Abstract
Purpose
To investigate interactions on gestational age among two environmental risk factors and four maternal genetic polymorphisms: organic solvents, passive smoking, CYP1A1 rs4646903 (MspI), EPHX1 rs2234922 (His139Arg), GSTT1 and PON2 rs12026 (Ala148Gly).
Methods
A pregnant women cohort was conducted at Beijing Yanshan Petrochemical Corporation, and 1,097 mothers with live singleton births were included in analysis. Generalized Multifactor Dimensionality Reduction (GMDR) method was used to explore interactions among these factors with adjustment for important potential confounders. Multiple linear regression models were used to estimate the association of interaction with gestational age.
Results
A three-factor model of organic solvents, GSTT1 and PON2 rs12026 had the highest testing balanced accuracy (57.05 %) and best cross-validation consistency (10/10). Compared with organic solvents unexposed mothers with GSTT1 non-null genotype and PON2 rs12026 CC genotype, organic solvents exposed mothers with GSTT1 null genotype and PON2 rs12026 CG + GG genotype had the largest reduction in gestational age (−0.36 weeks, 95%CI: −0.70 to −0.02). The significant reductions in different groups were from 0.24 weeks to 0.36 weeks.
Conclusions
Maternal genetic susceptibility GSTT1 and PON2 rs12026 could significantly modify the association of organic solvents with gestational age.
Keywords: Gene-environment interaction, Gestational age, Glutathione transferase, Paraoxonase
Introduction
Shortened gestational age, like preterm delivery (gestational age < 37 weeks), contributes greatly to infant mortality [1]. Because of its high prevalence, its association with mortality and morbidity, as well as the cost of hospitalization and long-term disability, shortened gestational age has become a major public health concern [2].
Identifying risk factors is essential to the prevention of shortened gestational age. Organic solvents and smoking are common environmental risk factors. Many epidemiological studies have found associations between organic solvents and adverse pregnancy outcomes [3–6] . As to smoking, both active smoking and passive smoking are associated with the risk of adverse birth outcomes [7–9].
Maternal genetic susceptibility is another type of risk factor. Four functional genes: CYP1A1, EPHX1, GSTT1 and PON2 have received attention. These genes all encode enzymes involving in metabolizing and detoxifying xenobiotics: CYP1A1 encodes a member of cytochrome P450 superfamily enzymes which serves as the major enzyme in phase I metabolism; EPHX1 encodes epoxide hydrolase which can active and detoxify epoxides from the degradation of toxins; GSTT1 encodes glutathione S-transferase (GST) theta-1, a member of superfamily enzymes playing an important role in phase II metabolism; and PON2 encodes a member of paraoxonase family which can protect cells from oxidative stress. Molecular epidemiological studies have provided evidences that these genes were associated with adverse pregnancy outcomes: Wang et al. found that CYP1A1 MspI Aa + aa genotype was associated with low-birth-weight infants (OR = 3.2, 95%CI: 1.6 to 6.4) [10]; Nukui et al. found that GSTT1 null genotype was associated with the risk of premature birth (P = 0.001) [11]; Wu et al. found an association between EPHX1 Tyr113His His/His genotype and reduced birth weight (−167.9 g, 95%CI: −329.6 to −6.1) [12]; and an association between PON2 S311C CC genotype and preterm delivery was found by Chen et al. (OR = 4.6, 95%CI: 1.5 to 14) [13].
There are also gene-gene interactions or gene-environment interactions among these four genes, organic solvents and smoking on adverse pregnancy outcomes. Epidemiological studies have found these interactions on gestational age: Tsai et al. reported a joint association between maternal smoking, CYP1A1 MspI (Aa/aa) and GSTT1 (null) genotypes and preterm delivery (OR = 5.8, 95%CI: 2.0 to 21.1) [14]; Wang et al. found that benzene exposure, CYP1A1 HincII (AA) and GSTT1 (null) could synergistically increase the risk of shortened gestational age (−0.79 weeks, P = 0.002) [15] .However, these studies only searched interactions among limited candidate risk factors. Interactions among all six factors on gestational age are not well studied.
Due to reasons above, in this study, we used the data from a prospective reproductive cohort study conducted in China to explore interactions among all six risk factors above. No subject in this study smoked, so we used passive smoking instead of active smoking. The genetic polymorphisms we chose were CYP1A1 rs4646903 (MspI), EPHX1 rs2234922 (His139Arg), GSTT1 and PON2 rs12026 (Ala148Gly). We hypothesized that there were combined association between these six risk factors and shortened gestational age.
Subjects and methods
Study site and population
This prospective cohort study was conducted at Beijing Yanshan Petrochemical Corporation (BYPC), located in a suburban area of Beijing, China. In BYPC, the major occupational exposure of organic solvents includes benzene, toluene, styrene and their derivatives. The level of organic solvents exposure in BYPC is very low. In study period, the time-weighted average for benzene during an 8-h shift for exposed workers ranged from 0.017 to 0.191 ppm, a centration far below the limit of 1 ppm as an 8-h time weighted average recommended by the Occupational Safety and Health Administration (OSHA) [16].
