INTRODUCTION Ethnicity influences the prevalence of GDM, being higher in non-European populations.4,5<\/sup> In addition, the progression to permanent type 2 diabetes may occur more rapidly in non-European subjects.2,6<\/sup> Ethnic variations in insulin sensitivity, blood pressure and aspects of the Metabolic Syndrome have also been described, independently of GDM.7,8 We previously found, using the insulin-modified frequently-sampled intravenous glucose tolerance test (FSIVGTT), a range of metabolic abnormalities, including insulin resistance, reduced \u03b2-cell function and dyslipidaemia in women of exclusively European origin with previous gestational diabetes.12<\/sup> Given the ethnic variations in insulin sensitivity and metabolic aspects, we wanted to assess whether similar metabolic abnormalities persist following gestational diabetes in other ethnic groups. We investigated insulin action and secretion by using the FSIVGTT and HOMA model following gestational diabetes in a group of South Asian and Afro-Caribbean women with normal fasting glucose and present these findings together with our previously studied European group.<\/p>\n SUBJECTS The two groups were matched for ethnicity, age, parity and time since delivery (Table 1<\/a>). Parents\u2019 and grand-parents\u2019 place of birth, language and religion were documented to ensure that recruitment in both groups was restricted to subjects with exclusively European, South Asian and Afro-Caribbean ancestry for at least 3 generations. No woman in either group was pregnant or breast-feeding at the time of the study. All subjects were assessed in the follicular part of their cycle, or in the \u201cpill-free\u201d interval for those receiving oral contraceptives. METHODS<\/strong><\/span> An insulin modified FSIVGTT was used to estimate insulin sensitivity (Si<\/sub>) and glucose effectiveness (Sg<\/sub>) from glucose and insulin data by the minimal model method of Bergman.16,17<\/sup> After an overnight fast, an intravenous line was inserted in each antecubital fossa. A glucose bolus (0.3 g\/kg of body weight, 50% dextrose solution) was administered intravenously over one minute and 25 minutes later insulin was injected at a dose of 0.02 U\/kg of body weight. Serum intact insulin and plasma glucose were measured in samples taken 15 minutes before and immediately before glucose injection. Blood samples for insulin and plasma glucose measurements were taken at: 2, 3, 4, 5, 6, 8, 10, 12, 14, 16, 19, 22, 24, 26, 27, 28, 29, 30, 32, 35, 38, 40, 50, 60, 70, 80, 90, 100, 120, 140, 160 and 180 min after the intravenous glucose administration, from the other arm. The minimal model parameters, Si <\/sub>and Sg<\/sub>, were derived using the IS-CIBA computer package (Gunther Mehring, Novartis Pharmaceuticals, Basel, Switzerland). Glucose disappearance rate (Kg<\/sub>) was calculated as the slope of glucose decline from 10 to 24 min of the FSIVGTT and acute insulin response to glucose (AIRg<\/sub>) was assessed as the incremental area under the curve (AUC) (calculated by the trapezoid rule) from 2 to 10 min of the FSIVGTT.18<\/sup> Of the 139 FSIVGTTs, data from 3 subjects (2 European women with previous GDM and 1 European control woman) could not be satisfactorily modelled. Therefore, there are no estimates for Si<\/sub> and Sg<\/sub> for these 3 subjects. The disposition index (Si <\/sub>x AIRg<\/sub>), an index of \u03b2-cell function which takes account of prevailing insulin sensitivity and which exploits the hyperbolic relationship between the two,19,20<\/sup> was calculated by the method described by Kahn et al.20 Glucose and lipid parameters were measured enzymatically on the Olympus AU5200 analyser (Olympus Opticals UK Ltd.). HDL-cholesterol levels were assayed following selective precipitation of non-HDL lipoproteins using dextran sulphate and magnesium. Specific insulin levels were measured by ELISA, using an assay with a sensitivity of 0.55 pmol\/l and inter-assay variation not exceeding 8.6% at low and high values.24<\/sup> The cross-reactivity of proinsulin with the intact insulin assay is negligible.