Durban, South Africa (SA) is a rapidly developing city with increased road traffic and industrial development in close proximity to residential areas. Durban is divided into a heavily industrial south region (also known as the South Durban Industrial Basin) and a less industrial north region; however, both regions are undergoing increased urbanisation and development. This is of major health concern due to increased levels of ambient air pollution (AAP) in both of these areas (Naidoo et al. 2013). Ambient air pollution and associated oxidative stress has been implicated in many pathological conditions, including cancer, asthma, acute respiratory infections, and adverse birth outcomes (Šrám et al. 2005; Kampa and Castanas 2008; Wu et al. 2009; Fleischer et al. 2014; Chen et al. 2015; Moorthy et al. 2015). Pregnant females are highly susceptible to oxidative stress due to increased basal oxygen and changes in energy consumption during placental and foetal development. Infants in utero are highly susceptible to the harmful effects of AAP; exposure has been associated with low birth weight (LBW), inter-utero growth restrictions, preterm birth (PTB), and preeclampsia (Glinianaia et al. 2004; Negi et al. 2012a; Proietti et al. 2013). High levels of pollutants, including sulphur dioxide, carbon monoxide (CO), particulate matter (PM), and nitric oxide (NO x ), have been reported (Kistnasamy et al. 2008). Oxides of nitrogen are of particular interest, because they are by-products of vehicle combustion, smoking, and cooking with gas and have a nitrogen-centred free radical. They interact directly with macromolecules (DNA, lipids, and proteins) and often result in a cascade of radical production and compromised cellular antioxidant function (Kelly 2003).

The increase in reactive oxygen species (ROS) and decrease in antioxidants lead to oxidative stress. The premutagenic deoxyguanosine DNA lesion is highly susceptible to oxidative stress, resulting in the hydroxylation of the guanosine residue at position C8. This produces oxidative 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-OHdG) and acts as a biomarker for oxidative DNA damage (Kim et al. 2005).

Glutathione-S-transferases (GST) are a family of phase II isoenzymes that protect against oxidative stress. This occurs through the conjugation of glutathione to electrophilic species that can react with and form protein or DNA adducts (Romieu et al. 2006). Two common and highly polymorphic antioxidants that have been implicated in health effects in response to chemical exposure are GST mu (M) and pi (P) 1(Wong et al. 2008). Common single-nucleotide polymorphisms (SNP) in the GSTM1 GSTP1 affect the enzymatic activity of GST and have been associated with AAP-associated health effects and increased oxidative stress (Romieu et al. 2006; Mustafa et al. 2010). A homozygous deletion in the GSTM1 gene results in the complete absence of enzyme activity (Seidegård et al. 1988; Xu et al. 1998), whilst a single nucleotide substitution of adenine (A) for guanine (G) results in the amino acid change of isoleucine for valine in GSTP1, at codon 105. This codon forms part of the active site of the enzyme; therefore, this change results in the alteration of the substrate-specificity of the enzymes’ binding site (Johanssona et al. 1998; Moyer et al. 2008). These polymorphisms have been associated with adverse birth outcomes due to AAP, such as increased risk of PTB (Sram et al. 2006; Mustafa et al. 2010; Slama et al. 2010).

To investigate a possible correlation between NO x and maternal oxidative stress, this study measured serum 8-OHdG in third trimester bloods of women living in the south and north regions of Durban. To determine whether NO x and maternal oxidative stress impacted neonatal birth weight (BW) and gestational age (GA), associations were investigated between NO x , 8-OHdG adduct concentration, and neonatal BW and GA. Multivariate analyses were performed to assess whether the relationship held when confounding factors were controlled. The study also investigated the prevalence of GST polymorphisms in South African women and their potential to affect the susceptibility of mothers, who were exposed to AAP, to oxidative stress.

