Maternal metabolic factors and the association with gestational diabetes: A systematic review and meta‐analysis

Abstract Gestational diabetes (GDM) is associated with several adverse outcomes for the mother and child. Higher levels of individual lipids are associated with risk of GDM and metabolic syndrome (MetS), a clustering of risk factors also increases risk for GDM. Metabolic factors can be modified by diet and lifestyle. This review comprehensively evaluates the association between MetS and its components, measured in early pregnancy, and risk for GDM. Databases (Cumulative Index to Nursing and Allied Health Literature, PubMed, Embase, and Cochrane Library) were searched from inception to 5 May 2021. Eligible studies included ≥1 metabolic factor (waist circumference, blood pressure, fasting plasma glucose (FPG), triglycerides, and high‐density lipoprotein cholesterol), measured at <16 weeks' gestation. At least two authors independently screened potentially eligible studies. Heterogeneity was quantified using I 2. Data were pooled by random‐effects models and expressed as odds ratio and 95% confidence intervals (CIs). Of 7213 articles identified, 40 unique articles were included in meta‐analysis. In analyses adjusting for maternal age and body mass index, GDM was increased with increasing FPG (odds ratios [OR] 1.92; 95% CI 1.39–2.64, k = 7 studies) or having MetS (OR 2.52; 1.65, 3.84, k = 3). Women with overweight (OR 2.17; 95% CI 1.89, 2.50, k = 12) or obesity (OR 4.34; 95% CI 2.79–6.74, k = 9) also were at increased risk for GDM. Early pregnancy assessment of glucose or the MetS, offers a potential opportunity to detect and treat individual risk factors as an approach towards GDM prevention; weight loss for pregnant women with overweight or obesity is not recommended. Systematic review registration: PROSPERO CRD42020199225.


| INTRODUCTION
Gestational diabetes mellitus (GDM) is defined as the onset or first recognition of glucose intolerance during pregnancy, primarily in the second or third trimester. 1 GDM is one of the most common metabolic complications in pregnancy, affecting 5%-25% of all pregnant women worldwide, depending on screening approaches and diagnostic criteria. 2 GDM has adverse maternal health consequences, including an increased risk for hypertensive disorders of pregnancy, preterm delivery, medicalised delivery, 3,4 as well as an increased risk for developing type 2 DM and cardiovascular events in the first decade following pregnancy. 5,6 Offspring of mothers with GDM are at greater risk for large for gestational age, 7-9 respiratory distress syndrome 10 and neonatal hypoglycaemia, 11 and tend to develop type 2 diabetes at younger ages. 12 Recognised risk factors for GDM include maternal obesity, advanced maternal age, excess gestational weight gain, Asian and African ethnicity, and a history of diabetes. [13][14][15][16][17] Fasting or postprandial blood glucose may be assessed early, but whether it is a suitable screening test for GDM has not been clarified. [18][19][20] Metabolic syndrome is a clustering of cardiovascular risk factors that includes atherogenic dyslipidemia, raised blood pressure, insulin resistance, and obesity, 21 increasing the risk of cardiovascular disease (CVD) 22 and diabetes by up to 5-fold. 23 In two pregnancy cohorts, Grieger et al. 24 and Schneider et al. 25 showed that MetS, measured in early pregnancy, increased the risk for GDM by 2-4 fold, even after adjusting for body mass index (BMI).
Several studies have demonstrated that individual metabolic markers such as raised triglycerides (TG) or low density lipoprotein cholesterol, or reduced high density lipoprotein cholesterol (HDL-C) pose a significant risk for developing GDM. [26][27][28] The relationship between MetS or its individual components as a risk factor for GDM is plausible given their shared relationship to future risk of CVD. Importantly, metabolic factors can be modified by diet, lifestyle 29,30 and pharmacological agents. 31 Consideration of assessing metabolic markers in early antenatal care may provide information about potential future risk for GDM, allowing for early detection and management.
While some systematic reviews have been conducted on similar topics, they did not specifically examine MetS factors, but rather explored biomarkers associated with placental pathology, 32 central obesity, 33 and predictive 34 or diagnostic biomarkers for metabolic diseases. 35 To date, there has been no systematic review or metaanalysis comprehensively evaluating whether MetS or its components, measured in early pregnancy, associate with risk for GDM.
This would be important given the current controversies surrounding early screening of GDM using conventional risk factors, and that intervention studies aimed at preventing GDM, predominantly through targeting hyperglycemia, have not been consistently successful. 36 Measurement of MetS or its components may offer a new approach to identify potential risk for GDM, and which could be used as a complementary component to standard routine antenatal care.
The aim of this systematic review and meta-analysis is to comprehensively evaluate the association between MetS and its components, measured in early pregnancy, and risk for GDM.