The study population was female workers who newly got married and wanted pregnancy between June 1995 and June 2000. The eligibility criteria for women who enrolled in this study were as follows: full-time employed workers; live singleton birth; and aged 20–34 years. Women who had multiple gestations, birth with major congenital defects or medically diagnosed gynecological or endocrine disorders were excluded from the study.
Data collection procedures
Eligible women were enrolled by the BYPC Staff Hospital, which is the only regional hospital that provides routine physical examination, family planning counseling, prenatal care and delivery services to female employees of BYPC. After informed consent was obtained, questionnaire investigation was administered by trained interviewers to collect pre-pregnancy baseline information: demographic characteristics, medical and reproductive history and personal habits. A physical examination was performed to obtain pre-pregnancy height and weight according to a standard protocol. If a woman reported a missed or late period or had early signs or symptoms of pregnancy, she was introduced to the hospital for a check-up and to give a urine sample to confirm pregnancy. A woman who was confirmed to be pregnant received regular prenatal care and delivery services according to standard clinical guidelines and was followed-up for pregnancy outcomes. Birth outcomes such as infant sex, gestational age, and birth weight were obtained from medical records by a trained nurse. Blood samples were obtained from mothers via venipuncture by a skilled phlebotomist. The study protocol was approved by the ethics committee of Peking University Health Science Center.
Exposure assessments
The measurement of maternal organic solvents exposure in this study was based on a specialized industrial hygiene method. Briefly, a checklist of 55 potential reproductive hazards present in the industry was developed based on toxicologic literature. This checklist was used to identify potential exposures for each subject. An industrial hygienist who was familiar with the BYPC production process but did not have knowledge of birth outcomes classified those women into exposed or unexposed group according to their workshops and job activities. Pregnant workers stopped working at about 28 weeks of gestation.
Information on other exposures such as passive smoking, noise, vibration, dust, and perceived work stress were also obtained from the questionnaire interview by trained interviewers. For passive smoking, the specific question is “Do your family members smoke at home?”; for noise, the specific question is “Do you work in a noisy environment?”; for vibration, the specific question is “Are you affected by vibration in your work?”; for dust, the specific question is “Are you affected by dust in your work?”; and for perceived work stress, the specific question is “Do you feel stressful when working?”.
Gestational age assessment
In this study, the first day of the last menstrual period recorded on the first prenatal visit was used to estimate the gestational age. This measurement was accurate in this population for several reasons. In China, married couples who plan to have a child need to apply for birth permission at the local family planning administration. Due to the one child policy, families highly concern about healthy pregnancy and healthy babies. All women received a human chorionic gonadotrophin (hCG) assay to verify pregnancy in the BYPC Staff Hospital soon after missing a menstrual period. Furthermore, the gestational age was calculated in exact days then divided by 7 instead of rounded completed weeks.
Genotyping methods
CYP1A1 rs4646903 polymorphism was detected according to the method of Kawajiri et al. [17]. This process resulted in 295-, 160-, and 135-bp products and was able to detect all three genotypes for the polymorphism.
EPHX1 rs2234922 polymorphism was detected according to the method of Hassett et al. [18]. This process resulted in 298-, 177-, and 121-bp products and was able to detect all three genotypes for the polymorphism.
GSTT1 polymorphism was detected according to the method of Nelson et al. [19]. The non-null genotype of GSTT1 was indicated by an amplified 457-bp product, while the null genotype was indicated by the absence of a 457-bp band.
PON2 rs12026 polymorphism was detected by the method of Sanghera et al. [20]. This process resulted in 153-, 123- and 30-bp products and was able to detect all three genotypes for the polymorphism.
Statistical methods
Considering the low number of preterm infants (1.64 %) and to preserve statistical power, we analyzed our data using gestational age as a continuous variable.
At first, we used the χ2 test for goodness-of-fit to check whether the distribution of genotypes in this population deviated from the Hardy–Weinberg Equilibrium. Then we used a univariate linear regression model to investigate individual association between gestational age and different risk factors.
