Abstract
Introduction
The main aim was to study whether the long‐term incidences of type 2 diabetes, pre‐diabetes and metabolic syndrome differed between women who were treated with metformin or insulin for gestational diabetes.
Material and methods
This 9‐year follow‐up study of two open‐label randomized trials compares metformin and insulin treatments of gestational diabetes. In all, 165 women, 88 previously treated with insulin and 77 treated with metformin in the index pregnancy, were included in the analyses. An oral glucose tolerance test was performed, and measures of anthropometry, glucose metabolism, serum lipids and inflammatory markers were compared between the treatment groups. Disorders of glucose metabolism (pre‐diabetes and type 2 diabetes) at the 9‐year follow‐up was the primary outcome of this study. This study was registered at ClinicalTrials.gov: NCT02417090.
Results
The incidences of pre‐diabetes and type 2 diabetes (40.3% vs. 46.6%, odds ratio [OR] 0.77, 95% CI 0.40–1.50, p = 0.51), type 2 diabetes (14.3% vs. 15.9%, OR 0.88, 95% CI 0.34–2.26, p = 0.94), pre‐diabetes (26.0% vs. 30.7%, OR 0.79, 95% CI 0.38–1.65, p = 0.62), and metabolic syndrome (45.9% vs. 55.2%, OR 0.69, 95% CI 0.35–1.35, p = 0.31) were comparable between the metformin and insulin groups. Moreover, there were no evident differences in the individual measures of anthropometry, glucose metabolism including HOMA‐insulin resistance, serum lipids or inflammatory markers between the two treatment groups.
Conclusions
Treatment of gestational diabetes with metformin vs. insulin during pregnancy is unlikely to have diverging long‐term effects on maternal anthropometry, glucose metabolism or serum lipids. From this perspective, both treatments may be considered in gestational diabetes.
Keywords: gestational diabetes, insulin, metabolic syndrome, metformin, type 2 diabetes
Abbreviations
- ALT
alanine aminotransferase
- ApoB
apolipoprotein B
- BMI
body mass index
- GDM
gestational diabetes
- HbA1c
glycated hemoglobin
- HDL
high‐density lipoprotein
- HOMA‐IR
homeostasis model assessment of insulin resistance
- hs‐CRP
high sensitivity C‐reactive protein
- LDL
low‐density lipoprotein
- MetS
metabolic syndrome
- OGTT
oral glucose tolerance test
- OR
odds ratio
- T2DM
type 2 diabetes
Key message.
Gestational diabetes is a risk factor for future abnormal glucose metabolism. Treatment with metformin compared with insulin during pregnancy does not appear to modify this risk.
1. INTRODUCTION
Gestational diabetes (GDM) is a common disorder in pregnancy 1 that increases the risk of adverse pregnancy outcomes such as macrosomia, preeclampsia, cesarean delivery, neonatal hypoglycemia, hyperbilirubinemia and need for neonatal intensive care. 2 , 3 In addition, prior GDM is a major risk factor for the development of type 2 diabetes (T2DM) 4 and metabolic syndrome (MetS) 5 later in life.
Impaired beta‐cell function and increased insulin resistance have been observed in women already prior to GDM diagnosis, 6 as well as after delivery. 7 Therefore, it has been questioned whether the treatment of GDM could modify the risk of long‐term metabolic morbidity. A follow‐up study 7 years after a randomized trial comparing treatment vs. no treatment of mild GDM found no differences in rates of T2DM, MetS, homeostasis model assessment of insulin resistance (HOMA‐IR) or body mass index (BMI) between the groups. 8 However, in that study the GDM was mild and only about 5% of the participants required pharmacological treatment. 9
Metformin, an alternative to insulin for GDM patients requiring pharmacological intervention, has been proven to be comparable regarding short‐term perinatal outcomes. 10 However, during pregnancy, metformin treatment has been associated with higher triglycerides, very‐low‐density lipoprotein lipids, low‐density lipoprotein (LDL) lipids, total fatty acids, saturated fatty acids and monounsaturated fatty acids in pregnancy. 11 , 12 One follow‐up study found similar maternal anthropometry, amount of body fat, and rates of self‐reported diabetes 7–9 years after the pregnancy in mothers randomized to metformin or insulin during the pregnancy; 13 however, the maternal glucose metabolism, lipids and inflammatory markers were not assessed.
