Prediction of pre-eclampsia in diabetic pregnant women

Gestational diabetes mellitus (GDM) is a glucose tolerance disorder that occurs or is diagnosed for the first time during pregnancy. It has been reported that GDM affects 1-14 per cent of all pregnancies, and its incidence has been steadily rising1,2. The patients with GDM have increased risk of developing type 2 diabetes mellitus (T2DM) in the years following the pregnancy. Pre-eclampsia is a multisystem complication and contributes significantly to maternal-foetal morbidity and mortality3. GDM is a risk factor for the development of T2 DM, and hypertension during pregnancy also predisposes to an increased risk for developing subsequent systemic hypertension4. Pre-eclampsia has been linked to glucose intolerance, and GDM itself is an independent risk factor for pre-eclampsia5. There are many common risk factors between the two conditions such as increased maternal age, nulliparity, multiple gestation and an increased pre-pregnancy body mass index5. Maternal serum biomarkers of these two conditions have been reported earlier. Sex hormone-binding globulin (SHBG) is a marker of insulin resistance6. Low levels of SHBG are associated with elevated insulin levels. The circulating levels of insulin-like growth factor-I (IGF-I) are found to be lower in pre-eclampsia7. Vitamin D deficiency is common in Indian population and has also been found to be associated with pre-eclampsia8.

The present study aimed to evaluate the clinical risk factors for the development of pre-eclampsia in cases of GDM/DM and to study levels and gene polymorphism of SHBG, IGF-I and 25 hydroxy vitamin D [25(OH)D] in the above cases for developing a predictive model for pre-eclampsia in GDM/DM cases.

Material & Methods

This prospective comparative study was conducted at the Antenatal Clinic of Maulana Azad Medical College and Lok Nayak Hospital, New Delhi, India, from February 2015 to January 2018. The study was approved by the Institutional Ethical Committee of Lok Nayak Hospital, a tertiary care hospital and a referral centre.

Inclusion criteria: The study group included all consecutively registered pregnant women between 12 and 20 wk of gestation, aged 18-40 yr and diagnosed with (i) GDM before 20 wk of gestation, and (ii) DM before pregnancy. The control group included healthy asymptomatic age-, parity- and period of gestation-matched pregnant women with no medical/metabolic/surgical diseases and had 2 h plasma glucose levels <140 mg/dl after 75 g oral glucose load at recruitment and at 24-28 wk gestation.

Exclusion criteria: Women with a history of hypertension, chronic renal disease, cardiovascular disease, thyroid disorders, tuberculosis, pre-existing calcium or parathyroid conditions, sarcoidosis and osteomalacia, bone disorders and urolithiasis and systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg at first visit and pregnant women with congenital malformation in foetus were excluded.

Diagnosis of pre-eclampsia: Pre-eclampsia is diagnosed when the systolic blood pressure is ≥140 mmHg or diastolic ≥90 mmHg after 20 wk of gestation at two occasions at least four hours apart in a woman with previously normal blood pressure along with the development of proteinuria (defined as >300 mg/24 h or ≥1+ on a clean catch dipstick testing in the absence of urinary infection)3.

Diagnosis of gestational diabetes mellitus (GDM): Gestational diabetes mellitus (GDM) is defined as hyperglycaemia that is first diagnosed during pregnancy after 75 g oral glucose load. Diabetes in Pregnancy Study Group, India (DIPSI) recommends 2 h venous blood sample for estimating plasma glucose during pregnancy. A value ≥140 mg/dl is taken as diagnostic for GDM9,10. This has also been included in the guidelines issued by the Ministry of Health and Family Welfare, Government of India11.

Diagnosis of diabetes mellitus (DM) in pre-pregnancy state: After obtaining a written informed consent, all the study participants were evaluated based on pre-designed and pre-tested standardized pro forma which included history for risk factors, dietary information, drug intake or supplementation information, obstetric history, clinical examination, routine laboratory investigations and biochemical tests during pregnancy. The socio-economic status (SES) of the subjects was assessed as per the Modified Kuppuswamy Scale12, which takes into consideration three variables and scores were given for the three variables namely, education (1-7), occupation (1-10) and income (1-12) and then added to get a composite score (maximum 29/29).

Sample size calculation: Grewal et al13 and Bodmer-Roy et al14 had shown that the prevalence of pre-eclampsia among GDM and healthy women was 15 and seven per cent, respectively. Taking it into consideration and 80 per cent power at five per cent level of significance with the case-to-control ratio as 1:1, the estimated sample size of 264 per group was calculated for a two sample proportion test using the following formula:

Further, considering about 15 per cent dropout during the study process, the approximate sample size of 300 per group was taken.

