1. Introduction
Women with gestational diabetes (GDM) have a high risk of developing manifest diabetes mellitus [
1,
2,
3,
4,
5], and a meta-analysis in 2009 reported a 7.5-fold increased risk of developing diabetes in women who were diagnosed with GDM during pregnancy [
6]. The prevalence of GDM is generally proportional to the prevalence of underlying type 2 diabetes, and the risk for women with GDM to develop diabetes mellitus later in life depends on several factors, like follow-up time, insulin need during pregnancy, and ethnicity [
7]. It is known that mothers with GDM are older and have a higher prevalence of obesity and chronic hypertensive disease than the background population [
8,
9]. GDM is also an independent risk factor for long-term cardiovascular morbidity [
10]. As obesity and GDM are closely related, rising rates of overweightness and obesity worldwide may result in more women with GDM. However, very long-term follow-up studies, which have also been influenced by the obesity epidemic during recent years, are scarce [
11]. We recently reported a prevalence of 25% of manifest diabetes mellitus in a large Swedish cohort of GDM women when followed up by a questionnaire 11 years after GDM diagnosis [
12]. A weakness of questionnaire studies is that they do not provide biochemical data for the diagnosis of glucose intolerance, i.e., impaired fasting glucose (IFG), impaired glucose tolerance (IGT), or diabetes type validated with autoantibodies (glutamic acid decarboxylase, GAD). Despite the well-known risk of future diabetes in these women, follow-up has not been optimal: a newly published report from primary care units in England indicated that only 20% of GDM women had a regular follow-up [
13]; a single-centre study recently reported that around 50% were followed up with oral glucose tolerance testing (OGTT) [
14]; and in our study, 60% reported a follow-up after GDM diagnosis [
12].
With the present study, we aimed to conduct a careful biochemical evaluation, including a 2 h 75 g OGTT, in a cohort of women in the southeast region of Sweden diagnosed with GDM 12 years earlier to assess the long-term risk of glucose metabolic abnormalities.
3. Discussion
This study shows a very high prevalence of glucose metabolic abnormalities manifesting in the long run after GDM diagnosis with a 2 h 75 g OGTT and using capillary blood glucose ≥9.0 mmol/L (≈plasma glucose ≥10.0 mmol/L) as a cut-off for diagnosis. At the follow-up after a median of 12 years, one-third of our GDM population had developed diabetes mellitus; this is in line with other studies showing a diabetes prevalence of 30–40% at follow-up of women with GDM 5–15 years after the index pregnancy [
3,
4,
11]. In addition to women with manifest diabetes, many women exhibit IGT or IFG—46% in our study, leaving only 24% with normal glucose tolerance. In a follow-up of GDM five years after pregnancy, Ekelund et al. found IFG/IGT in 22% of the women [
3], and Lauenborg et al. [
11] reported 27% IFG/IGT after a median follow-up time of 9.5 years. Several studies thus suggest that most women with GDM will have diabetes or glucose intolerance at a long-term follow-up. Both in our study and in the study by Linné et al. [
4], several undiagnosed cases were found, underlining the need for regular testing to detect diabetes. The results of the follow-up with 2 h 75 g OGTT thus show that GDM diagnosed using the Diabetic Pregnancy Study Group (DPSG) diagnostic criteria [
15,
16] is a strong predictor for future prediabetes/diabetes mellitus. There is no consensus regarding the use of fasting plasma glucose, OGTT, or HbA1c when following up on GDM. Glucose abnormalities defined by both IFG and IGT are risk predictors for diabetes, even if IGT defines a much larger target population for prevention [
17]. It should be noted that in our study, HbA1c could not distinguish between normal subjects and subjects with IGT and/or IFG. However, in the group with diabetes, HbA1c was elevated; it should therefore be considered for follow-up and is today generally recommended for the diagnosis of diabetes [
18]. Overall, the high frequency of glucose abnormalities found in this and previous studies indicates that women with GDM should be offered lifelong regular glucose measurements.
