Association Analysis of 14 Candidate Gene Polymorphism with Depression and Stress among Gestational Diabetes Mellitus

The association of candidate genes and psychological symptoms of depression, anxiety, and stress among women with gestational diabetes mellitus (GDM) in Malaysia was determined in this study, followed by the determination of their odds of getting psychological symptoms, adjusted for socio-demographical background, maternal, and clinical characteristics. Single nucleotide polymorphisms (SNPs) recorded a significant association between SNP of EPHX2 (rs17466684) and depression symptoms (AOR = 7.854, 95% CI = 1.330–46.360) and stress symptoms (AOR = 7.664, 95% CI = 1.579–37.197). Associations were also observed between stress symptoms and SNP of OXTR (rs53576) and (AOR = 2.981, 95% CI = 1.058–8.402) and SNP of NRG1 (rs2919375) (AOR = 9.894, 95% CI = 1.159–84.427). The SNP of EPHX2 (rs17466684) gene polymorphism is associated with depression symptoms among Malaysian women with GDM. SNP of EPHX2 (rs17466684), OXTR (rs53576) and NRG1 (rs2919375) are also associated with stress symptoms.


Introduction
Gestational diabetes mellitus (GDM) is one of the common complications in pregnancy. Its prevalence in Asia is 11.5% [1]. GDM is a known risk factor for neonatal adverse outcomes [2][3][4]. Additionally, a diagnosis of GDM is a stressful life event [5][6][7][8] which has an adverse impact on self-perception towards health and quality of life [6,9]; as well as increased odds of experiencing emotional distress. Previous studies reported that the prevalence of depression among women sufferers from GDM stood at 56.7%, while anxiety was 57.7%, and stress was even higher at 62.8% [10][11][12]. GDM and perinatal mental problems undeniably affect all members of the family [13]. This mental condition may reoccur or worsen to postpartum depression [14]. Multiple determinants such as socio-demographical background, maternal and clinical profiles have a reported positive association with psychological symptoms [15][16][17][18][19].
Genetic factors clearly play a substantial role in the etiology of psychological symptoms of depression, anxiety and/or stress, as evidenced by other studies, which indicate a heritability ranges from 45% to 50% for these disorders [20][21][22]. The genetic profile of the mother is particularly important if she wants to determine whether her child will be predispose to psychological disorders in the future. However, it is challenging to identify particular genetic variants underlying for symptoms of depression, anxiety and/or stress susceptibility because their psychological symptoms are not caused by single gene, but a complex interaction among multiple genes, socio-demographic background, clinical, and biological moderators [23]. The candidate gene-by-environment interaction hypothesis regarding psychological symptoms of depression, anxiety and/or stress has received widespread attention and acclaim; therefore, many studies to date have used this approach to underpin their findings for genetic effects on psychological symptoms of depression, anxiety, and/or stress [24].
Indeed, it is not difficult to find studies which have reported a significant association between candidate genes and these psychological symptoms, such as brain-derived neurotrophic factor (BDNF) [25,26] and oxytocin receptor genes (OXTR) [27,28]. These genes may be associated with depression or anxiety; however, there are ample studies which have failed to replicate the same results in the candidate gene literature [29][30][31]. One explanation for this lack of success in producing the replicable main effect of these genes is that the certain genetic variants are highly dependent on the gender, population, and disease-related outcomes [32]; even though these studies have recruited patients with major depressive disorder [27][28][29][30][31][32][33][34][35][36][37][38][39][40][41]; anxiety disorder [42][43][44]; and post-traumatic stress disorders [45] diagnosed according to Diagnostic and Statistical Manual of Mental Disorders and/or International Statistical Classification of Diseases. This has led to increasing skepticism about the true association or lack thereof between candidate genes and psychological symptoms of depression, anxiety and/or stress. Without testing the candidate genes in our population, it is difficult to conclude that the previous results are also applicable in our samples. One strategy that may aid in identifying the candidate genes in association with symptoms of depression, anxiety and/or stress is to interrogate several candidate genes thought to be associated with the underlying psychological symptoms of depression, anxiety and/or stress. To this end, we have constructed a custom of SNP array containing 18 genes that were chosen based on hypotheses regarding biological systems of relevance to depression [46][47][48][49][50]; anxiety [42,51,52] and stress [45,53]. These custom SNPs provide excellent coverage of many previously suggested and functionally important candidate genes for depression, anxiety and stress, including NPY5R [42,52]; ANO2 [42]; EPHX2 [42,51]; TPH2 [35]; NRG1 [34]; LHPP [38,39,54]; FKBP5 [41,45]; SDK2 [42]; RORA [33,55]; OXTR [27,28]; BDNF [56,57]; HTR2C [43]; TEX51 [42]; and PLEKHG1 [42]. Many of the genes represented on the array have also been reported to be involved in associated heritable phenotypes that are associated with symptoms of depression, anxiety and/or stress. Despite that, the putative susceptibility genes for depression, anxiety or stress have yet to be definitively identified among GDM women.
In light of the complications caused by GDM itself and the devastating consequences of depression and related psychological symptoms of anxiety and stress among women with GDM, we suggest performing a study of fourteen candidate genes to elucidate its genotypic effect on symptoms of depression, anxiety and/or stress among GDM women. The aim of the present study was to perform candidate gene analysis via mass array to evaluate the associations, if any, between phenotypes of threes psychological symptoms and fourteen candidate genes, as adjusted for socio-demographical background, maternal and clinical profile among GDM women.