In order to detect interactions among organic solvents exposure, passive smoking, CYP1A1 rs4646903, EPHX1 rs2234922, GSTT1 and PON2 rs12026 on gestational age, we used the Generalized Multifactor Dimensionality Reduction (GMDR) method [21]. Involving multiple factors in a traditional model means lots of multifactorial combinations, and some combinations may have few or no data. One major advantage of the GMDR method is that higher dimensions of multiple factors can be reduced to one dimension (low or high risk) within all possible combinations of factors. It permits adjustment for covariates and is applicable to both dichotomous and continuous phenotypes. Briefly, we at first calculated a score for gestational age with adjustment for covariates as following [15, 22, 23]: age (25, 25–28 and ≥ 28 years), education (elementary or middle, high school, and college or above), pregnancy history (no, yes), infant sex (male, female), pre-pregnancy BMI (<18.5, 18.5–24, 24–27 and ≥27), noise exposure (no, yes), vibration exposure (no, yes), and perceived work stress (no, yes). Then the data was divided into 10 sets: 9 for training and 1 for testing. N factors were selected from the training set and their combinations were represented in n-dimensional space. The GMDR classified each combination (multifactorial class) as ‘high risk’ or ‘low risk’, thus reduced the n-dimensional space to one dimension with two levels. For each possible model size (one-factor, two-factor, etc.), the model with the lowest misclassification error was selected. Leave-one-out cross-validation was used to calculate the prediction error and to evaluate the predictive ability of the model in the test data. The result was a set of models, a final model was chosen on the basis of minimum prediction error and maximum cross-validation consistency. After obtaining the final model, 1000 permutation tests were performed to obtain an empirical P value. Then we used multiple linear regression models to estimate the joint association of factors in the final model with gestational age with adjustment for covariates described above.
All P values were two-sided, and statistical significance was defined as P = 0.05. The GMDR procedure was performed by using the GMDR software Beta Version 0.7 (http://www.healthsystem.virginia.edu/internet/Addiction-Genomics/). The other statistical analyses were performed by using the SAS software Version 9.1 (SAS Institute Inc., Cary, North Carolina).
Results
The study population contained 1,334 mothers who delivered infants. 237 mothers were excluded for having multiple gestations, birth with major congenital defects, failure of extracting DNA, failure of genotyping or missing data. Finally, a total of 1,097 mothers were included in final analysis. Since GSTT1 had only two genotypes: non-null and null, we did not test its departure from the Hardy–Weinberg Equilibrium. The other three polymorphisms followed the Hardy–Weinberg Equilibrium (for CYP1A1 rs4646903, χ2 = 0.521, P = 0.471; for EPHX1 rs2234922, χ2 = 1.180, P = 0.277; and for PON2 rs12026 χ2 = 1.298, P = 0.255).
Individual associations between gestational age and different risk factors in these 1,097 mothers are shown in Table 1. On account of the low frequency of EPHX1 rs2234922 GG genotype, we combined them with AG genotype for analysis, and so as for PON2 rs12026. We found no significant association between gestational age and passive smoking. As to the four polymorphisms, we did not find significant individual association, neither. However, we found that those with BMI < 18.5 or with BMI ≥ 27 had shorter gestational age compared with subjects with normal BMI (18.5–24); organic solvents exposure was associated with shortened gestational age; mothers unexposed to stress tended to have shorter gestational age.
Table 1.
Gestational age by different groups of characteristics, exposures and polymorphisms
| Variables a | No. (%) | Mean (SD) b | β c | 95 % CI d | P value |
|---|---|---|---|---|---|
| Total sample | 1,097 (100.0) | 39.96 (1.30) | |||
| Age | |||||
| <25 | 245 (22.3) | 39.98 (1.47) | reference | ||
| 25–28 | 607 (55.3) | 39.98 (1.27) | 0.00 | −0.19, 0.19 | 0.997 |
| ≥28 | 245 (22.3) | 39.89 (1.20) | −0.09 | −0.33, 0.14 | 0.424 |
| Pre-pregnancy BMI | |||||
| <18.5 | 98 (8.9) | 39.59 (1.31) | −0.37 | −0.64, −0.10 | 0.008 |
| 18.5–24 | 661 (60.3) | 39.96 (1.22) | reference | ||
| 24–27 | 211 (19.2) | 39.95 (1.37) | −0.01 | −0.22, 0.19 | 0.884 |
| ≥27 | 127 (11.6) | 40.24 (1.54) | 0.28 | 0.03, 0.53 | 0.026 |
| Education | |||||
| ≤middle school | 276 (25.2) | 40.07 (1.24) | reference | ||
| high school | 529 (48.2) | 39.96 (1.34) | −0.12 | −0.31, 0.07 | 0.229 |