| MATERIALS AND METHODS
We performed a systematic review and meta-analysis of epidemiological studies examining the association between components of MetS and risk of GDM. The review was performed according to the PRISMA 2020 Guidelines (Preferred Reporting Items For Systematic Reviews and Meta-analyses). 37 The study protocol was registered on PROSPERO (International Prospective Register of Systematic Reviews) under the identification code: CRD42020199225 and is available online (www.crd.york.ac.uk/prospero).

| Selection criteria and search strategy
Potential studies were identified through electronic database searches on Cumulative Index to Nursing and Allied Health Literature, PubMed, Embase, and the Cochrane database, and manual searches of potentially eligible references in review articles. The search strategy included a combination of subject indexing terms (i.e., MeSH) and free text search terms relating to early pregnancy, prognostic factors, and GDM, along with search filters recommended for prognostic modelling. 38 The search strategy was itera- pregnancy outcomes in women after GDM diagnosis; or studies assessing measurements that are not assessed routinely in antenatal care (e.g., using bioelectrical impedance assay).

| Core outcomes
The primary outcome was GDM using any diagnostic criteria, measured at 24-28 weeks' gestation.

| Study selection, data collection and risk of bias assessments
All citations were imported into an Endnote file, duplicates were removed, and the remaining articles export into the Rayyan software database for blind screening. 42  year, country; type of study; study population and sample size; study duration and month/year the study was carried out; inclusion criteria; exclusion criteria; GDM diagnosis and time point; exposures in the model; and statistical adjustments. When risk estimates from more than one multivariable analysis were reported, data were extracted from the analysis adjusting for the largest number of confounders. If risk estimates from other routine antenatal factors were reported, only risk estimates related to MetS were extracted. Only standard cut-off values for categorical data were used, for example, the World Health Organization (WHO) categories for BMI. Outcome data reported as odds ratios (OR)/relative risks (RR) and 95% confidence intervals (CIs) were the primary output of interest.
Risk of bias assessment at the study-level was performed independently by two researchers, using the quality in prognostic factor studies (QUIPS) Risk of Bias tool. 43 The same two independent authors who completed the data extraction completed the risk of bias for the same set of studies. Domains included study participation; study attrition; prognostic factor measurement; outcome measurement; adjustment for other key prognostic factors; and statistical analysis and reporting. Each domain was judged as low, moderate or high risk, with more weighting given to the domains of 'Adjustment for other prognostic factors' and 'Statistical analysis and reporting', due to the observational nature of the included studies. Pilot testing was performed using three test articles to ensure consistency between the authors prior to formally commencing risk of bias assessments. For each study, the item scores were collated and an overall risk of bias (low, moderate, and high) was determined.

| Data analysis
All analyses were conducted using Review Manager (V5.4.1). For the primary meta-analyses, studies reporting OR or RR with 95% CIs were analysed as these data are best suited to address questions on prognosis. Data were pooled using the restricted maximum likelihood random-effects models to account for heterogeneity among the studies and outcome measures. 44 Unadjusted analyses were firstly reported followed by adjusted analyses with a core set of prognostic covariates (maternal age, maternal BMI, family history of diabetes, ethnicity). As many of the included studies did not adjust for all four covariates, we opted for at least one core covariate in each model.
Heterogeneity was quantified using the I 2 statistic. Significance for heterogeneity was set at p < 0.10, with an I 2 > 50% considered to be of relatively high heterogeneity. 45 Sources of heterogeneity were explored where outlier study/s were eliminated from the metaanalysis in a series of sensitivity analyses and the effect size was recalculated to determine the influence of those studies. 46 We considered outlier studies that had a different direction of effect, a high effect size, or studies judged to be at high risk of bias. 45 If ≥10 studies were available, we assessed publication bias by visual inspection of funnel plots. 47 Data reported as mean and SD/SE or 95% CI, or as median and interquartile range were included in the narrative synthesis.