Study Population
We performed a post-hoc exploratory sub-analysis of a cross-sectional study among GDM women (n = 343) to check which candidate SNPs may be associated with symptoms of depression, anxiety and/or stress in this particular population. We conducted a genetic association study using the cross-sectional study from the previously described "Prevalence and factors associated with depressive, anxiety and stress symptoms among women with gestational diabetes mellitus in tertiary care centres: A cross-sectional study", which was conducted between July 2018 and October 2018 in Malaysia [58]. The study participants were women enrolled in second or third trimester care and diagnosed with GDM at Hospital Kuala Lumpur or Hospital Serdang. All participants were native Malaysians and residents of surrounding areas. The detailed study protocol has been described previously [58]. In that study, 526 women agreed to participate. Upon completion of sample collection and analysis, data for depression, anxiety and stress score and polymorphisms of candidate genes were available for a total of 343 participants.
The general inclusion criteria were that the pregnant women were Malaysian, aged ≥18 years old, with a diagnosis of GDM. The diagnosis of GDM is defined as fasting plasma glucose ≥5.1 mmol/L or 75 g two-hours oral glucose tolerance test ≥ 7.8 mmol/L according to Malaysian Clinical Practice Guidelines [59,60]. The exclusion criteria were those with pre-existing diabetes.
Regarding patients and controls, patients with depression were defined as those with the DASS depression subscale score ≥10; otherwise, they were in control group if scoring <10 in the DASS-depression subscale. Similarly, they were categorized as a patient for anxiety if they scored ≥8 in the DASS anxiety subscale; they were in control group if the score was <8. They were categorized as a patient for stress if they scored ≥15 in the DASS stress subscale, and placed in the control group if scoring <15 in the DASS stress subscale.

Socio-Demographic Background and Clinical Characteristics
Socio-demographic backgrounds and clinical characteristics were recorded at enrollment to obtain information related to maternal profile, past-obstetrics history, concurrent medical problems, and family history. These data were obtained from the self-administered questionnaire and medical records.

Measurement of Depression, Anxiety and Stress Symptoms
The detailed sampling and assessment of depression, anxiety, and stress symptoms have been previously described [58]. We used an English [61] and Malay [62] version of the validated questionnaire on Depression, Anxiety, and Stress-21 items (DASS-21). DASS-21 is a valid and reliable measure to screen for depression, anxiety, and stress symptoms among both non-clinical and clinical populations. The English version of the questionnaire (DASS-21) has strong validation, with Cronbach's alpha values of 0.72 for depression; 0.77 for anxiety; and 0.70 for stress, and the overall Cronbach's alpha for DASS-21 is 0.88 [61]. The translated Malay version of the DASS-21 questionnaire has good Cronbach's alpha values, as well as among the Malaysian population (0.84 for depression; 0.74 for anxiety; and 0.79 for stress) [62] and among diabetic patients (0.75 for depression; 0.74 for anxiety; and 0.79 for stress) [63]. The participants rated on a 4-point severity scale their experiences over the preceding week. Scores for subscale for depression, anxiety, and stress were calculated. The depression symptoms defined to follow the depression subscale, ≥10; anxiety symptoms, ≥8; and stress symptoms, ≥15 [61].