| ≥college | 292 (26.6) | 39.86 (1.29) | −0.21 | −0.43, 0.00 | 0.053 |
| Infant sex | |||||
| male | 548 (50.0) | 39.89 (1.34) | reference | ||
| female | 549 (50.0) | 40.03 (1.26) | 0.15 | −0.01, 0.30 | 0.065 |
| Pregnancy history | |||||
| no | 636 (58.0) | 39.95 (1.30) | reference | ||
| yes | 461 (42.0) | 39.98 (1.31) | 0.03 | −0.13, 0.18 | 0.721 |
| Organic solvents | |||||
| no | 737 (67.2) | 40.03 (1.27) | reference | ||
| yes | 360 (32.8) | 39.82 (1.36) | −0.21 | −0.37, −0.05 | 0.012 |
| Passive smoking | |||||
| no | 590 (53.8) | 39.95 (1.31) | reference | ||
| yes | 507 (46.2) | 39.97 (1.30) | 0.02 | −0.13, 0.18 | 0.791 |
| Stress exposure | |||||
| no | 734 (66.9) | 39.90 (1.34) | reference | ||
| yes | 363 (33.1) | 40.09 (1.21) | 0.19 | 0.03, 0.36 | 0.021 |
| Noise exposure | |||||
| no | 831 (75.8) | 39.96 (1.27) | reference | ||
| yes | 266 (24.2) | 39.97 (1.40) | 0.01 | −0.17, 0.19 | 0.916 |
| Vibration exposure | |||||
| no | 1,018 (92.8) | 39.96 (1.29) | reference | ||
| yes | 79 (7.2) | 39.97 (1.48) | 0.02 | −0.28, 0.31 | 0.920 |
| Dust exposure | |||||
| no | 1,014 (92.4) | 39.98 (1.25) | reference | ||
| yes | 83 (7.6) | 39.75 (1.79) | −0.22 | −0.51, 0.07 | 0.136 |
| CYP1A1 rs4646903 | |||||
| TT | 419 (38.2) | 39.97 (1.41) | reference | ||
| TC | 527 (48.0) | 39.99 (1.23) | 0.02 | −0.14, 0.19 | 0.778 |
| CC | 151 (13.8) | 39.83 (1.23) | −0.13 | −0.38, 0.11 | 0.283 |
| EPHX1 rs2234922 | |||||
| AA | 871 (79.4) | 39.96 (1.28) | reference | ||
| AG + GG | 226 (20.6) | 39.94 (1.39) | −0.02 | −0.21, 0.17 | 0.811 |
| GSTT1 | |||||
| non-null | 585 (53.3) | 40.00 (1.28) | reference | ||
| null | 512 (46.7) | 39.91 (1.33) | −0.09 | −0.25, 0.06 | 0.248 |
| PON2 rs12026 | |||||
| CC | 713 (65.0) | 39.96 (1.29) | reference | ||
| CG + GG | 384 (35.0) | 39.95 (1.32) | −0.02 | −0.18, 0.15 | 0.842 |
aFor CYP1A1 rs4646903, C is the minor allele; for EPHX1 rs2234922, G is the minor allele; and for PON2 rs12026, G is the minor allele
bSD, Standard Deviation
cβ represents the difference in mean gestational age (weeks) for each group compared reference group in characteristics, exposures and polymorphisms
d95 % CI, 95 % confidence interval
Then we used the GMDR method to detect interactions among organic solvents exposure, passive smoking and those four polymorphisms. Best models of different sizes are shown in Table 2. We found the most significant (P = 0.001) interaction on gestational age among following three factors: organic solvents exposure, GSTT1, and PON2 rs12026. This three-factor model had the highest testing balanced accuracy (57.05 %) and best cross-validation consistency (10/10). It was the best and final model of interaction that we chose. Then we performed 1,000 permutation tests, and the empirical P value for this three-factor model was 0.01.
Table 2.
Results of GMDR analysis on gestational age a
| Model size | Factors in model | Testing balanced accuracy (%) | Cross-validation consistency | P value b |
|---|---|---|---|---|
| One factor | Organic solvents | 53.34 | 10/10 | 0.0547 |
| Two factors | GSTT1,PON2 | 55.03 | 9/10 | 0.0107 |
| Three factors | GSTT1,PON2, Organic solvents | 57.05 | 10/10 | 0.0010 |
| Four factors | CYP1A1,GSTT1,PON2, Organic solvents | 54.05 | 10/10 | 0.1719 |
| Five factors | CYP1A1, EPHX1,GSTT1, PON2, Organic solvents | 50.27 | 10/10 | 0.6230 |
| Six factors | CYP1A1, EPHX1,GSTT1, PON2, Organic solvents, passive smoking | 49.92 | 10/10 | 0.8281 |
aAdjustment for age (25, 25–28 and ≥ 28 years), education (≤middle, high school, ≥college), pregnancy history (no, yes), infant’s gender (male, female), pre-pregnancy BMI (<18.5, 18.5–24, 24–27 and ≥27), noise exposure (no, yes), vibration exposure (no, yes), and perceived work stress (no, yes)
b P value obtained using the nonparametric sign test for testing accuracy
Finally, we used multiple linear regression models to estimate the joint association between GSTT1, PON2 rs12026 and organic solvents exposure and gestational age, and the result is shown in Table 3. We found that except the combination “no organic solvents exposure, GSTT1 null genotype and PON2 rs12026 CG + GG genotype”, all the other combinations could shorten gestational age when compared with the reference combination “no organic solvents exposure, GSTT1 non-null genotype and PON2 rs12026 CC genotype”. Among these six combinations associated with shortened gestational age, four combinations had significant reductions (−0.36 weeks, −0.33 weeks, −0.27 weeks and −0.24 weeks, respectively). The largest significant reduction was found in the combination “organic solvents exposure, GSTT1 null genotype and PON2 rs12026 CG + GG genotype” (−0.36 weeks, 95 % CI −0.70 to −0.02).
Table 3.