| Patient and public involvement
This study was a systematic review and therefore did not include patients as study participants.

| Study selection and characteristics
The systematic search identified 7213 articles of which 106 were duplicates, 7029 were ineligible, leaving 78 articles in the systematic review and 40 articles in the meta-analysis ( Figure 1). Characteristics of the included studies are reported in Table 1. The majority of studies were conducted in China (k = 18 studies), USA (k = 9 studies), UK (k = 8 studies), and Australia (k = 8). Study population sizes ranged from 107 48 to 132,899 participants. 49 Of the 78 studies included, the majority (41.25%) reported on two or more MetS factors. Body mass index was the most common independently assessed risk factor (64.1%), while 10 studies (12.8%) reported on HbA1c and HDL-C.
In meta-analysis, the minimum set of confounding variables included were: maternal age for overweight and obesity analyses; maternal age and BMI for fasting plasma glucose (FPG), TG, HbA1c, HDL-C, and MetS analyses; and maternal age, BMI, ethnicity, and family history of diabetes, for systolic blood pressure analyses. HABIBI ET AL.  52 and around a quarter of included studies did not report on the diagnosis criteria, or used criteria from within their own institution.

| Narrative review and meta-analysis
Supporting Information (Figures S1-S7) illustrates the narrative results reporting on mean differences in each metabolic factor between women with and without GDM. Figures 2-10

| Waist circumference (WC)
Six cohort studies assessed WC with sample sizes ranging from 247 to 19,186. [53][54][55][56][57][58] Overall, women who developed GDM had a larger WC measured in early pregnancy, with a mean difference of 6.20 cm compared to women without GDM (p < 0.0001; Supplementary Figure S1). [56][57][58] Studies were not pooled in the meta-analysis as they did not report on OR or RR.

| Body mass index (BMI)
Body mass index was derived from recorded medical history data, self-report, or from a measurement at the first antenatal visit.

| Metabolic syndrome
Three prospective cohort studies with a sample size ranging from 498 to 3126 were pooled in the meta-analysis. 24,25,53 Metabolic syndrome in early pregnancy was associated with a higher odds of GDM in unadjusted (OR 2.58, 95% CI 1.97-3.37, I 2 = 0%, P het = 0.51, k = 2; Figure 10A) and analyses adjusted for age, BMI and up to another 7 confounders (aOR 2.52, 95% CI 1.65-3.84, I 2 = 67%, P het = 0.05, k = 3; Figure 10B). There were insufficient numbers of studies to perform sub-group analyses according to GDM criteria within any metabolic factor (Supporting information Figure S13).

| Principal findings
The purpose of this systematic review and meta-analysis was to examine the association between maternal MetS and its components with GDM, an independent risk factor for future type 2 diabetes and CVD. 6 Women with overweight or obesity had up to a 4-fold increased risk for GDM, and increasing FPG or having the MetS as a clustering of factors posed up to a 2.5 times higher likelihood for developing GDM. Findings were consistent in adjusted analyses and persisted in sensitivity analyses to reduce heterogeneity.

| Strengths and limitations
Strengths of this review include the extensive and thorough litera- review enroled younger and older pregnant women or women across the BMI spectrum, the studies did not specifically recruit women who were either younger or older, or with low or high BMI, thus subgroups could not be created. While many studies also included women across different ethnic groups, studies did not always report on the proportion of different ethnicities included, and where they did, they were not sufficiently homogenous across studies to make comparisons. Thus, the effect of age, BMI, or ethnicity, on the strength of the association with GDM could not be determined.
Nevertheless, we did find that even after adjusting for age and BMI, the effect of the different MetS risk factors on GDM was similar to unadjusted analyses.