Blood Sample Collection and DNA Extraction
Samples of 5 mL of blood were collected from the participants' peripheral blood using a 21-gauge needle with a 5.0 mL syringe by a qualified phlebotomist into EDTA tubes (Becton Dickinson, East Rutherford, NJ, USA). The samples were kept in portable icebox at 4 • C during the transportation and there were stored at −20 • C in laboratory for further analysis. Genomic DNA was isolated by using the QIAamp Blood DNA Mini Kit (QIAGEN, Hilden, Germany). The quantity and purity of extracted DNA were checked using a Biophotometer (Eppendorf, Hamburg, Germany). First, readings of a blank using distilled water against A260 and A280 of the genomic DNA were obtained. The DNA absorbed UV light with a maximum absorbance of 260 nm, while the protein absorbed UV light with a maximum absorbance of 280 nm. By dividing the amount of UV absorption at 260 nm by the absorption at 280 nm, the standard measure of the purity of the genomic DNA could be calculated. The genomic DNA was measured to be relatively free of protein impurity when the ratio of optical density was between 1.7 and 2.0.

Statistical Analysis
We used IBM SPSS Statistics version 21.0 to perform the data analysis. A chi-square goodness-of-fit test was performed to assess the agreement of the genotype distribution among candidate genes using Hardy-Weinberg equilibrium, if the p-value for chi-square goodness-of-fit tests is significant (p < 0.05), the population is not in Hardy-Weinberg equilibrium. If the genotype distribution of candidate genes is not fit to Hardy-Weinberg equilibrium based on equal distribution, expected values for genotype distribution will be adjusted according to the global population. Univariate analysis was used to analyze the association between candidate genes and the presence of depression, anxiety, or stress symptoms among GDM women. The significant difference was set to p-value < 0.05. In addition, we tested the candidate gene polymorphism associations with depression phenotypes and any polymorphism adjusted for socio-demographical and clinical moderator effects. Variables with a p-value of less than 0.25 in univariate analysis underwent multiple logistic regression [64], because a p-value set at <0.05 may miss any variables known to be important [65,66]. A backward stepwise regression method was used [67]. All analyses were made with a 95% CI, and the level of significance was set at p < 0.05.

Ethical Consideration
The study was conducted after written informed consent was obtained from all participants. The Medical Research Ethics Committee (MREC), Ministry of Health Malaysia approved the study protocol (NMRR-17-2264-37814).