The aim of the present study was to evaluate whether the rates of T2DM, pre‐diabetes and MetS differ between the metformin‐ and insulin‐treated GDM patients 9 years after delivery. Furthermore, we compared anthropometry, serum lipids and inflammatory markers between these two groups, as these parameters are known to predict subsequent T2DM. 14 , 15 , 16 We hypothesized that there would be no large differences between the groups.
2. MATERIAL AND METHODS
This study is a follow‐up of two Finnish randomized trials conducted between 2005 and 2009 at Oulu University Hospital (patients were also recruited and treated at Kainuu Central Hospital), 17 and between 2006 and 2010 at Turku University Hospital. 18 The original recruitment, randomization and perinatal outcomes have been reported previously in detail. 17 , 18 Briefly, women with a singleton pregnancy and need for medical treatment for GDM were randomized to receive either metformin or insulin. Originally, 97 women were included in the study in Oulu and 217 in Turku. Randomization was performed in 1:1 ratio using sealed envelopes at the study hospital when medical treatment was indicated. In the original trials, the sample sizes were calculated to demonstrate 30% difference in the rates of macrosomia (birthweight >4000 g) in Oulu 17 and non‐inferiority in birthweights in Turku. 18
GDM was diagnosed by a 75‐g 2‐h oral glucose tolerance test (OGTT). In Oulu, one elevated value was required for diagnosis; the thresholds for fasting, 1‐h, and 2‐h glucose values in capillary plasma were 5.3, 11.0 and 9.6 mmol/L. In Turku, two elevated values were required; the corresponding threshold values in venous plasma were 4.8, 10.0 and 8.7 mmol/L until December 2008, when the Finnish national guidelines were published, and 5.3, 10.0 and 8.6 mmol/L thereafter. The maximum metformin doses were 2250 mg/day in Oulu and 2000 mg/day in Turku. Insulin therapy was performed using neutral protamine Hagedorn insulin and/or rapid‐acting insulin analogues insulin lispro or insulin aspart. In the metformin groups, 32% and 21% of women required additional insulin to achieve glycemic goals in Oulu and Turku, respectively. In our follow‐up study, these women are included in the metformin group. There were no differences between the treatment groups in the primary outcomes; birthweight 18 and rates of macrosomia and large‐for‐gestational‐age neonates, 17 respectively. For this analysis, birthweight was adjusted for gestational weeks and fetal sex according to new Finnish population‐based references, and represented as SD units and centiles. 19
At the follow‐up study visit, weight, height, abdominal circumference and waist circumference were measured three times, and the mean values of the three measurements were used in the analyses. Three blood pressure measurements were also taken, and the mean values of the second and third measurements were used. Weight categories were defined as normal weight (BMI <25 kg/m2), overweight (BMI ≥25 and <30 kg/m2) and obese (BMI ≥30 kg/m2).
A 75‐g OGTT was performed, and glucose values were measured at fasting and at 30 and 120 min. Glycated hemoglobin (HbA1c), C‐peptide, insulin, alanine aminotransferase (ALT), apolipoprotein A‐1 (ApoA1), apolipoprotein B (ApoB), total cholesterol, LDL cholesterol, high‐density lipoprotein (HDL) cholesterol, triglycerides, high sensitivity C‐reactive protein (hs‐CRP), ferritin and adiponectin were measured from fasting samples. Analysis details are described in Appendix S1. HOMA‐IR was calculated as fasting glucose (mmol/L) x fasting insulin (mU/L)/22.5, and non‐HDL cholesterol as total cholesterol – HDL cholesterol.
Pre‐diabetes was defined as impaired fasting glucose (fasting glucose 6.1–6.9 mmol/L), impaired glucose tolerance (OGTT 2‐h glucose 7.8–11.0 mmol/L) or HbA1c 6.0–6.4% (42–47 mmol/mol). T2DM was defined as fasting glucose ≥7.0 mmol/L, OGTT 2‐h glucose >11.0, or HbA1c ≥6.5% (48 mmol/mol). MetS was defined according to International Diabetes Federation as waist circumference ≥80 cm plus at least two of the following four criteria: (1) raised triglyceride concentration (≥1.7 mmol/L), (2) reduced HDL cholesterol concentration (<1.29 mmol/L) or treatment for dyslipidemia, (3) raised blood pressure (systolic ≥130 or diastolic ≥85 mmHg) or treatment for previously diagnosed hypertension, and (4) raised fasting plasma glucose concentration (≥5.6 mmol/L) or previously diagnosed T2DM. 20
Parity and educational level of the participants were recorded. Breastfeeding duration was recorded only in the participants recruited at Turku. 21 The women were also asked to describe their well‐being as “extremely poor”, “poor”, “average”, “good” or “excellent”, and physical activity was assessed by asking the women on how many days per week exercises were done. Data regarding the offspring anthropometry, glucose and lipid metabolism have been published in a separate report. 22
Primary outcome of the study was the rate of disorder of glucose metabolism, ie composite outcome of pre‐diabetes and T2DM. Main secondary outcomes were incidences of pre‐diabetes, T2DM and MetS—long‐term consequences of GDM 4 , 5 with clinical significance. Exploratory secondary outcomes were measures of maternal anthropometry, components of MetS according to IDF, 20 individual measures of glucose metabolism, insulin resistance, low‐grade inflammation and lipid homeostasis, and adiponectin (detailed in Appendix S1).