The original reason for defining and detecting GDM was to be able to identify those women at risk of diabetes mellitus later in life, a risk which depends on several factors [
5]. In the last consensus statement by the International Association of the Diabetes and Pregnancy Study Groups IADPSG [
19], endorsed by the American Diabetes Association (ADA) [
20] and World Health Organization (WHO) [
21], GDM is diagnosed if one or more of the following glucose values is met or exceeded: a fasting venous plasma glucose ≥5.1 mmol/L and/or a 1 h value ≥10.0 mmol/L and/or 2 h value ≥8.5 mmol/L post 75 g OGTT. This recommendation identifies GDM in 16–18% of all women if applied to the Hyperglycemia and Adverse Pregnancy Outcome Study (HAPO) data [
19]. This much higher prevalence of GDM than with older criteria like those used in the present study has been criticized, as the HAPO study was observational in design and cannot provide information regarding the effectiveness of treatment of women with glucose concentrations that are lower than the former thresholds for GDM diagnosis [
22,
23,
24,
25]. The IADPSG criteria for the diagnosis of GDM derived from the HAPO study lack data about the future risk of diabetes [
26,
27]. It is important to note that our results could not be directly applicable to these lower cut-off values for GDM used in the HAPO study.
Overweightness and metabolic syndrome are more common in GDM women [
28,
29]. The metabolic burden was high in our patients, with 33% overweight subjects and 43% obese individuals at follow-up. This high prevalence of remaining overweight/obese, together with the strong correlations found in our study between C-peptide/proinsulin and cardiovascular risk markers like body composition, lipids, and blood pressure, indicates that the GDM diagnosis overlaps with the metabolic syndrome to a great extent [
30] and is potentially linked to the insulin-like growth factor (IGF) system [
31].
Type 2 diabetes could to some extent be prevented by lifestyle interventions, while type 1 diabetes, which was 10 times more common after GDM (5%) than in the Swedish general population (0.5%) [
12], is not preventable. We measured GAD antibodies in the present study, and despite excluding subjects with known type 1 diabetes, we found additional individuals with signs of autoimmunity against islet cells. Despite excluding women with known type 1 diabetes from the study, the finding of further GAD-positive women strengthens the data that around 5–6% in a Scandinavian GDM population are GAD positive postpartum [
32]. These results, showing a high prevalence of type 1 diabetes in GDM women, could to some extent explain the results from a newly published Finnish study, which also found a high risk for diabetes in non-obese women five years after GDM [
33].
The major strength of this study was the careful biochemical and clinical analysis of the women with previous GDM. We measured plasma glucose with a capillary technique as both capillary blood and venous plasma samples can be used for the diagnosis of diabetes mellitus, but are not considered to be interchangeable [
34,
35]. Reliable ambulant methods for capillary glucose speed up the diagnostic procedure for diabetes mellitus, are economically favorable compared with venous plasma glucose methods, and avoid the instability of glucose in plasma after blood sampling [
36]. How representative the cohort of 51 women is of all GDM women in Sweden may be questioned [
12]. We previously reported that high BMI, 2 h OGTT blood glucose, and having been born outside Europe are risk factors for the progression of GDM to overt diabetes [
12]. Comparing the baseline data of the cohort of 51 women who were examined in this study with those of the larger group, we found that slightly more women were of non-European heritage in the larger group and their 2 h OGTT glucose was significantly higher (
Table 1). This suggests that if there is any bias, we have underestimated the prevalence of diabetes.
In conclusion, GDM diagnosed from a capillary blood glucose level of ≥9.0 mmol/L (≈plasma glucose ≥10.0 mmol/L) after a 75 g OGTT indicates the later development of impaired glucose metabolism in most of the women diagnosed with GDM, and our data indicates that all women with prior GDM should be offered lifelong regular glucose measurements. Moreover, type 1 diabetes should be regarded as a diagnostic alternative.
4. Materials and Methods
4.1. Participants
In a prospective nationwide registration of GDM in Sweden, 2085 pregnancies in 2025 women were reported from 1 January 1995 to 31 December 1999 [
12]. For the diagnosis of GDM, a 75 g oral glucose load was used and a 2 h capillary blood glucose value ≥9.0 mmol/L (≈plasma glucose ≥10 mmol/L) was taken as a GDM-positive result according to the European Association for the Study of Diabetes (EASD) study group for gestational diabetes [
15,
37]. Capillary B-glucose used for diagnosis at the maternal healthcare clinics was analysed with HemoCue
® (HemoCue Ltd., Ängelholm, Sweden), a method with a high precision and accuracy and total CV less than 5% [
38]. The GDM cohort was followed up 8.5–13.5 years after initial diagnosis with a questionnaire, which was answered by 1324 GDM women (65%) [
12].