Results
Overall, we found that almost 50% of women with GDM suffered from anxiety symptoms, which was notably higher than symptoms of either depression (13.4%) or stress (11.7%). We also found a significant association between a specific SNP of gene EPHX2 and depression, as well as SNPs of EPHX2, OXTR, NRG1 with stress symptoms.
Analyses of the socio-demographic background and clinical characteristics of the final 343 participants were stratified by psychological problem, as shown in Table 1. Among the various backgrounds and clinical characteristics evaluated, significant differences were observed only in terms of self-monitoring with a glucometer, ethnicity, religion, marital status, underlying with allergy and family history of depression and anxiety (p < 0.05) in between those with and without depression symptoms. After a Bonferroni adjustment in the context of family-wise error, these variables (ethnicity, religion, marital status, underlying with allergy and family history of depression and anxiety) still had an adjusted p-value < 0.05, except self-monitoring with glucometer (p-value = 0.08). Likewise, there were significant differences among ethnicity, religion, smoking habit, and underlying asthma among those with and without anxiety symptoms (p < 0.05). After a Bonferroni adjustment in the context of family-wise error for anxiety symptoms among GDM women, variables with adjusted p-value < 0.05 included ethnicity and smoking habit, while adjust p-value for religion was 0.066 and underlying asthma (p-value = 0.058). Further, significant differences were observed in terms of religion, past history of GDM and underlying allergy among those with and without stress symptoms (p < 0.05). After a Bonferroni adjustment in the context of family-wise error for stress symptoms among GDM women, the adjusted p-value for religion was 0.073, with past history of GDM (p-value = 0.048) and underlying allergy history (p-value < 0.0001). Bonferroni correction was used to reduce risk of multiple testing error. Even though some of the variables (self-monitoring with glucometer in depression, religion, and underlying asthma in anxiety symptoms) showed significant results with p-values < 0.05 after Bonferroni correction, we still proceeded with multiple logistic regression as we did not want to miss any variables known to be important as one of the predictors in our study.
The distribution of candidate gene genotypes satisfied the Hardy-Weinberg equilibrium (p > 0.05) ( Table A2). Analyses of the genotypes in SNPs of genes EPHX2, NPY5R, ANO2, NRG1, FKBP5, RORA, OXTR and BDNF among women with GDM were stratified by psychological symptoms and for candidate genotypes with p-value > 0.25 using univariate analysis is shown in Table 2. The analyses of the genotypes in SNPs of genes LHPP, SDK2, HTR2C, TEX51, PLEKHG1 and TPH2 genotype among women with GDM stratified by presence of psychological symptoms with p-value > 0.25 using univariate analysis are shown in (Table A3).
Notably, the proportion of the TT or TC genotypes was higher than that of the CC genotype in SNP of NRG1 (T > C in rs17466684) among GDM women with stress symptoms (13.2% versus 2.2%; p = 0.031). Similarly, the proportion of the TT genotype was higher compared with TG or GG genotypes in the SNP of FKBP5 (T > G in rs3800373) among GDM women with stress symptoms (57.5% versus 42.5%; p = 0.047) as shown in Table 2. On the other hand, there was no significant association between SNPS for candidate genes: [EPHX2, NPY5R, ANO2, FKBP5 (rs947008), RORA, OXTR and BDNF] and stress symptoms (p > 0.05). There was also no association between candidate genes and depression or anxiety symptoms (p > 0.05).
The association between specific SNPs' genotype of candidate genes and psychological symptoms of depression, anxiety and/or stress adjusted for socio-demographical and clinical moderators is shown in Table 3. GDM women with the AA genotype in specific SNP of EPHX2 (G > A in rs17466684) are 7.9 times more likely to suffer from depression symptoms compared to those who carry G allele in the SNP, when adjusted for ethnicity, religion, practice of home glucose monitoring, planned pregnancy, marital status, past obstetric history of abortion, underlying with allergy, a family history of depression and anxiety and GDM. Likewise, GDM women with the AA genotype in specific SNP of EPHX2 (G > A in rs17466684) is at 7.7 times odds more likely of getting stress symptoms compared to those who carry G allele in the SNP adjusted for ethnicity, religion, marital status, treatment regimens, past obstetric history of GDM, underlying with allergy and asthma and a family history of depression and anxiety. Not only that, we also found that GDM women with the either AA or AG genotypes in specific SNP of OXTR (A > G in rs53576) are 3.0 times more likely to suffer from stress symptoms compared to those who carry GG genotype in the SNP, as well as to those who carry either TT or TC genotypes in SNP of NRG1 (T > C in rs2919375), is at 9.9 times odds to experience stress symptoms compared to those who carry CC genotype in the SNP.
After a Bonferroni adjustment in the context of family-wise error for depression symptoms among GDM women, the adjusted p-value for self-monitoring with glucometer was 0.083, ethnicity (p-value = 0.003), religion (p-value = 0.004), marital status (p-value = 0.012), allergy history (p-value = 0.031) and family history of depression and/or anxiety (p-value = 0.002).
After a Bonferroni adjustment in the context of family-wise error for anxiety symptoms among GDM women, the adjusted p-value for ethnicity with was 0.004, religion (p-value = 0.066), smoking habit (p-value = 0.007), and asthma (p-value = 0.058).
After a Bonferroni adjustment in the context of family-wise error for stress symptoms among GDM women, the adjusted p-value for religion was 0.073, history of GDM (p-value = 0.048), and allergy (p-value < 0.0001).
After a Bonferroni adjustment in the context of family-wise error for stress symptoms among GDM women, the adjusted p-value for NRG1 (rs2919375) was 0.066.