Methodology

Study Population

The Mother and Child in the Environment (MACE) longitudinal cohort study recruited pregnant women from public sector anti-natal clinics in the heavily polluted south Durban (n = 225). A comparison sample of women with similar socioeconomic statuses was recruited from the less-industrialised north Durban (n = 152). The women were residents of the geographical area for the full duration of the pregnancy. Women with hypertension, multiple pregnancies, diabetes, placenta previa, genital tract infections, and other complications that result in adverse growth effects were excluded from the study.

The study was approved by the Biomedical Research Ethics Committee of the University of Kwa-Zulu Natal (BF263/12). Informed consent from study participants was obtained.

Nitric Oxide Levels

Land use regression modelling was used to determine the exposure levels of NO x for individual study participants. The method was developed following the ESCAPE approach (Beelen and Hoek 2010). Model development entailed measurement of NO x at selected locations and regressing these measurements against site-specific a priori defined (i.e., direction of effect) geographic predictors, such as road length, land use types, topography, population, and housing density, in a multivariate regression model. NO x measurements were conducted over two, 2-week periods during mid-winter and mid-summer using Ogawa samplers, which were deployed at 40 randomly selected sites in the north and south Durban areas. The sampling periods selected are representative of the two distinct seasons to occur in Durban (Tyson and Preston-Whyte 2004; Tularam and Ramsay 2013), thus accounting for seasonal variation. The adjusted NO x measurements were then used in model development of which the regression coefficients were applied to each participant. This determined individual NO x exposure levels for each study participant (Muttoo et al. 2017).

Collection and Preparation of Samples

Third-trimester blood was collected from pregnant women during the period between 2013 and 2015. The serum and whole blood was stored (− 80 °C) for analyses. Isolated serum was used for 8-OHdG adduct quantification and whole blood for genotyping of polymorphisms.

Polymorphisms of GSTM1 and GSTP1

DNA was isolated from whole blood using the Qiagen FlexiGene® DNA Kit (as per manufacturer’s instructions). Isolated DNA was quantified using the Nanodrop 2000 spectrophotometer and standardised to 10 ng/µl.

Differentiation polymerase chain reaction (PCR) was performed to assess the GSTM1 polymorphism (n = 372), using β-globin as a reference gene. GSTM1 (215 base pair (bp)) and β-globin (268 bp) PCR products were amplified using 40 pmol of GSTM1 and β-globin primers (Inqaba Biotech, SA; Table 1) in a 30-µl reaction (1 × Green GoTaq Flexi buffer, 1.25 mM MgCl2, 0.5 U GoTaq DNA polymerase (Promega), 200 µM of each deoxyribonucleotide (dNTP), 10 ng DNA template). Initial denaturation was applied (96 °C, 5 min), followed by 30 cycles of denaturation (96 °C, 30 s), annealing (57 °C, 30 s), and extension (72 °C, 30 s), concluding with final extension occurring at 72 °C for 5 min. Amplification products were electrophoresed on agarose gel (4%, 2 µl GelRed) and visualised on the Bio-Rad ChemiDoc™ XRS + System, using the Image Lab™ software. The presence of a single 268 bp is indicative of homozygous null genotype, and the presence of 218 bp indicates either a homozygous positive or heterozygous (wild-type (wt)) genotype.