2.1. Statistical analyses
Baseline characteristics of both cohorts during pregnancy were compared between the participants and non‐participants of the follow‐up study and between the metformin and insulin groups. Additional comparisons between the participants and non‐participants were performed in the metformin and insulin groups separately. Outcome variables were compared between the metformin and insulin groups in the whole population. Between‐group comparisons were performed using the t‐test, the Mann–Whitney U‐test, the chi‐square test or Fisher’s exact test depending on the data. In cases of slightly skewed data, the P‐values were calculated after log‐transformation.
We hypothesized that there would be no differences between the treatment groups and therefore no adjustments for the P‐values were done. For odds ratios (OR), 95% confidence intervals (CI) were calculated with Fisher’s exact test. All analyses were performed using R statistical software (version 4.0.3).
2.2. Power analyses
This being a follow‐up study, we performed post‐hoc power analysis regarding the primary outcome (composite of pre‐diabetes and T2DM). To demonstrate a clinically significant 30% difference in the rate of outcome between 47.0% (observed rate in the insulin group) and 32.6% with power of 80%, 192 participants would have been needed per group.
2.3. Ethical approval
This study was approved by the ethics committee of The Hospital District of Southwest Finland (ETMK 31/2015) on April 27, 2015 and registered at ClinicalTrials.gov (NCT02417090). All study participants provided an informed consent.
3. RESULTS
A flowchart of the study is presented in Figure 1. Three mother–child dyads were excluded from the original data due to the child's trisomy‐21, valproate syndrome and stillbirth. Between August 2015 and November 2019, 311 dyads were contacted and 173 (55.6%) were willing to participate. Thirty‐seven mothers were not reached, 65 were not willing to participate, 12 canceled their participation, three did not participate due to undisclosed disease of the child, and for 21 the reason was not recorded. The reasons not to participate were similar between the metformin and insulin groups. Finally, 82 and 91 mother–child dyads participated in the 9‐year follow‐up in the metformin and insulin groups, respectively. In two cases, only the child participated in the follow‐up study. Two mothers were diagnosed with autoimmune diabetes after the index pregnancy, one was pregnant at the time of follow‐up, and three had undergone a bariatric surgery; these women were excluded from the analyses. Fifteen patients with previously diagnosed T2DM were excluded from the comparison of glucose metabolism, insulin resistance, inflammation and ALT. Eleven patients with medications affecting lipids (ie antihyperglycemic drugs, antihyperlipidemic drugs, systemic hormonal contraceptives) were also excluded from the analysis of serum lipids.
FIGURE 1.
Study flow chart. ALT, alanine aminotransferase; LADA, latent autoimmune diabetes in adults; RCT, randomized controlled trial; T1DM, type 1 diabetes; T2DM, type 2 diabetes
3.1. Baseline characteristics
Following randomization, the participants of the follow‐up study were older (32.6 vs. 31.3 years, p = 0.046) and weighted less (80.4 vs. 84.3 kg, p = 0.043) compared with the non‐participating women (Table 1). They also had a GDM diagnosis later in gestation (27 vs. 26 weeks, p = 0.015) and a slightly lower fasting glucose (5.42 vs. 5.55 mmol/L, p = 0.051) and HbA1c (37.4 vs. 38.3 mmol/mol, p = 0.074). When the metformin and the insulin groups were analyzed separately, these differences between the participants and non‐participants were in the same direction, although with reduced statistical power (P‐values 0.052–0.36; data not shown). Among the follow‐up participants, the only difference was the higher rate of labor inductions in the insulin (58.2%) than in the metformin group (39.0%, p = 0.018) (Table 2).
TABLE 1.
