Screening and diagnosis of gestational diabetes mellitus in an African population

Abstract

Background: Gestational diabetes mellitus (GDM) is defined as hyperglycaemia that is discovered for the first time during pregnancy and is not overt. The prevalence of GDM in South Africa varies between 9.1% and 25%. Mothers with GDM are at increased risk of preeclampsia, caesarean delivery, and preterm birth. Similarly, babies born to GDM mothers are at increased risk of macrosomia, hypoglycaemia, respiratory distress, hypocalcaemia, hypomagnesaemia and Type 2 diabetes(T2D) later in life. Therefore, efficient screening and diagnosis remain the cornerstone for reducing GDM-related maternal and neonatal morbidity and mortality. The oral glucose tolerance test (OGTT) is the current gold standard for screening for GDM. However, in South Africa, the test is not universally accessible as only pregnant women with GDM-related risk factors are tested. This approach is likely to miss a significant number of women with GDM. Aim: The overall aim of this research is to improve the current screening and diagnostic methods for GDM. Consequently, our first objective was to evaluate the performance of several point-of-care (POC) glucometers against the standard laboratory-based glucose measurements during an OGTT. The second objective was to evaluate the use of glycaemic markers namely, fasting plasma glucose (FPG), glycated albumin (GA) and glycated haemoglobin (HbA1c), for the diagnosis of GDM. The third objective was to determine the associations of various candidate biomarkers with glycaemia in both early and late pregnancy in Black African pregnant women. Methods: This is a prospective cross-sectional study that was carried out at five different health facilities in Johannesburg, South Africa. Pregnant Black African women under 21 weeks of gestation (early pregnancy) and at 24-32 weeks of gestation (late pregnancy) were recruited and stratified into three groups. The women in late pregnancy (groups 1 and 2) were tested for GDM using the 75 g OGTT and classified into GDM and non-GDM following the International Association of Diabetes and Pregnancy Study Group (IADPSG) guidelines. Seven glucometers were compared to laboratory- based glucose measurements for GDM diagnosis using blood samples from the OGTT (Group 1). In the second group of late-pregnancy women (Group 2), candidate biomarkers (TNFα, IL-6, free fatty acids (FFAs), branched-chain amino acids and acylcarnitines) and known markers of glycaemia (FPG, HbA1c and GA) were assessed for their ability to diagnose GDM using the OGTT as the reference. Fasting and random blood samples were taken in women in early pregnancy (group 3), and candidate biomarkers and blood glucose were measured. Univariate correlations and multivariate regression analysis were performed to determine the relationship between glycaemia and ii candidate biomarkers in early and late pregnancy, and their discriminatory power was assessed by receiver-operator characteristic (ROC) curve analysis. Results: A total of 1716 women were recruited: 1076 from pregnancy (group 1), 477 also from late pregnancy (group 2), and 163 from early pregnancy (group 3). For the glucometer study (Group 1), 83 (7.7%) were classified as GDM. The sensitivity of the glucometers for GDM diagnosis ranged from 17.6% to 87.18% and their specificity from 62.7% to 99.8%. The area under the ROC curve (AUC) was between 0.59 and 0.79. However, the laboratory- based FPG showed better performance than the glucometers in terms of sensitivity, specificity, and AUC, with levels of 94.0%, 100%, and 0.98, respectively. Building on the observations from the Group 1 study, which showed potential for diagnosing GDM using a single test i.e. FPG, we subsequently compared the efficacy of FPG, glycated haemoglobin (HbA1c), and glycated albumin (GA), for the diagnosis of GDM in the Group 2 women. Overall, FPG at a 5.1mmol/L cut point performed better than HbA1c and GA, with sensitivity and specificity of 87% and 100% respectively. At the 5.5.% cut-off point, HbA1c had sensitivity and specificity of 54% and 80% respectively and GA, had a sensitivity and specificity of 54% and 83%, respectively at a cut point of 37.6%. Based on the selective screening criteria currently used, all the women in group 2 had one or more risk factors for GDM and as a result, would have qualified for OGTT. If FPG was then used as a second screening test, it would have averted the need for OGTTs in 77% of these women. Thereafter, we sought to identify novel biomarkers for GDM diagnosis. Metabolites and proteins known to be linked to GDM were measured in 163 women below 21 weeks of pregnancy (Group 3) and 477 women between 24-32 weeks of pregnancy (Group 2). Free fatty acids (FFAs), and isoleucine were positively associated with GDM whilst acylcarnitine C2 demonstrated an inverse association. The sensitivity and specificity of these candidate biomarkers for the diagnosis of gestational diabetes mellitus (GDM) were inadequate, and they did not exhibit a significant association with fasting or random plasma glucose levels during early pregnancy, with the exception of valine, which demonstrated a significant correlation with random plasma glucose. (p=0.0029) Conclusions: The current study showed that the performance of laboratory-based FPG measurements for GDM diagnosis is better than HbA1c, GA, and glucometers. Furthermore, we observed that a two-tiered approach utilising FPG would reduce the number of OGTTs required for iii women diagnosed with GDM. In addition, valine has the potential to be used as an early marker of GDM, however, this would need to be validated in longitudinal studies.