Table 1 Primer sequences for PCR
Full size table

PCR restriction fragment length polymorphism (RFLP) was used to investigate GSTP1 genotypes (n = 377). A 176-bp PCR product was amplified using 10 pmol of GSTP1 primers (Inqaba Biotech, SA; Table 1) in a 25-µl reaction (1 × Green GoTaq Flexi buffer, 1.5 mM MgCl2, 0.5 U GoTaq DNA polymerase (Promega), 200 µM of each dNTP, 10 ng DNA template). Initial denaturation was applied (96 °C, 5 min), followed by 30 cycles of denaturation (96 °C, 30 s), annealing (55 °C, 30 s), and extension (72 °C, 30 s), concluding with final extension (72 °C, 5 min). Amplification products were electrophoresed on agarose gel (3%, 1 µl GelRed) and visualised. The PCR amplicon underwent restriction endonuclease digestion to determine the presence of the polymorphic restriction site. An overnight digestion (37 °C) was performed in 28-µl reactions: 10 µl PCR product and 18 µl (18 µl nuclease-free water, 2 µl 10 × Buffer Tango, 1 µl Alw261 (BsmA1); Thermo Fisher Scientific). Amplicons completely digested into two restriction fragments (91 bp and 85 bp) were homozygous for G105 allele. The restriction fragments were electrophoresed on an agarose gel (3%, 2 µl GelRed) and visualised on the Bio-Rad ChemiDoc™ XRS + System, using the Image Lab™ software.

Determination of 8-Hydroxydeoxyguanosine

The amount of serum 8-OHdG adduct was determined using a competitive OxiSelect™ Oxidative DNA Damage ELISA Kit (Cell Biolabs, Inc.), with a sensitivity range of 100 pgmL−1–20 ngmL−1. Non-haemolysed serum samples (n = 166) were chosen at random by region from the study population; diluted (1:5) in assay diluent and assayed as per manufacturer’s instructions. A 1:5 dilution was recommended as per manufacturer protocol. An initial experiment was performed using this dilution, and the levels of 8-OHdG fell within the range of the 8-OHdG standards. Therefore, this dilution was used for all subsequent analysis. A standard curve was prepared by using known concentrations of 8-OHdG standards (0–20 ng/ml). The logarithmic equation for the best fit line was used to extrapolate the concentrations of the unknown samples. Each 96-well plate that was used had its own set of standards to ensure that human-error and variation between experiments was accounted for. The final concentration of 8-OHdG adduct was the anti-log multiplied by the dilution factor of 5.

Statistical Analysis

Statistical analyses were performed by using GraphPad Prism V5 Software Package (GraphPad Software Inc., San Diego, CA). Comparisons between north and south groups for maternal and neonatal characteristics, atmospheric NO x levels, and maternal serum 8-OHdG adduct concentrations were determined by using the Student t test. Data were log transformed to ensure normalcy and to allow graphical representation of the data. Correlations among atmospheric NO x , maternal serum 8-OHdG, and neonatal BW and neonatal GA, and genotypes of GSTM1 and GSTP1 were performed by using the nonparametric Spearman correlation. Chi square and Fischer’s exact tests were used to test the significant difference in the prevalence of GSTM1 and GSTP1 genotypes between the north and south groups. One-way ANOVA was performed to determine the level of difference for maternal serum 8-OHdG adduct concentrations among the genotypes of GSTM1 and GSTP1. All statistical tests were two-sided. Multivariate linear regression was used to determine whether the relationship between atmospheric NO x and maternal serum 8-OHdG was affected by potential confounders, namely: maternal age, maternal body mass index (BMI), HIV status, area, maternal systolic and diastolic blood pressure (BP), and haemoglobin (Hb) levels. The relationship between neonatal GA and atmospheric NO x and maternal serum 8-OHdG also was determined by using linear regression, and potential confounders were controlled. The linear regression analyses were performed using STATA version 13.1.

Results

Maternal and neonatal characteristics of study participants are described in Table 2. The maternal age was slightly higher in the north compared with the south (p = 0.0844). Mothers in the north were significantly shorter with higher body mass index (BMI) levels compared with the south mothers who were taller and had lower BMI levels (p < 0.0001). Maternal Hb levels were higher in the north compared with the south (p = 0.0002). The systolic and diastolic BP measurements were significantly lower in the north compared with the south (p < 0.05). Mean GA and BW were lower in the north compared with the south (p = 0.0540 and 0.0951 (Table 2), respectively) but did not reach significance.