Baseline characteristics at index pregnancy: comparison of follow‐up participants and non‐participants
| Participants | Non‐participants | p‐value | |
|---|---|---|---|
| Age (years) | 32.6 ± 5.2 (173) | 31.3 ± 5.6 (141) | 0.046 |
| Parity (n) | 1 [0–2] (173) | 1 [0–2] (141) | 0.93 |
| Nulliparous (n/total) | 68/173 (39.3) | 57/141 (40.4) | 0.92 |
| Smoking (n/total) | 21/171 (12.3) | 25/137 (18.2) | 0.19 |
| Weight (kg) | 80.4 ± 15.2 (169) | 84.3 ± 18.0 (130) | 0.043 |
| BMI (kg/m2) | 29.2 ± 5.4 (173) | 30.5 ± 5.8 (141) | 0.057 |
| Glucose metabolism | |||
| OGTT fasting glucose (mmol/L) | 5.42 ± 0.55 (172) | 5.55 ± 0.64 (140) | 0.051 |
| OGTT 60‐min glucose (mmol/L) | 11.2 [10.5–12.0] (171) | 11.2 [10.5–12.3] (141) | 0.93 |
| OGTT 120‐min glucose (mmol/L) | 8.22 ± 1.85 (172) | 8.02 ± 1.74 (140) | 0.35 |
| HbA1c at OGTT (mmol/mol) | 37.4 ± 4.3 (173) | 38.3 ± 4.5 (140) | 0.074 |
| HbA1c at OGTT (%) | 5.57 ± 0.40 (173) | 5.65 ± 0.41 (140) | |
| Gw at OGTT (weeks) | 27 [26–28] (172) | 26 [25–28] (140) | 0.015 |
| HbA1c at 36 gw (mmol/mol) | 38.3 ± 3.9 (155) | 39.1 ± 4.2 (127) | 0.084 |
| HbA1c at 36 gw (%) | 5.65 ± 0.36 (155) | 5.73 ± 0.39 (127) | |
| Treatment | |||
| Gw at randomization (weeks) | 31 [29–32] (173) | 31 [29–32] (141) | 0.46 |
| Treatment | 0.36 | ||
| Insulin (n/total) | 91/173 (52.6) | 66/141 (46.8) | |
| Metformin (n/total) | 82/173 (47.4) | 75/141 (53.2) | |
| Additional insulin (n/total, of metformin‐treated patients) | 22/82 (26.8) | 16/75 (21.3) | 0.54 |
| Obstetric outcomes | |||
| Weight gain in pregnancy (kg) | 8.5 ± 4.9 (173) | 7.9 ± 5.2 (140) | 0.33 |
| Hypertensive disorders (n/total) | 13/173 (7.5) | 16/141 (11.3) | 0.33 |
| Induction of labor (n/total) | 85/173 (49.1) | 64/141 (45.4) | 0.58 |
| Operative vaginal delivery (n/total) | 20/173 (11.6) | 8/141 (5.7) | 0.10 |
| Cesarean delivery (n/total) | 36/173 (20.8) | 25/141 (17.7) | 0.59 |
| Gestational weeks at delivery (n/total) | 39.1 [38.4–40.1] (173) | 39.1 [38.1–40.0] (141) | 0.34 |
| Birthweight (g) | 3590 ± 500 (173) | 3630 ± 460 (141) | 0.55 |
| Adjusted birthweight (SD) |
0.11 ± 1.11 (173) |
0.27 ± 1.17 (141) | 0.23 |
| Adjusted birthweight (centiles) | 50 [27–79] (173) | 58 [34–84] (141) | 0.19 |
| SGA (n/total) | 19/173 (11.0) | 13/141 (9.2) | 0.74 |
| LGA (n/total) | 30/173 (17.3) | 26/141 (18.4) | 0.92 |
| Breastfeeding a (months) | 3.5 [2–6] (115) | 3 [2–6] (66) | 0.33 |
Notes: Continuous data are expressed as mean ± SD (total n) or median [interquartile range] (total n) and binary data as n/total n (%). Adjusted birthweight was adjusted for gestational age at delivery and neonate sex according to Finnish population reference charts. p‐values are given for the t‐test, the Mann–Whitney U‐test, the chi‐square test or Fisher’s exact test where appropriate.
Abbreviations: BMI, body mass index; Gw and gw, gestational week; HbA1c, glycated hemoglobin; LGA, large‐for‐gestational‐age (adjusted birthweight >90th centile); OGTT, oral glucose tolerance test; SD, standard deviation; SGA, small‐for‐gestational age (adjusted birthweight <10th centile).