Joint association of organic solvents exposure, GSTT1 and PON2 rs12026 with gestational age a
| Organic solvents | GSTT1 | PON2 rs12026 | Adjusted mean | β b | 95 % CI c | P value |
|---|---|---|---|---|---|---|
| No | Non-null | CC | 40.11 | reference | ||
| No | Non-null | CG + GG | 39.84 | −0.27 | −0.52,-0.02 | 0.037 |
| No | Null | CC | 39.87 | −0.24 | −0.46,-0.02 | 0.034 |
| No | Null | CG + GG | 40.25 | 0.15 | −0.13, 0.42 | 0.290 |
| Yes | Non-null | CC | 39.97 | −0.14 | −0.41, 0.14 | 0.326 |
| Yes | Non-null | CG + GG | 39.97 | −0.14 | −0.47, 0.19 | 0.403 |
| Yes | Null | CC | 39.77 | −0.33 | −0.61,-0.06 | 0.017 |
| Yes | Null | CG + GG | 39.75 | −0.36 | −0.70, −0.02 | 0.037 |
aAdjustment for age (25, 25–28 and ≥ 28 years), education (≤middle, high school, ≥college), pregnancy history (no, yes), infant’s gender (male, female), pre-pregnancy BMI (<18.5, 18.5-24, 24–27 and ≥27), noise exposure (no, yes), vibration exposure (no, yes), and perceived work stress (no, yes)
bβ represents the difference in adjusted mean gestational age (weeks) for each combination compared reference combination
c95 % CI, 95 % confidence interval
Discussion
In our study, we explored the interaction among maternal organic solvents exposure, passive smoking and four polymorphisms on gestational age by using the GMDR method. This study has some unique features: 1) it is a study researching on the interaction on gestational age among several candidate environmental risk factors and maternal genetic risk factors; 2) it uses the GMDR method to reduce the dimensionality of multifactor, which can solve the problem “curse of dimensionality” [24]; 3) the study population is an overall low-risk population: non-smoking, non-drinking, optimal maternal age, planned pregnancy among married couples and early prenatal care, which provides a precious opportunity to detect the interaction without potential socio-demographic and environmental confounders.
In this study, we could not provide evidence that there’s individual association among passive smoking, CYP1A1 rs4646903, EPHX1 rs2234922, GSTT1 and PON2 rs12026. This result might due to following reasons: 1) we used passive smoking instead of active smoking. Passive smoking may have less harm than active one that we could not find it in our population. 2) Some significant genetic variants were found in populations other than Chinese population [10, 11, 14]. The allele frequencies in different populations are different. For example, the MAF of CYP1A1 rs4646903 in European population is 0.1; however, it is 0.417 in Asian population.
We found a three-way gene-environment interaction among GSTT1, PON2 rs12026 and organic solvents exposure on the length of gestational age. Organic solvents exposure, even at a level at least five times lower than the safety level required by the Occupational Safety and Health Administration, was significantly associated with shortened gestational age. When considering the two polymorphisms in addition, the largest statistically significant reduction was observed in the combination “organic solvents, GSTT1 null genotype and PON2 rs12026 CG + GG genotype”, suggesting that these two polymorphisms together could modify the association between organic solvents and shortened gestational age.
Detecting risk factors of preterm birth is the primary interest of many studies. However, in our study population, the percentage of preterm birth was 1.64 %, and the magnitude of reduction in gestational age was not of major significance. Since the population is low-risk as described above, the low percentage of preterm birth is reasonable. However, our findings have some important implications: 1) organic solvents exposure and susceptible genotypes are prevalent in general population, so the 0.2–0.4 weeks leftward shift in the gestational age distribution curve among the high-risk population could lead a significantly increased number of preterm births; 2) the result reveals that genetic susceptibility can modify the association between organic solvents exposure and reduced gestational age, which supports the importance of considering genetic susceptibility when evaluating reproductive toxicants.
To our knowledge, this study firstly provides evidence for a gene-environment interaction among GSTT1, PON2 rs12026 and organic solvents on gestational age. Although with no other studies supporting our results, our finding is biologically plausible. Organic solvents have been proved to be reproductive toxicants by epidemiological studies [3, 6, 23, 25]. Since organic solvents are liquid soluble, they can pass through the placenta and can reach the fetus and induce harmful effects [26]. They can induce DNA damage, including sister chromatid exchanges [27], micronuclei formation [28] and chromosomal aberrations [29]. The reproductive toxicity of organic solvents also involves oxidative stress [30, 31]. Glutathione s-transferase serves as a major role in the detoxification of organic solvents. The GSTT1 enzyme is important in protecting against organic solvents damage, such as sister chromatid exchange and the formation of hemoglobin adducts [32, 33]. The GSTT1 gene encodes GSTT1 enzyme, and its complete deletion leads to the loss of enzyme activity [34, 35]. Paraoxonase-2, encoded by PON2 gene, is not well studied for its function so far. Several studies have demonstrated its antioxidative potential [36, 37], which can reduce the toxicity of organic solvents. Little information is available for the association between PON2 rs12026 and paraoxonase-2 activity. However, according to the linkage disequilibrium (LD) information available for Han Chinese in Beijing China in the International HapMap Database (http://www.hapmap.org), PON2 rs12026 and PON2 rs7493 (Ser311Cys, G allele is the minor allele) are in strong LD (D = 1.0, r2 = 1.0), and rs12026 G allele and rs7493 G allele are in one haplotype. Stoltz et al. found that PON2 rs7493 G allele could impair the lactonase activity of paraoxonase-2 [38]. Thus it is plausible that PON2 rs12026 G allele associates with the activity reduction of paraoxonase-2 and the detoxification impairment of organic solvents. Regarding the biological plausibility described above, subjects with GSTT1 null genotype and PON2 rs12026 G allele would be less resistant to organic solvents toxicity. Consistently, our study found that subjects with GSTT1 null genotype and PON2 rs12026 G allele had the shortest gestational age when exposed to organic solvents.