Table 2 Maternal and neonate characteristics
Full size table

The level of atmospheric NO x in the south (37.04 ± 7.46) was significantly greater than in the north (33.26 ± 8.51, p < 0.0001; Fig. 1a). This corresponds to a significant increase in maternal serum 8-OhdG concentration observed in the south (20.26 ± 40.88 ng/mL) compared with the north (11.51 ± 18.51 ng/mL, p = 0.0197; Fig. 1b).

Fig. 1
figure 1

Concentrations of a atmospheric NO x (µg/cm3) and b maternal serum 8-OHdG (log (ng/mL)) for patients living in the north (a n = 142; b n = 59) and south (a n = 185; b n = 97). Statistical significance: ***p < 0.0001 and *p < 0.05

Full size image

A significant positive correlation (Spearman r = 0.2173; p = 0.0158) was found between the levels of maternal serum 8-OHdG and atmospheric NO x (Fig. 2a). When the specific area was taken into consideration, the south (Spearman r = 0.2337; p = 0.0466) atmospheric NO x correlated significantly with maternal serum 8-OHdG concentration (Fig. 2c), whereas the north (Spearman r = 0.1270; p = 0.3795) showed a positive trend, although not significant (Fig. 2b). No relationship was observed between NO x and neonate BW (Spearman r = 0.09865; 95% confidence interval [CI] − 0.016 to 0.210; p = 0.0814) and NO x and neonatal GA (Spearman r = − 0.0716; 95% CI − 0.185 to 0.043; p = 0.0814). When area was considered, a negative trend is suggested for NO x and BW in both north (Spearman r = − 0.1135; 95% CI − 0.282 to 0.062; p = 0.19) and south (Spearman r = − 0.060; 95% CI 0.211 to 0.093; p = 0.43). A similar negative trend is observed between NO x and GA in both the north (Spearman r = − 0.1135; 95% CI − 0.282 to 0.062; p = 0.19) and south (Spearman r = − 0.060; 95% CI − 0.211 to 0.093; p = 0.43). Maternal serum 8-OHdG was not shown to correlate with BW (Spearman r = 0.058; 95% CI − 0.1521 to 0.2635; p = 0.58) and GA (Spearman r = 0.044; 95% CI − 0.121 to 0.207; p = 0.59).

Fig. 2
figure 2

Relationship between the concentrations of atmospheric NO x (log (µg/cm3)) and maternal serum 8-OHdG (log (ng/mL)) for all patients (a) (Spearman r = 0.2173; 95% CI 0.03657–0.3843; * p = 0.0158; n = 123), patients living in the north (b) (Spearman r = 0.1270; 95% CI − 0.1652–0.3987; p = 0.3795; n = 50) and south (c) (Spearman r = 0.2337; 95% CI − 0.003195–0.4457; *p = 0.0466; n = 73). Dotted lines represent 95% CI interval. Statistical significance: *p < 0.05

Full size image

Maternal HIV status was thought to influence maternal serum 8-OHdG adduct concentration; however, no difference between HIV-positive (16.02 ± 21.39 ng/mL) and HIV-negative (18.48 ± 40.23 ng/mL) 8-OHdG concentration was observed (Fig. 3).

Fig. 3
figure 3

Maternal serum 8-OHdG adduct [log (ng/mL)] concentration between HIV-positive (n = 52) and HIV-negative (n = 113) patients in the total sample group

Full size image

The genotypic prevalence of GSTM1 and GSTP1 amongst study participants are shown in Tables 3 and 4, respectively. For all subjects, the prevalence of the GSTM1 null (0/0) and wt (0/+ or +/+) type was 25.3 and 75.7%, respectively. The frequencies of the A105 and G105 allele of GSTP1 were 44.2 and 55.8%, respectively. A significantly greater fold increase (p = 0.0144) was observed in the south for the GSTP1 A105 allelotype compared to G105 allelotype (1.8 and 1.2, respectively). The prevalence of GSTP1 AA (wt) and AG+GG (variant) (p = 0.0281) was significantly different between the north and south; whilst the prevalence of GSTM1 (p = 0.9038) did not differ significantly. The GSTP1 AG and GG genotypes were combined for analysis, because subjects with a single GSTP1 G105 allele have reduced enzyme activity compared with those with the GSTP1 A105 allele (Zimniak et al. 1994).