Data available only from Turku.
TABLE 2.
Baseline characteristics at index pregnancy of the follow‐up study participants
| Insulin | Metformin | p‐value | |
|---|---|---|---|
| Age (years) | 32.5 ± 5.5 (91) | 32.7 ± 4.9 (82) | 0.83 |
| Parity (n) | 1 [0–2] (91) | 1 [0–2] (82) | 0.56 |
| Nulliparous (n/total) | 40/91 (44.0) | 28/82 (34.1) | 0.24 |
| Smoking (n/total) | 15/90 (16.7) | 6/81 (7.4) | 0.11 |
| Weight (kg) | 79.3 ± 14.9 (90) | 81.5 ± 15.6 (79) | 0.36 |
| BMI (kg/m2) | 28.9 ± 4.9 (91) | 29.6 ± 5.9 (82) | 0.41 |
| Glucose metabolism | |||
| OGTT fasting glucose (mmol/L) | 5.45 ± 0.46 (91) | 5.38 ± 0.63 (81) | 0.37 |
| OGTT 60‐min glucose (mmol/L) | 11.2 [10.3–11.8] (90) | 11.3 [10.6–12.3] (81) | 0.11 |
| OGTT 120‐min glucose (mmol/L) | 8.10 ± 1.78 (90) | 8.35 ± 1.93 (80) | 0.38 |
| HbA1c at OGTT (mmol/mol) | 37.5 ± 4.1 (91) | 37.2 ± 4.6 (82) | 0.64 |
| HbA1c at OGTT (%) | 5.59 ± 0.38 (91) | 5.56 ± 0.42 (82) | |
| Gw at OGTT (weeks) | 27 [26–28] (91) | 27 [26–28] (81) | 0.96 |
| HbA1c at 36 gw (mmol/mol) | 38.5 ± 4.1 (80) | 38.1 ± 3.7 (75) | 0.54 |
| HbA1c at 36 gw (%) | 5.67 ± 0.38 (80) | 5.64 ± 0.34 (75) | |
| Treatment | |||
| Gw at randomization (weeks) | 31 [29–32] (91) | 31 [29–32] (82) | 0.56 |
| Additional insulin (n/total, of metformin treated patients) | 22/82 (26.8) | ||
| Obstetric outcomes | |||
| Weight gain in pregnancy (kg) | 8.6 ± 5.2 (91) | 8.4 ± 4.6 (82) | 0.85 |
| Hypertensive disorders (n/total) | 9/91 (9.9) | 4/82 (4.9) | 0.26 |
| Induction of labor (n/total) | 53/91 (58.2) | 32/82 (39.0) | 0.018 |
| Operative vaginal delivery (n/total) | 10/91 (11.0) | 10/82 (12.2) | 0.99 |
| Cesarean delivery (n/total) | 19/91 (20.9) | 17/82 (20.7) | 0.99 |
| Gestational weeks at delivery (n/total) | 39.1 [38.5–40.3] (91) | 39.0 [38.4–40.1] (82) | 0.61 |
| Birthweight (g) | 3570 ± 520 (91) | 3610 ± 480 (82) | 0.60 |
| Adjusted birthweight (SD) | 0.08 ± 1.16 (91) | 0.15 ± 1.08 (82) | 0.67 |
| Adjusted birthweight (centiles) | 50 [21–80] (91) | 52 [31–79] (82) | 0.46 |
| SGA (n/total) | 12/91 (13.2) | 7/82 (8.5) | 0.46 |
| LGA (n/total) | 17/91 (18.7) | 13/82 (15.9) | 0.77 |
| Breastfeeding a (months) | 3 [2–6] (58) | 4 [1.5–7] (57) | 0.42 |
Notes: Continuous data are expressed as mean ± SD (total n) or median [interquartile range] (total n) and binary data as n/total n (%). Adjusted birthweight was adjusted for gestational age at delivery and neonate sex according to Finnish population reference charts. p‐values are given for the t‐test, the Mann–Whitney U‐test, the χ 2 test or the Fisher's exact test where appropriate.
Abbreviations: BMI, body mass index; Gw and gw, gestational week; HbA1c, glycated hemoglobin; LGA, large for gestational age (adjusted birthweight >90th centile); OGTT, oral glucose tolerance test; SD, standard deviation; SGA, small for gestational age (adjusted birthweight <10th centile).
Data available only from Turku.