Several methodological limitations should be taken into consideration when interpreting our results. First, subjects of this study were low-risk, so it should take caution to generalize our results to other populations. Second, we combined benzene, toluene, etc. as one exposure, which made it hard to detect the effect of a particular organic solvent. Third, many exposures were assessed based on self-report, which might generate bias. In addition, we adjusted many covariates in the GMDR analysis and regression analysis. However, we could not rule out the possible confounding effects by other uncontrolled or inadequately controlled risk factors.
In summary, searching among six risk factors, we find that maternal genetic susceptibility GSTT1 and PON2 rs12026 significantly modify the association of organic solvents with gestational age, which provides a new insight of gene-environment interaction on gestational age.
Acknowledgment
We sincerely thank all of the participants and their families for participating in this study, and we gratefully acknowledge the invaluable assistance of clinical, field and laboratory staff who contributed to making this work possible. This study is supported in part by grants R825818 from the Environmental Protection Agency, 1R01 HD32505-01 from the National Institute of Child Health and Human Development, 1R01 ES08337-01 from the National Institute of Environmental Health Science and 1R01 OH03027 from the National Institute of Occupational Safety and Health.
Ethical issue
This study was approved by the ethics committee of Peking University Health Science Center.
All participants gave their informed content prior to participating in this study.
Conflict of interest
The authors declare that they have no conflict of interest.
Abbreviations
- CI
confidence interval
- CYP1A1
cytochrome P-450 1A1
- EPHX1
epoxide hydrolase 1
- GMDR
generalized multifactor dimensionality reduction
- GSTT1
glutathione S-transferase theta-1
- hCG
human chorionic gonadotrophin
- OSHA
Occupational Safety and Health Administration
- PON2
paraoxonase2
Footnotes
Capsule Organic solvents exposure and maternal genetic variants: GSTT1 and PON2 rs12026 have an interaction on gestational age in Chinese women.
Contributor Information
Yonghua Hu, Phone: 86-10-82801189, Email: yhhu@bjmu.edu.cn.
Dafang Chen, Phone: 86-10-82802644, Email: dafangchen@bjmu.edu.cn.
References
- 1.Callaghan WM, MacDorman MF, Rasmussen SA, Qin C, Lackritz EM. The contribution of preterm birth to infant mortality rates in the United States. Pediatrics. 2006;118(4):1566–73. doi: 10.1542/peds.2006-0860. [DOI] [PubMed] [Google Scholar]
- 2.Crider KS, Whitehead N, Buus RM. Genetic variation associated with preterm birth: a HuGE review. Genet Med. 2005;7(9):593–604. doi: 10.1097/01.gim.0000187223.69947.db. [DOI] [PubMed] [Google Scholar]
- 3.Llop S, Ballester F, Estarlich M, Esplugues A, Rebagliato M, Iniguez C. Preterm birth and exposure to air pollutants during pregnancy. Environ Res. 2010;110(8):778–85. doi: 10.1016/j.envres.2010.09.009. [DOI] [PubMed] [Google Scholar]
- 4.Forand SP, Lewis-Michl EL, Gomez MI. Adverse birth outcomes and maternal exposure to trichloroethylene and tetrachloroethylene through soil vapor intrusion in New York State. Environ Health Perspect. 2012;120(4):616–21. doi: 10.1289/ehp.1103884. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ahmed P, Jaakkola JJ. Exposure to organic solvents and adverse pregnancy outcomes. Hum Reprod. 2007;22(10):2751–7. doi: 10.1093/humrep/dem200. [DOI] [PubMed] [Google Scholar]
- 6.Slama R, Thiebaugeorges O, Goua V, Aussel L, Sacco P, Bohet A, et al. Maternal personal exposure to airborne benzene and intrauterine growth. Environ Health Perspect. 2009;117(8):1313–21. doi: 10.1289/ehp.0800465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Miyake Y, Tanaka K, Arakawa M. Active and passive maternal smoking during pregnancy and birth outcomes: the Kyushu Okinawa Maternal and Child Health Study. BMC Pregnancy Childbirth. 2013 doi: 10.1186/1471-2393-13-157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Salmasi G, Grady R, Jones J, McDonald SD. Environmental tobacco smoke exposure and perinatal outcomes: a systematic review and meta-analyses. Acta Obstet Gynecol Scand. 2010;89(4):423–41. doi: 10.3109/00016340903505748. [DOI] [PubMed] [Google Scholar]