Table 3 Genotypic frequency of GSTM1 in patients living in the North and South (n = 372)
Full size table
Table 4 Genotype and allele frequencies of GSTP1 in patients living in the north and south (n = 377)
Full size table

No difference was observed in the levels of maternal serum 8-OHdG among the GSTM1 genotypes between the north and south (Fig. 4a). Maternal serum 8-OHdG concentration was higher in AA genotyped mothers (29.31 ± 10.10 ng/mL, n = 38) compared with AG+GG genotyped mothers (13.16 ± 1.772, n = 127, p = 0.1589), although not significant. When area was considered, the level of maternal serum 8-OHdG was higher in the north GSTP1 AA (21.11 ± 8.9 ng/mL) mothers compared with the GSTP1 AG+GG (8.70 ± 2.0 ng/mL) mothers but did not reach significance (p = 0.1421). The south GSTP1 AG+GG (13.50 ± 2.0 ng/mL) mothers had significantly greater levels of maternal serum 8-OHdG compared with the north (8.70 ± 2.0 ng/mL, p = 0.0188; Fig. 4b).

Fig. 4
figure 4

Maternal serum 8-OHdG adduct [log (ng/mL)] concentration between the GSTM1 (a) and GSTP1 (b) genotypes for patients living in the North and South. Statistical significance: * p < 0.05

Full size image

The relationship between the atmospheric NO x and maternal serum 8-OHdG concentration was investigated among the different genotypes in Table 5. A significant correlation was observed between the atmospheric NO x and the level of maternal serum 8-OHdG within the GSTM1 null genotype (Spearman r = 0.4227, p = 0.0199) and the GSTP1 AA + GG genotype (Spearman r = 0.2105, p = 0.0395). No relationship was observed when north and south was taken into consideration.

Table 5 Relationship between the concentrations of atmospheric NO x (log (µg/cm3)) and maternal serum 8-OHdG (log (ng/mL)) among the different SNP genotypes (GSTM1 and GSTP1) for the total patient sample and those living in the north and south
Full size table

The relationship between atmospheric NO x and maternal serum 8-OHdG concentration, controlled for potential confounders, is described in Table 6. The β-coefficient and regression model for the total samples was strengthened when maternal characteristics were controlled (unadjusted = 1.14, adjusted2 = 1.21, p < 0.05). Every 1% increase in NO x would lead to a 1.21% increase in maternal serum 8-OHdG. This almost doubled when area was taken into consideration: a 1% increase in atmospheric NO x would result in a 1.99% increase in maternal 8-OHdG in the south (p = 0.004). When polymorphisms were considered, a 1% increase in atmospheric NO x would lead to a 1.14% increase in maternal 8-OHdG for the variant GSTP1 genotype (p = 0.026), whereas an almost equal change in percent was observed between NO x and 8-OHdG for the GSTM1 wt genotype (p = 0.076). The neonate gender was shown to influence the concentration of maternal serum 8-OHdG: a 1% increase in NO x would result in a 2.48% increase in maternal serum 8-OHdG for women carrying female neonates (p = 0.002).

