger sample sizes and more objective measures of exposures and outcomes at multiple time points. The majority of the current papers measured exposures and outcomes at one time by self-report. This is problematic as both exposures and outcomes change over time. Moreover, further studies should also consider recruiting participants in specific age ranges especially people over the age of 65 years, because antioxidant capacity is known to decrease with age [59] and the magnitude of gene-environment interaction may also change over time. More publications, with transparent methods, in varied populations in terms of age and ethnicity should be encouraged to obtain a clearer picture of the true magnitude of the relationship.

Conclusions

There is evidence that carriers of GSTM1 null, GSTT1 null, and GSTP1 val genotypes are more susceptible to indoor air pollution exposure, having a higher risk of asthma and lung function deficits, although some findings are conflicting in terms of risk alleles and specific exposures. These interactions may help to explain the inconsistent effects seen when either indoor air pollution or GST genes are studied in isolation for their effects on respiratory health. The recognition of specific genetic risk alleles may allow for targeted preventive measures in susceptible subpopulations to improve respiratory health by avoiding indoor air pollutants or improving their antioxidant defenses through diet or supplemental vitamins.