Sample collection: Venous blood (5 ml) was drawn at the time of recruitment and divided into two vacutainers. Serum was separated from 3 ml blood sample and stored at −70°C for the estimation of SHBG, IGF-I and 25(OH)D levels. Remaining 2 ml blood was collected in heparinized vial and stored at 2°C-8°C for genomic DNA extraction and polymorphism analysis. Bilateral uterine artery Doppler velocimetry was performed at 20-24 wk of gestation. These women were followed up and managed as per hospital protocol.

Estimation of sex hormone-binding globulin (SHBG), insulin-like growth factor (IGF)-I and 25 hydroxy vitamin D [25(OH)D] by ELISA: The levels of SHBG (DRG International, Inc., USA), IGF-I (DRG International) and 25(OH)D (Calbiotech, Spring Valley, CA 91978, USA) were estimated by using commercially available ELISA kits according to the manufacturer’s protocol. The sensitivity of SHBG, IGF-I and 25(OH)D kits was 0.1 nmol/l, 9.75 ng/ml and 0.67 ng/ml, respectively.

DNA extraction and polymorphism: DNA was extracted from whole blood samples by commercially available DNA extraction kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s protocol. The quality of extracted DNA was checked spectrophotometrically (Biophotometer, Eppendorf, Germany) and also by running on three per cent agarose gel. A ratio (OD260/OD280) of 1.8-2.0 denotes that the absorption in the ultraviolet (UV) range is due to nucleic acids. For gene polymorphism, single-nucleotide polymorphisms (SNPs) were determined using sequence-specific oligonucleotide primers15-17.

All the DNA samples were individually extracted and to check for genotyping error rate, approximately 10 per cent of the samples were randomly selected and subjected to genotyping in duplicate.

Polymorphism of vitamin D receptor (VDR) gene: The vitamin D receptor (VDR) genotype at BsmI, FokI and TaqI SNP sites was determined using PCR-restriction fragment length polymorphism (RFLP) analysis as per standard protocol.

Polymorphism of sex hormone-binding globulin (SHBG) gene: Amplification of the TAAAA repeat region within the Alu sequence in the SHBG promoter and CAG region within the AR was done with a forward primer (5′-GCTTGAACTCGAGAGGCAG-3′) and a reverse primer (5′-CAGGGCCTAAACAGTCTAGCAGT-3′) for the SHBG(TAAAA)n polymorphism and a forward primer (5′-TCCAGAATCTGTTCCAGAGCGTGC-3′) and a reverse primer (5′-GCTGTGAAGGTTGCTG TTCCTCAT-3′) for the AR(CAG)n polymorphism. DNA fragments amplified with PCR were of 159-184 bp in size for SHBG(TAAAA)n polymorphism and 240-318 bp in size for AR(CAG)n polymorphism.

Polymorphism of insulin-like growth factor-I (IGF-I): The IGF-I CA repeat was amplified using PCR primers forward 5′-GCTAGCCAGCTGGTGTTATT-3′ and reverse 5′-ACCACTCTGGGAGAAGGGTA-3′ as per standard protocol.

Statistical analysis: Data analyses were carried out using statistical software STATA version 12.0 (StataCorp LLC, Texas, USA) . Continuous variables were tested for normality assumptions using Kolmogorov–Smirnov test. Descriptive statistics such as mean, standard deviation and range values were calculated for normally distributed data. Comparison of two-group means was tested using Student’s t independent test. More than two means were tested using one-way ANOVA followed by Bonferroni pairwise comparison test. For non-normal data, median and interquartile range values were calculated. Comparison of two median values was tested using Mann-Whitney U test/Wilcoxon rank-sum test and more than two medians were compared using Kruskal-Wallis test. Categorical data were expressed as frequency and per cent values and Chi-square/Fisher’s exact test was used as appropriate. For assessing significant variables for pre-eclampsia, univariate followed by multivariable logistic regression analyses were carried out. While carrying out univariate logistic regression analysis, the variables that were found to be significant at P<0.10 were taken for multivariable logistic regression analysis. Adjusted odds ratios (aORs) with 95 per cent confidence limits were calculated. Based on logistic regression coefficients, risk probability values were calculated. To decide cut-off probability of optimum sensitivity and specificity, receiver operating characteristic (ROC) analysis was carried. A two tailed P<0.05 was considered significant.