Table 6 Impact of atmospheric NO x (log) concentration on maternal serum 8-OHdG adduct (log) concentration—linear regression analysis for total mothers and subdivided into area, GSTM1 and GSTP1 SNP with adjustments for maternal characteristics
Full size table

The relationship between atmospheric NO x and GA, controlled for potential confounders, is described in Table 7. Controlling for potential maternal and neonate confounders strengthened the β-coefficient and regression model for the total samples (unadjusted = − 0.464, adjusted2 = − 1.64, p < 0.05). Therefore, for every 1% increase in NO x , there would be a 0.0164% reduction in GA. The maternal serum 8-OhdG did not significantly impact GA. However, when it was not included in the regression model, NO x was not found to influence GA. Controlling for 8-OHdG, therefore, was an important factor for NO x influencing GA. When area was considered, a 1% increase in NO x would result in a 0.0196% decrease in GA (p = 0.023). The GSTP1 variant caused a 0.013% reduction in GA, whilst the GSTM1 wt genotype caused a 0.0126% decrease in GA if a 1% increase in NO x is observed (p < 0.05; Table 7). Mothers carrying a male foetus were significantly associated with a 0.0156% decrease in GA if NO x were to increase by 1% (p = 0.027). This decrease was almost doubled (0.0203% change) with mothers carrying a female foetus, although not significant (p = 0.097).

Table 7 Impact of atmospheric NO x (log) concentration on gestational age—linear regression analysis for total mothers and subdivided into area, GSTM1, and GSTP1 SNP with adjustments for maternal characteristics
Full size table

Discussion

Exposure to NO x was shown to influence maternal serum 8-OHdG concentrations directly in pregnant women living in Durban, SA. Gestational age of these women also was shown to decrease significantly as a result of increased NO x exposure. This is the first study in Durban, SA to link the increase in oxidative stress in pregnant women to increased NO x pollution.

Oxides of nitrogen, a by-product of combustion, have been linked to several adverse health conditions, including respiratory diseases, cardiovascular diseases, low birth weight, and preterm birth (Seo et al. 2007; Wu et al. 2009; César et al. 2015). This nitrogen-centred free radical, upon inhalation, is absorbed in lung fluids producing free radical products that enter the blood stream. These free radicals are then able to react directly with macromolecules (protein, lipids, and DNA) present resulting in ROS production (Tabacova et al. 1998). Guanine, having the lowest redox potential among the nucleic bases, is highly susceptible to oxidation by ROS, which results in the production of 8-OHdG mutagenic lesions (Kershaw and Hodges 2012; Ba et al. 2015). Therefore, this accounts for the significant increase in maternal serum 8-OHdG adduct concentration as a result of NO x exposure (Figs. 1 and 2).

The Durban south region has been shown previously to have higher levels of air pollution (Naidoo et al. 2013) and in the present study with significantly increased NO x concentration (Fig. 1a) compared with the north. Previously, pregnant women in the south region have been shown to exhibit increased markers of oxidative stress compared with the north (Nagiah et al. 2015). This finding was corroborated in our study, where pregnant women in the south had increased levels of serum 8-OHdG compared with the north (Fig. 1b). Our study went further and investigated the relationship between NO x and 8-OHdG, the influence of GST polymorphisms, as well as linking NO x to GA.

Several studies have found an association of pollution (polycyclic aromatic hydrocarbon (PAH), diesel-exhaust smoke, and smoking) to increased 8-OHdG and oxidative stress (Risom et al. 2005; Lewtas 2007; Leonardi-Bee et al. 2008; Ren et al. 2010). Studies also have observed that a dose–effect relationship occurs between PAH exposure and levels of urinary 8-OHdG (Kuang et al. 2013; Li et al. 2015). Our study found a significantly positive correlation between atmospheric NO x and maternal serum 8-OHdG (Fig. 2a, p = 0.158). As mentioned above, the south region is considered to have a higher pollution level than the north. When area was taken into consideration, the relationship between NO x and 8-OHdG was strengthened in the heavily industrialised south (Fig. 2c, p = 0.0466) whilst the relationship was lost in the less industrialised north (Fig. 2b). This relationship was further investigated by controlling for potential confounding factors to determine whether this effect was indeed a response to NO x exposure. The results confirmed that NO x exposure caused a significant increase in 8-OHdG concentration when controlled for maternal characteristics (Table 7). It was found that a 1% increase in NO x results in a 1.21% (p = 0.015) increase in 8-OHdG, with an even higher increase (1.99%, p = 0.004) observed for south living mothers. The results provide evidence that exposure to atmospheric NO x increases serum 8-OHdG levels in pregnant women.