3.2. Maternal characteristics at follow‐up
At the follow‐up visit 9 years after the index pregnancy, there were no differences between the groups in the number of pregnancies after the index pregnancy, rates of smoking, educational level or physical activity (Table 3). The perception of one’s own health was “poor” in only 7.9% and 6.8% of the women in the metformin and insulin groups, respectively, and none considered their health “extremely poor”.
TABLE 3.
Maternal characteristics at follow‐up
| Insulin | Metformin | p‐value | |
|---|---|---|---|
| Age (years) | 42.1 ± 5.4 (88) | 42.3 ± 4.8 (77) | 0.86 |
| Parity (n) | 2 [2–3] (88) | 3 [2–3] (77) | 0.69 |
| ∆Parity from the study recruitment (n) | 1 [1–2] (64) | 1 [1–2] (60) | 0.95 |
| Smoking (n/total) | 11/88 (12.5) | 9/77 (11.7) | 0.99 |
| Education post‐secondary or higher (n/total) | 32/84 (38.1) | 22/71 (30.1) | 0.45 |
| Exercise (days/week) | 2.75 [1.0–4.0] (88) | 3.0 [1.5–5.0] (77) | 0.17 |
Notes: Continuous data is expressed as mean ± SD (total n) or median [interquartile range] (total n), and binary data as n/total n (%). p‐values are given for the t‐test, the Mann–Whitney U test or the chi‐square test where appropriate.
3.3. Primary and main secondary outcomes
Nine years after the index pregnancy, the observed rate of composite outcome of pre‐diabetes and T2DM was comparable between the metformin and insulin groups (40.3 vs. 46.6%, OR 0.77, 95% CI 0.40–1.50, p = 0.51) (Table 4). The rates of pre‐diabetes (26.0 vs. 30.7%, OR 0.79, 95% CI 0.38–1.65, p = 0.62), T2DM (14.3 vs. 15.9%, OR 0.88, 95% CI 0.34–2.26, p = 0.94) and MetS (45.9 vs. 55.2%, OR 0.69, 95% CI 0.35–1.35, p = 0.31) were also comparable between the groups. Seven (9.1%) and eight (9.1%) women had previously diagnosed T2DM, and four (5.2%) and six (6.8%) had an incident, ie newly diagnosed, T2DM (based on OGTT and/or HbA1c at 9‐year study visit) in the metformin and insulin groups, respectively.
TABLE 4.
Primary and main secondary outcomes at 9‐year follow‐up
| Insulin | Metformin | P‐value | Odds ratio [95% CI] | Risk difference [95% CI] | |
|---|---|---|---|---|---|
| Primary outcome | |||||
| Pre‐diabetes and type 2 diabetes | 41/88 (46.6) | 31/77 (40.3) | 0.51 | 0.77 [0.40–1.50] | −6.3% [−21.5–8.8] |
| Main secondary outcomes | |||||
| Pre‐diabetes | 27/88 (30.7) | 20/77 (26.0) | 0.62 | 0.79 [0.38–1.65] | −4.7% [−18.4–9.0] |
| Type 2 diabetes | 14/88 (15.9) | 11/77 (14.3) | 0.94 | 0.88 [0.34–2.26] | −1.6% [−12.6–9.3] |
| Metabolic syndrome | 48/87 (55.2) | 34/74 (45.9) | 0.31 | 0.70 [0.35–1.35] | −9.2% [−24.7–6.2] |
Note: Data are expressed as n/total n (%), p‐values are given for the chi‐square test.
Abbreviation: CI, confidence intervals.
3.4. Exploratory secondary outcomes at follow‐up
At the 9‐year follow‐up, over half of the women were obese and only 13.4% normal weight; there were no differences between the metformin and insulin groups. There were no evident differences between the groups in maternal anthropometry, glucose metabolism, insulin resistance, ALT, inflammation, serum lipids and adiponectin (details in Table 5). Lipid profile was, however, marginally better in the metformin group, with lower ApoB, LDL and non‐HDL cholesterol, and higher HDL cholesterol (p‐values 0.057–0.099).
TABLE 5.