- 9.Stillerman KP, Mattison DR, Giudice LC, Woodruff TJ. Environmental exposures and adverse pregnancy outcomes: a review of the science. Reprod Sci. 2008;15(7):631–50. doi: 10.1177/1933719108322436. [DOI] [PubMed] [Google Scholar]
- 10.Wang X, Zuckerman B, Pearson C, Kaufman G, Chen C, Wang G, et al. Maternal cigarette smoking, metabolic gene polymorphism, and infant birth weight. JAMA. 2002;287(2):195–202. doi: 10.1001/jama.287.2.195. [DOI] [PubMed] [Google Scholar]
- 11.Nukui T, Day RD, Sims CS, Ness RB, Romkes M. Maternal/newborn GSTT1 null genotype contributes to risk of preterm, low birthweight infants. Pharmacogenetics. 2004;14(9):569–76. doi: 10.1097/00008571-200409000-00001. [DOI] [PubMed] [Google Scholar]
- 12.Wu T, Hu Y, Chen C, Yang F, Li Z, Fang Z, et al. Passive smoking, metabolic gene polymorphisms, and infant birth weight in a prospective cohort study of Chinese women. Am J Epidemiol. 2007;166(3):313–22. doi: 10.1093/aje/kwm090. [DOI] [PubMed] [Google Scholar]
- 13.Chen D, Hu Y, Chen C, Yang F, Fang Z, Wang L, et al. Polymorphisms of the paraoxonase gene and risk of preterm delivery. Epidemiology. 2004;15(4):466–70. doi: 10.1097/01.ede.0000129509.59912.b2. [DOI] [PubMed] [Google Scholar]
- 14.Tsai HJ, Liu X, Mestan K, Yu Y, Zhang S, Fang Y, et al. Maternal cigarette smoking, metabolic gene polymorphisms, and preterm delivery: new insights on GxE interactions and pathogenic pathways. Hum Genet. 2008;123(4):359–69. doi: 10.1007/s00439-008-0485-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang X, Chen D, Niu T, Wang Z, Wang L, Ryan L, et al. Genetic susceptibility to benzene and shortened gestation: evidence of gene-environment interaction. Am J Epidemiol. 2000;152(8):693–700. doi: 10.1093/aje/152.8.693. [DOI] [PubMed] [Google Scholar]
- 16.Occupational exposure to benzene, Docket no.H-059c, 29 CFR part 1910 (1985).
- 17.Kawajiri K, Nakachi K, Imai K, Yoshii A, Shinoda N, Watanabe J. Identification of genetically high risk individuals to lung cancer by DNA polymorphisms of the cytochrome P450IA1 gene. FEBS Lett. 1990;263(1):131–3. doi: 10.1016/0014-5793(90)80721-T. [DOI] [PubMed] [Google Scholar]
- 18.Hassett C, Aicher L, Sidhu JS, Omiecinski CJ. Human microsomal epoxide hydrolase: genetic polymorphism and functional expression in vitro of amino acid variants. Hum Mol Genet. 1994;3(3):421–8. doi: 10.1093/hmg/3.3.421. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nelson HH, Wiencke JK, Christiani DC, Cheng TJ, Zuo ZF, Schwartz BS, et al. Ethnic differences in the prevalence of the homozygous deleted genotype of glutathione S-transferase theta. Carcinogenesis. 1995;16(5):1243–5. doi: 10.1093/carcin/16.5.1243. [DOI] [PubMed] [Google Scholar]
- 20.Sanghera DK, Aston CE, Saha N, Kamboh MI. DNA polymorphisms in two paraoxonase genes (PON1 and PON2) are associated with the risk of coronary heart disease. Am J Hum Genet. 1998;62(1):36–44. doi: 10.1086/301669. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lou XY, Chen GB, Yan L, Ma JZ, Zhu J, Elston RC, et al. A generalized combinatorial approach for detecting gene-by-gene and gene-by-environment interactions with application to nicotine dependence. Am J Hum Genet. 2007;80(6):1125–37. doi: 10.1086/518312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Chen D, Cho SI, Chen C, Wang X, Damokosh AI, Ryan L, et al. Exposure to benzene, occupational stress, and reduced birth weight. Occup Environ Med. 2000;57(10):661–7. doi: 10.1136/oem.57.10.661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Qin X, Wu Y, Wang W, Liu T, Wang L, Hu Y, et al. Low organic solvent exposure and combined maternal-infant gene polymorphisms affect gestational age. Occup Environ Med. 2008;65(7):482–7. doi: 10.1136/oem.2007.032474. [DOI] [PubMed] [Google Scholar]