<\/p>\n Statistical analysis<\/strong><\/em><\/p>\n Since the distribution of variables was not normal, we log-transformed data to achieve normality. Data are presented as geometric mean (SD range). Statistical analyses were performed in SPSS 8.0 for Windows using unpaired Student\u2019s t-test and analysis of variance, as appropriate, on log-transformed data. Repeat analysis was performed incorporating: a. BMI as a covariate and b. BMI, age, parity and time since delivery as covariates.<\/p>\n RESULTS<\/strong><\/span><\/p>\n European women with previous GDM displayed clinical and metabolic abnormalities compared to European control women: significantly higher BMI (p-value =0.033), waist\/hip (p-value =0.001), fasting triglycerides (p-value =0.001), fasting insulin (p-value =0.004) and lower HDL-cholesterol (p-value =0.024) (Table 2<\/a>). Allowing for BMI and additional analysis allowing for BMI, age, time since delivery and parity, reduced but did not abolish differences between the groups; apart from HDL-cholesterol, all remained statistically significant (Table 2<\/a>). Insulin sensitivity was significantly lower in women with a history of GDM compared to controls as suggested by the lower HOMA (%S) (p-value=0.004, adjusted for BMI p-value, was 0.006; adjusted for BMI, age, time since delivery and parity the p-value was 0.003). Insulin sensitivity (Si) derived from the FSIVGTT was lower [p-value =0.010, adjusted for BMI p-value, was suggestive (borderline) = 0.054, and adjusted for BMI, age, time since delivery and parity the p-value was 0.051] in the post-GDM group. Although \u03b2-cell function was higher in the basal state in the post-GDM group, as suggested by higher HOMA (%B) (p-value =0.013, adjusted for BMI, the p-value was 0.015; adjusted for BMI, age, time since delivery and parity the p-value was 0.007), the disposition index was lower (p-value =0.017), suggesting \u03b2-cell dysfunction following glucose stimulation in the post-GDM group (Table 2<\/a>). Since the disposition index represents a measure of \u03b2-cell function, which already takes account of prevailing insulin sensitivity, we did not perform analyses that adjusted disposition index for measures of adiposity. There was no difference in glucose effectiveness (Sg<\/sub>), acute insulin response (AIRg<\/sub>) and glucose disappearance rate (Kg<\/sub>) between the two groups.<\/p>\n In South Asian women there was no statistically significant difference in BMI, waist\/hip ratio, fasting lipids, fasting insulin and fasting glucose between the GDM and the control group (Table 3<\/a>). There was no statistically significant difference in insulin sensitivity derived by both HOMA (%S) and the FSIVGTT (Si) between post-GDM and control women; however, South Asian post-GDM women displayed lower \u03b2-cell HOMA (%B) (p-value=0.048) and borderline-lower AIRg<\/sub> (p-value=0.052), suggesting decreased \u03b2-cell function (Table 3<\/a>). After adjustment for BMI and for BMI, age, time since delivery and parity, these differences became suggestive (p-values of borderline significance) of decreased \u03b2-cell function (Table 3<\/a>). There was no significant difference in Sg<\/sub>, Kg<\/sub> and the disposition index between the two groups.<\/p>\n In Afro-Caribbean women there was no statistically significant difference in BMI, fasting lipids, fasting insulin and fasting glucose between the GDM and the control group, but post-GDM women displayed significantly higher waist\/hip ratio than control women (Table 4)<\/a> . There was no statistically significant difference in insulin sensitivity and \u03b2-cell function, derived by FSIVGTT (Si<\/sub>, AIRg<\/sub> and disposition index) and HOMA (%S and %B) or Sg<\/sub>, but Kg<\/sub> derived by the FSIVGTT was significantly lower in the GDM group compared to Afro-Caribbean control women (p-value =0.003, adjusted for BMI p-value =0.003 and adjusted for BMI, age, time since delivery and parity p-value =0.006), suggesting subtle glucose intolerance in the Afro-Caribbean post-GDM women (Table 4<\/a>).<\/p>\n