Exploratory secondary outcomes
| Insulin | Metformin | p‐value | |
|---|---|---|---|
| Maternal anthropometry | |||
| Weight (kg) | 83.4 ± 16 (88) | 85.5 ± 16.8 (76) | 0.43 |
| BMI (kg/m2) | 30.8 ± 5.34 (88) | 31.4 ± 6.1 (76) | 0.50 |
| BMI category (n/total) | 0.60 | ||
| Normal, <25 kg/m2 | 11/88 (12.6) | 11/76 (14.5) | |
| Overweight, 25–30 kg/m2 | 32/88 (36.4) | 22/76 (28.9) | |
| Obese, ≥30 kg/m2 | 45/88 (51.1) | 43/76 (56.6) | |
| Waist circumference (cm) | 95 ± 12 (88) | 96 ± 14 (76) | 0.71 |
| Hip circumference (cm) | 105 ± 13 (88) | 106 ± 14 (76) | 0.83 |
| ∆Weight from 1st antenatal visit in the index pregnancy (kg) | 3.6 ± 6.6 (87) | 4.2 ± 8.0 (73) | 0.59 |
| Systolic blood pressure (mmHg) | 124 ± 13.9 (87) | 122 ± 14.8 (75) | 0.40 |
| Diastolic blood pressure (mmHg) | 74.9 ± 8.49 (87) | 74.6 ± 9.17 (74) | 0.86 |
| Metabolic syndrome (individual components) | |||
| Elevated fasting glucoseb (n/total) | 52/88 (59.1) | 45/77 (58.4) | 0.99 |
| Elevated blood pressurec (n/total) | 31/87 (35.6) | 25/74 (33.8) | 0.94 |
| Dyslipidemiad (n/total) | 48/88 (54.5) | 36/77 (46.8) | 0.40 |
| Waist circumference ≥80 cm (n/total) | 80/88 (90.9) | 66/76 (86.8) | 0.56 |
| Glucose metabolism | |||
| IFG (n/total) | 21/80 (26.2) | 19/70 (27.1) | 0.99 |
| IGT (n/total) | 17/80 (21.2) | 11/70 (15.7) | 0.51 |
| Elevated HbA1c (n/total) | 14/79 (17.7) | 11/70 (15.7) | 0.91 |
| OGTT fasting glucose (mmol/L) | 5.70 ± 0.79 (80) | 5.69 ± 0.60 (70) | 0.92a |
| OGTT 30‐min glucose (mmol/L) | 9.09 ± 1.74 (80) | 8.92 ± 1.29 (70) | 0.66a |
| OGTT 120‐min glucose (mmol/L) | 6.51 ± 1.96 (80) | 6.19 ± 1.92 (70) | 0.24a |
| HbA1c (mmol/mol) | 38.6 [36.7–40.5] (79) | 38.6 [36.7–40.5] (70) | 0.94 |
| HbA1c (%) | 5.60 [5.40–5.79] (79) | 5.60 [5.40–5.79] (70) | 0.96 |
| Fasting insulin (mU/L) | 13.8 ± 8.6 (80) | 13.1 ± 8.4 (70) | 0.56a |
| Fasting C‐peptide (μg/L) | 2.46 [1.94–3.22] (80) | 2.35 [1.85–2.96] (70) | 0.37 |
| HOMA‐IR | 3.55 ± 2.21 (80) | 3.42 ± 2.46 (70) | 0.61a |
| ALT and inflammation | |||
| ALT (U/L) | 23.6 ± 15.8 (80) | 24.2 ± 16 (70) | 0.76a |
| Ferritin (μg/L) | 39.0 [22.5–76.1] (78) | 40.8 [18.9–66.2] (70) | 0.51 |
| hs‐CRP (mg/L) | 1.67 [0.60–4.43] (77) | 1.0 [0.35–3.6] (67) | 0.086 |
| Lipids and adiponectin | |||
| ApoA1 (g/L) | 1.43 ± 0.22 (72) | 1.46 ± 0.25 (67) | 0.46 |
| ApoB (g/L) | 0.97 ± 0.25 (72) | 0.90 ± 0.20 (67) | 0.068 |
| Total cholesterol (mmol/L) | 5.40 ± 1.09 (72) | 5.24 ± 0.84 (67) | 0.34 |
| HDL cholesterol (mmol/L) | 1.34 ± 0.39 (72) | 1.46 ± 0.39 (67) | 0.057 |
| LDL cholesterol (mmol/L) | 3.44 ± 0.92 (72) | 3.21 ± 0.72 (67) | 0.099 |
| Non‐HDL cholesterol (mmol/L) | 4.06 ± 1.09 (72) | 3.78 ± 0.88 (67) | 0.096 |
| Triglycerides (mmol/L) | 1.23 ± 0.58 (72) | 1.18 ± 0.66 (67) | 0.55a |
| Adiponectin (μg/L) | 7.38 ± 4.72 (72) | 7.69 ± 3.98 (67) | 0.39a |
Notes: Continuous data are expressed as mean ± SD (total n) or median [interquartile range] (total n) and binary data as n/total n (%). p‐values are given for the t‐test (where the data that were log‐transformed prior to the test are marked with superscript letter a), the Mann–Whitney U‐test, or the chi‐square test where appropriate. The individual components of metabolic syndrome are based on the International Diabetes Federation criteria; bfasting glucose ≥5.6 mmol/L or previously diagnosed type 2 diabetes, cblood pressure ≥130 systolic or ≥85 diastolic mmHg or treatment for hypertension, dplasma triglycerides ≥1.7 mmol/L or HDL cholesterol <1.29 mmol/L or treatment for dyslipidemia.