Antioxidants are an important controller of oxidative stress by helping to reduce and eliminate oxidants to prevent oxidative stress-related damage. However, genetic susceptibility plays an important role in determining the effect and responses an individual has to oxidative damage. Therefore, the risk for cancer, adverse reproductive outcomes, and cardiovascular diseases are a consequence of air pollution exposure and genetic susceptibility (Lewtas 2007; Lagadu et al. 2010). GSTs are antioxidant enzymes that protect against oxidative stress by conjugating electrophilic species, thereby neutralising their effect. These GSTs have two common polymorphisms that influence an individual’s genetic susceptibility to oxidants. A homozygous deletion in GSTM1 results in the enzyme inactivation (Mustafa et al. 2010), whereas the substitution of isoleucine for valine at codon 105 in GSTP1 reduces substrate specificity (Wong et al. 2008).

Our results indicate that GSTP1 is associated with the heavily industrialised south (Table 4, p = 0.0281), with increased serum 8-OHdG observed in the GSTP1 AG+GG genotypes compared with the AA genotype. This suggests that wt GSTP1 is able to scavenge oxidants more effectively than the variant, which leads to reduced serum 8-OHdG concentrations observed. When subdivided into areas, no difference in 8-OHdG was observed in north and south wt GSTP1 whilst the north variant mothers had significantly lower 8-OHdG levels compared with the south variant mothers (Fig. 4b, p < 0.05). This suggests that at low pollution levels (north) the variant GSTP1 enzymes, with its reduced specificity is still able to conjugate electrophiles and reduce their effect whilst at higher pollution levels (south) the variant genotype is overwhelmed and becomes inefficient at scavenging, which leads to increased oxidants present that attack DNA leading to increased 8-OHdG levels.

It has been shown that the variant GSTP1 allele has a sevenfold greater efficacy against PAH diol epoxides than wt allele while threefold less effective against 1-chloro-2,4-diinitrobenzene (Strange et al. 2000). However, this study showed that the variant genotype is less efficient compared with the wt at detoxifying oxidant products. A significant positive correlation was observed between NO x and 8-OHdG levels in AG+GG GSTP1 genotyped mothers (Table 5; p = 0.0395), further confirming that the variant GSTP1 mothers when exposed to NO x are unable to effectively scavenge oxidants leading to increased DNA damage. No association was observed between GSTM1 and the heavily industrialised south (Table 3), with no difference in serum 8-OHdG levels observed between the null and wt GSTM1 genotypes. A significant positive correlation was observed between NO x and serum 8-OHdG for null GSTM1 mothers (Table 5; p = 0.0199), thus suggesting that the inactive enzyme GSTM1 was unable to neutralise oxidants leading to increased DNA damage. These results were further analysed in multivariate analyses to determine whether the results observed in bivariate correlations remained when controlling for maternal characteristics as potential confounders.

The GSTP1 variant mothers had increased serum 8-OHdG (1.14%) with increasing NO x (1%, p = 0.026; Table 6), whilst in GSTM1 null mothers this relationship was lost. This could be a result of small sample size, once all confounding factors were taken into account, only 30 GSTM1 null mothers remained, which could account for the lack of association between NO x and 8-OHdG. Other studies have shown no association between GSTM1 null individuals, DNA damage, and pollution (PAH (Marczynski et al. 2002; Garte et al. 2007) and particulate matter (Sørensen et al. 2003)). The multivariate analysis, however, suggests a parallel increase in 8-OHdG (1.01%) as a result of NO x (1%) exposure for GSTM1 wt mothers (Table 6, p = 0.076). The results provide evidence that GSTP1 variant genotype increases the susceptibility of mothers to NO x exposure, leading to increased oxidative stress.