- 24.Moore JH, Ritchie MD. The challenges of whole-genome approaches to common diseases. JAMA. 2004;291(13):1642–3. doi: 10.1001/jama.291.13.1642. [DOI] [PubMed] [Google Scholar]
- 25.Chevrier C, Dananche B, Bahuau M, Nelva A, Herman C, Francannet C, et al. Occupational exposure to organic solvent mixtures during pregnancy and the risk of non-syndromic oral clefts. Occup Environ Med. 2006;63(9):617–23. doi: 10.1136/oem.2005.024067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kumar S. Occupational exposure associated with reproductive dysfunction. J Occup Health. 2004;46(1):1–19. doi: 10.1539/joh.46.1. [DOI] [PubMed] [Google Scholar]
- 27.Erexson GL, Wilmer JL, Kligerman AD. Sister chromatid exchange induction in human lymphocytes exposed to benzene and its metabolites in vitro. Cancer Res. 1985;45(6):2471–7. [PubMed] [Google Scholar]
- 28.Yager JW, Eastmond DA, Robertson ML, Paradisin WM, Smith MT. Characterization of micronuclei induced in human lymphocytes by benzene metabolites. Cancer Res. 1990;50(2):393–9. [PubMed] [Google Scholar]
- 29.Lazutka JR, Lekevicius R, Dedonyte V, Maciuleviciute-Gervers L, Mierauskiene J, Rudaitiene S, et al. Chromosomal aberrations and sister-chromatid exchanges in Lithuanian populations: effects of occupational and environmental exposures. Mutat Res. 1999;445(2):225–39. doi: 10.1016/S1383-5718(99)00128-X. [DOI] [PubMed] [Google Scholar]
- 30.Kovacic P, Jacintho JD. Reproductive toxins: pervasive theme of oxidative stress and electron transfer. Curr Med Chem. 2001;8(7):863–92. doi: 10.2174/0929867013372878. [DOI] [PubMed] [Google Scholar]
- 31.Murata M, Tsujikawa M, Kawanishi S. Oxidative DNA damage by minor metabolites of toluene may lead to carcinogenesis and reproductive dysfunction. Biochem Biophys Res Commun. 1999;261(2):478–83. doi: 10.1006/bbrc.1999.1041. [DOI] [PubMed] [Google Scholar]
- 32.Schroder KR, Wiebel FA, Reich S, Dannappel D, Bolt HM, Hallier E. Glutathione-S-transferase (GST) theta polymorphism influences background SCE rate. Arch Toxicol. 1995;69(7):505–7. doi: 10.1007/s002040050205. [DOI] [PubMed] [Google Scholar]
- 33.Xu X, Wiencke JK, Niu T, Wang M, Watanabe H, Kelsey KT, et al. Benzene exposure, glutathione S-transferase theta homozygous deletion, and sister chromatid exchanges. Am J Ind Med. 1998;33(2):157–63. doi: 10.1002/(SICI)1097-0274(199802)33:2<157::AID-AJIM7>3.0.CO;2-V. [DOI] [PubMed] [Google Scholar]
- 34.Seidegard J, Vorachek WR, Pero RW, Pearson WR. Hereditary differences in the expression of the human glutathione transferase active on trans-stilbene oxide are due to a gene deletion. Proc Natl Acad Sci U S A. 1988;85(19):7293–7. doi: 10.1073/pnas.85.19.7293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Sprenger R, Schlagenhaufer R, Kerb R, Bruhn C, Brockmoller J, Roots I, et al. Characterization of the glutathione S-transferase GSTT1 deletion: discrimination of all genotypes by polymerase chain reaction indicates a trimodular genotype-phenotype correlation. Pharmacogenetics. 2000;10(6):557–65. doi: 10.1097/00008571-200008000-00009. [DOI] [PubMed] [Google Scholar]
- 36.Ng CJ, Wadleigh DJ, Gangopadhyay A, Hama S, Grijalva VR, Navab M, et al. Paraoxonase-2 is a ubiquitously expressed protein with antioxidant properties and is capable of preventing cell-mediated oxidative modification of low density lipoprotein. J Biol Chem. 2001;276(48):44444–9. doi: 10.1074/jbc.M105660200. [DOI] [PubMed] [Google Scholar]
- 37.Horke S, Witte I, Wilgenbus P, Kruger M, Strand D, Forstermann U. Paraoxonase-2 reduces oxidative stress in vascular cells and decreases endoplasmic reticulum stress-induced caspase activation. Circulation. 2007;115(15):2055–64. doi: 10.1161/CIRCULATIONAHA.106.681700. [DOI] [PubMed] [Google Scholar]
- 38.Stoltz DA, Ozer EA, Recker TJ, Estin M, Yang X, Shih DM, et al. A common mutation in paraoxonase-2 results in impaired lactonase activity. J Biol Chem. 2009;284(51):35564–71. doi: 10.1074/jbc.M109.051706. [DOI] [PMC free article] [PubMed] [Google Scholar]