Discussions
Over the years, an increasing number of polymorphisms in candidate genes related to the psychological problems have been discovered. Even so, most candidate gene association studies have been either overpowered or underpowered to detect the odds of genotypic heterogeneity for psychological symptoms. In this study, we performed simple logistic regression for every candidate gene, followed by multiple logistic regressions to elucidate the actual effect size of genotypes on the presence of depression, anxiety and/or stress symptoms. To our knowledge, this is the first study to examine the symptoms of depression, anxiety and/or stress among GDM women in Malaysia, and is also the first study to use the gene-environmental interaction hypothesis.
It is noteworthy that anxiety symptoms were the most commonly reported symptoms among the population of pregnant women with GDM (57.4% vs. 42.6%), whereas depressive symptoms (86.6% vs. 13.4%) and stress (88.3% vs. 11.7%) were much lower.
Based on logistic regression in this study, we reported that there is significant between SNP (rs17466684) of Epoxide Hydrolase 2 gene (EPHX2) with depression symptoms (AOR = 7.854, 95% CI = 1.330-46.360) and stress symptoms (AOR = 7.664, 95% CI = 1.579- 37.197). This is different finding compared with a study done in Japan where the carrier of AA genotype in SNP (rs17466684) of EPHX2 was found to be a risk variant of anxiety particularly panic disorder [42,68]. However, according to our genotypic analysis, this candidate gene was not associated with anxiety symptoms among Malaysian women. Polymorphism in EPHX2 contributes to the odds of suffering from depression, anxiety, and stress symptoms in the Japanese and Malaysian population. A possible explanation for these findings is that EPHX2 encodes for a key gat-keeper enzyme (soluble epoxide hydrolase) which functions in the catabolism of epoxy-fatty acids to their corresponding diols [69][70][71]. Soluble epoxide hydrolase is localized in neurons of central amygdala and this enzyme plays a vital role in neuronal firing [72] and it is hence believed that polymorphism in EPHX2 reduce the potency of anti-inflammatory activity of epoxy-fatty acids in brain [73], thus affecting the release of functional neurotransmitters that influence neuropsychiatric disorders [74].
Neuregulin 1 (NRG1) is an important gene signaling numerous neurodevelopment processes such as neurotransmitter receptor expression regulation and synaptic plasticity [75]. In our study, there was a significant association between SNP (rs2919375) of NRG1 and stress symptoms (AOR = 9.894, 95% CI = 1.159-84.427). To date, the C allele in SNP of NRG1 (T > C in rs2919375) is a minor allele and also a risk allele for major depression disorder among the Han Chinese population [34] was not found in our study. The reason for this difference is unknown. Apart from the population factor, the possible reason might be due to minor allele frequency in this study was 0.366, compared to 0.410 among Han Chinese population [34], therefore the effect of risk allele or genotype might be underestimated in our study. The minor allele frequency has influent on the power to detect genetic effects, SNPs with minor allele frequency ranges from 25% to 50% might give a false-positive rate ranging from 69.2% to 70.8% [76]. Therefore, the analysis for genes NRG1 (T > C in rs2919375) indicates that either TT or TC genotypes are determinants for stress symptoms, which might inflate false positive concerns.
Oxytocin receptor genes (OXTR) were found to have an association with neuropsychiatry disorders [27,28]; a possible explanation is that OXTR regulates the expression of OXTR p53, a potent transcription factor for the oxytocinergic pathway in neurons [77][78][79]. Emerging evidence also shows that OXTR rs53576 was associated with the structural coupling of the hypothalamus and amygdala, alteration to this structure is potentially to inflict neuropsychiatric disorders [80][81][82]. In our study, we found a positive association between OXTR rs53576 and stress symptoms among GDM women. Our finding contradicts with previous studies among the Japanese population [27] and Caucasian in Italy [28]. In a Japanese study, the G allele is the minor allele and presence of either AA or AG genotypes in SNP rs53576 were associated with panic disorders among the Japanese population [27]. In comparison to the finding done among Caucasians in Italy, a allele is a minor allele among Caucasians in Italy and the presence of either AA or AG genotypes is the protective factor for depression (OR = 0.67, 95% CI = 0.45-0.99) [28]. The findings of this study are of potential clinical and scientific importance as the identification of a significant association between particular candidate genes polymorphism with depression and stress among GDM women in Malaysia have certainly helped in the understanding of genetic aetiology among GDM women in local settings. Future studies should be conducted to validate the value of these candidate genes polymorphism in terms of genetic screening, so that the clinicians can send those GDM women at risk of having depression and stress for a genetic study.

Study Strength and Limitations
The present study contains multiple logistic regression analysis, adjusted for all socio-demographic backgrounds, and maternal and clinical profiles that potentially modulate the presentation of psychological symptoms. Therefore, the results shown on significant genotype related to depression and stress symptoms are clinically relevant despite this is an unmatched comparative case-control study, a sub-analysis from a cross-sectional study. The study demonstrates an association between candidate genes and the presence of depression, anxiety, or stress symptoms among GDM women. The interpretation of these association is limited by the screening nature of the psychometric tools used in measuring for these psychological symptoms, and not the diagnoses per se. Thus, the results should be interpreted cautiously. Future studies should be conducted with the inclusion of more SNP numbers per candidate gene to confirm the epigenetics-environmental moderator effects.

Conclusions
A significant association was observed between SNP (rs17466684) of EPHX2 and depression symptoms when adjusted for ethnicity, religion, the practice of home glucose monitoring, planned pregnancy, marital status, past obstetric history of abortion, underlying with allergy, a family history of depression, and anxiety with GDM. SNPs in EPHX2 (rs17466684), OXTR (rs53576) and NRG1 (rs2919375) are also associated with stress symptoms adjusted for ethnicity, religion, marital status, treatment regimens, past obstetric history of GDM, underlying with allergy and asthma and a family history of depression and anxiety.