Pregnancy, a physiological state characterised by increased basal oxygen demand and high-energy requirement, favours ROS production and has been shown to exhibit increased susceptibility to oxidative stress in normal pregnancies (Saker et al. 2008). This already highly susceptible condition in the presence of high AAP would exasperate antioxidant stores and lead to increased oxidative stress. Exposure to traffic-related air pollution (i.e., NO x , CO, and primary exhaust particles) have been implicated in deceased foetal growth, LBW, and PTB (Seo et al. 2007; Darrow et al. 2011). Several studies also have reported oxidative stress as a potential mechanism for LBW and PTB, with reports showing increased 8-OHdG in LBW and PTB (Kim et al. 2005; Mustafa et al. 2010; Rossner et al. 2011).

Our study first set out to find an association between NO x and serum 8-OHdG with neonate BW and GA. Using simple correlations, no association was observed between serum 8-OHdG and BW, with a negative trend suggested between NO x and BW in north and south. This negative trend was not significant; however, this could be due to our small sample size, because previous studies have shown a link between AAP and BW reduction (Lacasana et al. 2005; Darrow et al. 2011; Wilhelm et al. 2012). Next, our study used bivariate linear regression to determine whether NO x was associated with reductions in GA. The results from the unadjusted bivariate analysis again suggested a negative trend but remained nonsignificant, which also was observed using simple correlation. We then controlled for maternal and neonatal characteristics, which revealed a small, significant reduction in gestational age as a result of NO x exposure (Table 7, p = 0.042). This relationship was only found to be significant when maternal serum 8-OHdG was controlled for, suggesting 8-OHdG may affect GA.

Previous studies found increased 8-OHdG in mothers who give birth prematurely (Matsubasa et al. 2002; Nassi et al. 2009; Darrow et al. 2011; Negi et al. 2012b). When area was considered, the Durban north mothers showed a significantly higher reduction in GA compared with the total mothers, whilst Durban south mothers showed no reduction. This study only measured the levels of NO x ; however, other pollutants or environmental factors may have been present in high concentrations within the north area. These pollutants or environmental factors may be potent enough to cause the reduction in GA observed within the north. This would be an ideal follow-up study. The GTSP1 variant and GSTM1 wt mothers were susceptible to GA reduction as a consequence of NO x exposure (p = 0.034 and p = 0.038, respectively; Table 7). Neonatal gender also was found to be associated with reduced gestational age due to NO x exposure; mothers carrying male infants exposed to NO x had a significant reduction in GA. These reductions in GA observed were small (< 0.1 weeks per percent change in NO x ) and, therefore, would not impact clinical significance for individual neonate. However, a negative shift in GA on the population level could result in increased PTB nationwide.

The findings of our study must be interpreted in light of the following limitations. First, upon subdivision, by area and genotype, our sample size becomes relatively small. Second, due to our relatively small sample size, ethnicity was not taken into account; this could be addressed in a future study with increased sample numbers. Although significant results are obtained, increasing population numbers in future studies could give further insight into the conclusions observed. Third, measuring other pollutants in study areas could give insight into their effects on 8-OHdG levels and GA. A combined effect of NO x and other pollutions, such as particulate matter, also would be interesting to investigate.

Conclusions

This study demonstrated increased maternal serum 8-OHdG in pregnant women exposed to higher levels of NO x pollution in the south. This increase in DNA damage was found to be a direct consequence of increased NO x exposure, with increased susceptibility found in GSTP1 variant carriers and GSTM1 wt carriers. Gestational age also was found to be reduced as a consequence of NO x exposure; male neonates making mothers more susceptible to GA reduction. This study highlights the need for better systems in place to reduce traffic-related air pollution close to residential areas, so that vulnerable individuals are better protected against oxidative stress related injury.