Results

A total number of 2050 pregnant women between 12 and 20 wk of gestation were evaluated. Of these, 30 women were lost to follow up and therefore excluded from the study. A total of 316 (15.41%) pregnant women (296 pregnant women had GDM and 20 pregnant women had type 1 DM before pregnancy) were included in the study group (Fig.1). The women with GDM were found to have venous blood glucose values more than 140 mg/dl at two hours after oral intake of 75 g glucose. Out of the remaining 1704 normal healthy women, a cohort of 723 women was obtained in such a way that age, parity and gestation at recruitment matched with that of the GDM group. Then, 316 women were selected as controls by adopting simple random numbers generation technique using Epi Info software developed by Centers for Disease Control and Prevention, Atlanta, USA.

Close

Fig. 1

Consort flow diagram of study population. DM, diabetes mellitus; GDM, gestational diabetes mellitus

Export to PPT

The mean values of age of participants, gestational age at enrolment and distribution by parity status did not differ significantly between the control and GDM patients. The basic characteristics and biochemical profile of study population are shown in Table I. DNA could be extracted only from 600 blood samples. The genotypic frequencies of SHBG, IGF-I and VDR genes are presented in Table II.

During antenatal follow up, 96 women (30.38%) in the study group and 44 (13.92%) women in the control group developed pre-eclampsia (P<0.001). The basic characteristics of women who developed pre-eclampsia and data of serum titres and genotypic frequencies of SHBG, IGF-I and VDR genes are shown in Tables III and IV, respectively.

Univariate and multivariate logistic regression analysis for pre-eclampsia among controls and women with GDM are shown in Tables V, VIa and VIb, respectively. To predict risk of development of pre-eclampsia in women in the study group, the fitted risk model by backward elimination procedure used to calculate a risk score for each patient. The estimated regression coefficients for an adequate model are given in Table VIb.

After calculation, the final model for pre-eclampsia prediction among GDM patients was given as:

Risk score = −1.05 + [−0.68×1 (education >10 yr)] + [0.38×2 (upper lower SES) − 0.40×3 (middle SES) + 1.50×4 (upper middle SES) + 1.81×5 (upper SES) + 0.85×1 (history of diabetics) + 1.52×1 (history of pre-eclampsia)].

Risk probability = Exponation (risk Score) ÷ {1+exponation (risk score)}

ROC analysis was carried out with the available study variables by considering the probability risk score, which was based on multivariable logistic regression amongst GDM women. ROC curve for pre-eclampsia showed that area under the curve (AUC) is 0.68 [95% confidence interval (CI): 0.63-0.73]; P<0.001. The probability of detecting pre-eclampsia is 0.68 at the same time the probability of detecting normal cases (Fig. 2). In the present analysis, for a risk probability of more than 0.3, the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), diagnostic accuracy, likelihood ratio (LR)+ve and LR−ve were 59, 72, 48.3, 80.3, 68.4, 2.14 and 0.56 per cent, respectively.

Close

Fig. 2

ROC for predicting pre-eclampsia by risk score in women with DM/GDM. DM, diabetes mellitus; GDM, gestational DM; ROC, receiver operating characteristic.

Export to PPT

The pregnancy outcome of the two groups is shown in Table VII.

Discussion

This study was conducted at a tertiary care centre catering to a large population of lower and middle socio-economic status of population and provides an insight about the risk factors for the occurrence of pre-eclampsia among patients with GDM or DM in pregnancy in the Indian set up. Out of 2050, 316 (15.41%) pregnant women had GDM/DM. During antenatal follow up, 96 (30.38%) women in the study group and 44 (13.92%) women in the control group developed pre-eclampsia (OR 2.83, 95% CI: 1.90-4.2; P<0.001). It was observed that women from upper-middle and upper class SES were likely to be in the higher risk of developing pre-eclampsia (4.50 and 6.10 times, respectively) compared to lower SES category (P=0.001 and 0.006). The chance of getting pre-eclampsia among those who had a history of DM before pregnancy and pre-eclampsia in previous pregnancy was about 2.34 and 4.56 times higher as compared to those who had no such history of DM and pre-eclampsia, respectively (P=0.027 and 0.030) (Table VIb).

A systematic review and meta-analysis18 on the estimates of GDM prevalence across India evaluated 12 studies using DIPSI criteria and reported a prevalence of 7.37 per cent (95% CI: 5.2-10.16) (range: 3.01-16.26%). In a 16 yr follow up study of the participants of Nurses’ Health Study II, 5.6 per cent (1414 women) developed GDM with a 26 per cent increased risk of hypertension19. After controlling the variables such as age, nationality, job status, smoking, parity, multi-foetal pregnancy, pre-pregnancy weight status and gestational weight gain, in a retrospective study of 647,392 pregnancies in the German Perinatal Quality Registry, the odds of pre-eclampsia were found to be increased among women with GDM [adjusted odds ratio (aOR): 1.29, 95% CI: 1.19-1.41]5. Similarly, results from the birth registry study in Canada reported that GDM is an independent risk factor for pre-eclampsia20.