\n<\/strong><\/span>
\nGestational diabetes (GDM) is defined as carbohydrate intolerance first recognised in pregnancy.1<\/sup> Although the majority of women with GDM return to normal glucose tolerance after delivery, they remain, as a group, at substantially increased risk of developing type 2 diabetes in later life.2,3<\/sup> GDM may therefore be considered a forerunner of type 2 diabetes and the study of women in the post-GDM state serves as a model for detection of early metabolic abnormalities associated with subsequent disease progression.<\/p>\n
\n<\/sup>
\nWhilst insulin action and secretion are both clearly impaired during pregnancies complicated by GDM, the extent and nature of the metabolic disturbances persisting after pregnancy remain unclear. Even when the confounding effects of hyperglycaemia are avoided by studying women with a history of GDM who have normal glucose tolerance postpartum, conflicting conclusions have been reached.9-11<\/sup> Some of these disparities may have resulted from the study of small numbers of subjects tested and of mixed ethnic origin.<\/p>\n
\n<\/strong><\/span>
\nOver a two-year period, 34 European, 16 South Asian and 10 Afro-Caribbean women with previous GDM were retrospectively recruited from West London antenatal databases (Table 1<\/a>). Although the clinical criteria and protocols used for diagnosis and treatment of GDM in different recruitment centres vary to a limited extent, only women who fulfilled WHO (1999) criteria for glucose intolerance during pregnancy (glucose >7.8mmol\/l following 75g of glucose) were considered to have had proven GDM for the purposes of this study.13<\/sup> All women were invited to attend for a standardized clinical examination which included anthropometry and a 75g oral glucose tolerance test (OGTT) to assess current (i.e. postpartum) glucose homeostasis. Body mass index (BMI), calculated as weight (kg)\/height(metres)2<\/sup>, and waist (the least circumference between the costal margin and the iliac crest) to hip (the circumference at the level of the greater trochanters) ratio were recorded. Where the OGTT could not be completed (in 2 European, 6 South Asian and 5 Afro-Caribbean subjects), a fasting plasma glucose and an HbA1c were performed. Based on these measures, interpreted according to WHO (1999) criteria using all available results,13<\/sup> all women had normal glucose homeostasis at the time of the metabolic studies. Autoimmune diabetes was excluded on the basis of measurement of anti-GAD antibody titre.14
\n<\/sup>
\nOver the same period, 44 European, 16 South Asian and 19 Afro-Caribbean control women were recruited retrospectively from the same sources. All these women had successfully passed the modified O\u2019Sullivan screening test15<\/sup> during the most recent pregnancy and had maintained normal glucose levels throughout all previous pregnancies. In addition, all control subjects had a normal fasting glucose according to WHO (1999) criteria13<\/sup> and a normal HbA1c at the time of this study. There was no family history of type 2 diabetes in 1st degree relatives in any of the women.<\/p>\n
\nThe study protocol had been approved by the local ethics committee for each recruitment hospital. Informed consent was obtained from all subjects studied.<\/p>\n
\n
\nFrequently sampled intravenous glucose tolerance test (FSIVGTT)<\/strong><\/em><\/p>\n
\n<\/sup>
\nHOMA
\n<\/strong><\/span>
\nFasting glucose and specific insulin measures (the average from 2 samples taken with a 15 minute collection interval) were used to derive estimates of \u03b2-cell function (%B) and insulin sensitivity (%S) using the \u201chomeostasis model assessment\u201d (HOMA).21<\/sup> HOMA-derived parameters have been shown to correlate well with other measures of insulin resistance and \u03b2-cell function.22,23
\n<\/sup>
\nBiochemical assays<\/strong><\/em><\/p>\n