| Comparison with other studies
Women with overweight or obesity had a 2-4-fold greater likelihood for development of GDM. A recent meta-analysis using 33 observational studies demonstrated up to a 3.2-fold increased odds for GDM with increasing pre-pregnancy BMI category, and a 19% increased risk of GDM per unit of increase in pre-pregnancy BMI. 122 Our results for early pregnancy overweight or obesity are in line with findings on prepregnancy BMI, albeit, a smaller increase in odds for GDM (8%) per unit increase. Importantly, for our review, we deliberately focussed on early pregnancy BMI, because losing weight before pregnancy does not appear to alter risk for GDM compared to women who are weight stable. 123,124 Pregnant women with overweight or obesity have higher FPG, insulin, and TG, compared to normal weight pregnant women. 125 However, several of the individual studies in this review demonstrated that metabolic risk factors increased risk for GDM, independent of BMI. Since weight loss is not recommended during pregnancy, 126 and targeting pre-conception women with overweight or obesity is likely to be challenging, our findings reinforce the need to identify other important modifiable risk factors for GDM.
An approximate 2-fold risk for GDM was demonstrated with increasing level of FPG, which persisted in adjusted and sensitivity analyses. Across gestation, glucose levels reduce due to the maternal adaptations of pregnancy and because of the increased glucose utilisation by the foetal-placental unit. 127 There is also an increase in insulin resistance. 128 These maternal adaptations potentially limit the use of fasting glucose in early pregnancy for early diagnosis of GDM.
However, there is data to show that maternal hyperglycaemia before the routine diagnosis of GDM increases the rate of foetal growth 129 and infant adiposity. 130,131 Thus whether to diagnose GDM in early pregnancy is an ongoing point of contention. Our results provide some evidence that testing for early FPG may be useful to intervene in women with high glucose to ameliorate the adverse short and long-term effects of prolonged intrauterine exposure to hyperglycaemia, but the strength of this evidence is insufficient to alter clinical practice or guide timing of early testing. For measurement of HbA1c, a longer-term measure of glucose control, three studies were included in the meta-analysis of which one study had a very small OR with large CIs. Thus, whether HbA1c is useful for early screening of future GDM cannot be established from this analysis and requires further investigation.
Increasing fasting TG was associated with a 1.2-fold increased likelihood for GDM, however from two studies, triglyceride levels >1.7 mmol/L was associated with a 2-fold increased risk. Increased TG are associated with insulin resistance, 132 which not only drives the process for MetS, 133 but is also an important factor underlying the development of type 2 diabetes and CVD. [134][135][136] In a recent study of 500 adults in China, TG positively correlated with insulin resistance in participants with normal glucose tolerance, with a negative, independent correlation with beta cell function in individuals with dyslipidaemia. 132 Indeed, a clustering of abnormalities (i.e. metabolic syndrome) which is related to insulin resistance and/or hyperinsulinemia coupled with dyslipidaemia, may be unfavourable to GDM and overall cardiometabolic health. Our systematic review identified only three studies investigating MetS, and pooling of these studies showed a 2.5-fold higher likelihood for developing GDM. This odds ratio is higher than the individual risk associated with elevated FPG or TG, but lower to that of obesity. While these observations are important and highlight a potentially important relationship between MetS in early pregnancy and risk for GDM, the studies available were few, warranting further investigation.

| Recommendations or clinical implications
Overall, our review cannot provide explicit recommendations or implications for practice for women who may benefit early screening and assessment of MetS factors to identify potential risk for GDM.
While the effect estimates remained largely unchanged in sensitivity analyses, the overall high heterogeneity could not be sufficiently tested given the available data from the studies. Moreover, subgroup analyses in populations at higher risk of GDM, such as older maternal age, higher BMI, and women of minority ethnicities, could not be undertaken. The indication that MetS as a cluster of risk factors demonstrated a doubled risk for GDM, warrants further exploration, both in women with or without obesity.

| CONCLUSION
The meta-analysis provides some evidence that early pregnancy assessment of FPG or the MetS, as a clustering of factors, offers a potential opportunity to detect and treat individual risk factors as an important approach towards GDM prevention. Women with HABIBI ET AL.
-35 of 41 overweight or obesity in pregnancy are also at risk for GDM, however weight loss in pregnancy is not recommended. Given the overall number and quality of studies included, there is a need for further, larger, and higher quality studies to corroborate these results.