Ross et al21 in a recent study aimed to assess whether SES, race/ethnicity and pre-natal cardio-metabolic disease (pre-eclampsia, gestational hypertension, gestational diabetes) interact in the prediction of postpartum cardio-metabolic risk and concluded that African-American women and especially those who experienced pregnancy complications were at a higher risk. On further analysis, higher household income did not appear to confer protection against worse postpartum cardio-metabolic risk for this group21. It appears that there is a complex interplay between SES and race/ethnicity with respect to understanding health disparities21. A sample of 718,604 Black and White women was evaluated from a population-based California cohort of singleton birth. Black women were found to have an increased risk of pre-eclampsia as compared to Caucasian women. It is possible that higher SES attenuated the risk for pre-eclampsia among White women, but not for Black women. Ross et al22 concluded that higher SES has less impact on cardio-metabolic disease among Black women compared to White women, an effect called ‘diminishing returns’.

Both endothelial dysfunction and insulin resistance are common aetiopathogenetic factors for DM, pre-eclampsia and cardiovascular disease23,24. Obesity, advanced maternal age, non-White race, chronic hypertension and DM/GDM have been shown to be associated with insulin resistance and are reported as the risk factors of pre-eclampsia23,24. Furthermore, metformin has been reported to be efficacious in the treatment of GDM and possibly reduces the risk of pre-eclampsia by reducing soluble fms-like tyrosine kinase-1 levels25,26.

In the present study, the serum titres of SHBG and 25(OH)D were not significantly different in the control group and women with GDM who developed pre-eclampsia, except those of IGF-1. Although SHBG is a sensitive biomarker of insulin resistance and metabolic syndrome, the contribution of SHBG and 25(OH)D levels in pre-eclampsia and GDM has not been well established. SHBG level between 13 and 16 wk is found to be valuable for screening women for GDM risk27,28, but with poor sensitivity in low-risk pregnancies29. A significant reverse correlation between serum SHBG and pre-eclampsia (aOR=0.99; 95% CI: 0.98-1.00; P=0.04) was reported in Iranian women independent of insulin resistance30, whereas comparable SHBG levels were observed between normotensive women and patients with pre-eclampsia with significantly higher insulin resistance in pre-eclampsia31. There is reportedly a considerable heterogeneity in polymorphisms of SHBG, indicating the multiplicity of factors influencing SHBG variation32. Serum concentrations of IGF-I are reported to be abnormal long before women manifested clinical evidence of pre-eclampsia and might be related to abnormalities in trophoblastic invasion. A nested case–control study suggested a potential role of IGFs and their binding proteins in the prediction of pregnancies complicated by pre-eclampsia33. Maternal IGF-I concentrations in GDM pregnancies were found to be significantly higher than in controls, suggesting that the IGF axis might be playing a role in the development of this condition34. Maternal vitamin D status is influenced by many factors. Women with circulating 25(OH)D level less than 50 nmol/l in pregnancy had an increased risk of pre-eclampsia and GDM8. Vitamin D insufficiency of <30 ng/ml was associated with higher blood pressure in a cohort of pregnant women with GDM, and serum 25(OH)D was an independent marker of blood pressure in White but not in dark skinned women35. There is wide variation and complex relationship of these potential biomarkers of pre-eclampsia in the presence of insulin resistance and glucose intolerance. The present study could not find any significant association of serum biomarkers and their genotypic polymorphisms in the prediction of pre-eclampsia in women with GDM.

In a study to evaluate the risk factors of pre-eclampsia in women with GDM, first-trimester BMI≥27 kg/m2, GDM diagnosed within 20 wk of gestation and poor glycaemic control were associated with pre-eclampsia on multivariable analysis36. When these three factors were put into a risk-scoring model (ranged from 0 to 3 points), the sensitivity, specificity and AUC for pre-eclampsia were high at 76.9 per cent and 0.849, respectively (considering the optimal cut-off score of ≥2)36. In the present study, ROC curve for pre-eclampsia showed that AUC was 0.68. Since the model was based on the available variables in the study, it is possible due to the lack of other potential risk variables; the AUC of 0.68 may lead to moderate PPV and NPV.

Overall, diabetic status during pregnancy predisposes for the development of pre-eclampsia. The risk may increase in women belonging to higher SES with history of pre-eclampsia in previous pregnancy or GDM. The serum biomarkers [SHBG, IGF-I and 25(OH)D] and respective gene polymorphisms were not found to be useful in prediction of pre-eclampsia in women with GDM in this study. Healthcare providers are thus challenged with controlling maternal glucose levels and blood pressure in ways that optimize both maternal and foetal outcome.