We identified women (
n = 195) from the Swedish cohort living in our regional area of the southeast of Sweden, near Linköping University Hospital (1 million inhabitants). Of these, the first questionnaire was answered by 146 women (75%). In the present study, we invited women from the original cohort of 195 individuals living within 100 km of Linköping University Hospital for clinical examination. Individuals with known type 1 diabetes (
n = 7) were not included. In all, 51 women accepted the invitation to participate in the clinical part of the study (
Table 4 and
Figure 1).
4.2. Measurements
The 51 women invited were examined by way of fasting plasma glucose and a 75 g OGTT. Additional laboratory measurements, including lipids, blood pressure, pulse rate, weight, length, waist circumference, and sagittal abdominal diameter, were collected. Duplicate samples of 5 μL capillary blood were collected in HemoCue Glucose cuvettes and analysed in a HemoCue Glucose 201 Analyser (HemoCue Ltd., Ängelholm, Sweden) [
38], which converts blood glucose concentrations to equivalent plasma glucose concentrations by multiplying by an adjustment factor of 1.117. For the diagnosis of diabetes mellitus, a capillary fasting plasma glucose level ≥7.0 mmol/L and/or a 2 h capillary plasma glucose value ≥12.2 mmol/L after a 75 g oral glucose load was considered as diabetes mellitus, while fasting plasma glucose <7.0 mmol/L and a 2 h capillary plasma glucose value ≥8.9 and <12.1 mmol/L was defined as impaired glucose tolerance (IGT). Impaired fasting glucose (IFG) was defined as a fasting plasma glucose between 6.1 and 6.9 mmol/L with a 2 h capillary plasma glucose value <8.9 mmol/L [
39]. HbA1c was analysed using a TOSOH G7 automated hemoglobin analyser (Tosoh Bioscience, Tokyo, Japan). Proinsulin was measured using Mercodia Proinsulin ELISA (Mercodia, Uppsala, Sweden) and C-peptide using Mercodia C-peptide ELISA (Mercodia, Uppsala, Sweden), and GAD antibodies were measured as described by Kordonouri et al. [
40]. Blood lipids were analysed using the routine method in the department of clinical chemistry. Blood pressure was measured in the supine position after a 5 min rest and the mean of two measurements in the right arm was recorded. Hypertension was defined as blood pressure >140/90.
Overweightness was defined as 25 ≤ BMI < 30 kg/m2 and obesity as BMI ≥ 30 kg/m2. A waist circumference in a standing position of >80 cm was considered a moderate risk and >88 cm a high risk. Data regarding the development of diabetes, treatment for diabetes mellitus, treatment during pregnancy, other diseases, smoking, any concomitant medication, follow-up after pregnancy, contraceptives, later pregnancies, heredity for diabetes, the mother’s birth weight, married/unmarried, and occupation were collected at the visit.
4.3. Ethics
The ethics committe at Linköping University identification code M147-04 (approval date 10 November 2004), approved the study. The participants in the study were informed about the purpose of the study and gave their written informed consent.
4.4. Statistical Analysis
Continuous variables are presented as mean ± standard deviation (SD) or median with interquartile range (IQ), as appropriate. Number and percentage are reported for categorical variables, and differences between groups were evaluated using the Chi-square test. Comparisons between groups of continuous data were made using an unpaired Student’s t-test. Statistical analysis of differences between three or more groups was performed using ANOVA with Bonferroni post hoc testing, significant at the 0.05 level. For the estimation of linear associations, the Pearson correlation coefficient was calculated. Values for serum proinsulin were transformed to their corresponding natural logarithm (ln) to accomplish a normal distribution. A significance level of p < 0.05 (two-sided) was used. Statistics were calculated on a PC using the Statistical package for Social Science (SPSS Statistics 23, IBM, Stockholm, Sweden).