Abbreviations: ALT, alanine aminotransferase; ApoA1, apolipoprotein A‐1; ApoB, apolipoprotein B; BMI, body mass index; HDL, high‐density lipoprotein; HOMA‐IR, homeostasis model assessment of insulin resistance; hs‐CRP, high‐sensitivity C‐reactive protein; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; LDL, low‐density lipoprotein; OGTT, oral glucose tolerance test; SD, standard deviation.
4. DISCUSSION
In this 9‐year follow‐up of two randomized trials, we found comparable rates of pre‐diabetes, T2DM and MetS between women who were treated with metformin or insulin for GDM. Additionally, individual measures of glucose metabolism, insulin resistance, lipids, blood pressure and anthropometry were comparable. In this respect, choosing either metformin or insulin treatment for GDM probably has only small effects, if any, on the long‐term maternal health. Furthermore, the differences between metformin and insulin groups in maternal circulating triglycerides in late pregnancy 11 , 12 disappear during the first decade after pregnancy.
Approximately half of the original study population participated in the 9‐year follow‐up, which is close to the drop‐out rate reported in similar studies. 8 , 13 The participating women were older, weighted less at baseline and tended to have lower fasting glucose and HbA1c compared with those not participating in the follow‐up study. Our findings at 9 years are, however, in line with a previous report from the Turku study population at 1 year postpartum where the participation rate was 83% and the rates of impaired fasting glucose and impaired glucose tolerance were similar between the insulin and metformin groups. 21
It is currently not known whether GDM treatment improves long‐term outcomes, 8 or whether differences between metformin and insulin treatments exist. Dietary interventions and increased physical activity in pregnancy may have long‐lasting positive effects on metabolism, and treatment allocation to metformin or insulin could affect differently on maternal adherence to lifestyle treatment. Metformin not only alleviates hyperglycemia but also affects low‐grade inflammation and the gastrointestinal tract, 23 which may have long‐term effects. In our study, 26.8% of the participants in the metformin group also received insulin, which might have attenuated the differences in the outcome measures. Exclusion of these women from the metformin group, however, would have introduced bias.
We found neither of the treatments to be superior in affecting the burden of adverse metabolic traits in the long term, although we observed numerically marginally lower hs‐CRP, lower ApoB and higher HDL cholesterol in the metformin group. We cannot exclude small differences in glucose and lipid metabolism between the treatment groups, but metformin treatment seems very unlikely to be inferior to insulin, to a clinically significant extent.
Possible small long‐term effects of treatment may have been obscured by modifying factors such as weight gain, breastfeeding and changes in diet. Therefore, a larger study would be needed to compare the possible long‐term outcomes with more certainty. Given that follow‐up of the largest randomized trial comparing metformin and insulin in GDM included 208 subjects (109 at 7 and 99 at 9 years postpartum), 13 sufficiently powered long‐term follow‐up studies are unlikely to emerge and perhaps meta‐analyses will provide more conclusive data in the future.
Rowan et al similarly compared metformin‐ and insulin‐treated GDM from two separate study centers at 7 or 9 years after the index pregnancy, respectively. 13 They found no differences in maternal anthropometrics or rate of self‐reported diabetes between the treatment groups. Maternal glucose metabolism was not, however, assessed in that study. Our results strengthen the findings of similar maternal anthropometry and rates of diabetes 13 and provide a more detailed insight into maternal metabolism 9 years after the index pregnancy. Moreover, as approximately half of the T2DM cases in our study represented incident T2DM, the self‐reported diagnosis in the study by Rowan et al may be an underestimation.
Maternal serum triglycerides in late pregnancy are higher in metformin‐ than in insulin‐treated woman with GDM. 11 , 12 This difference disappeared in our data during the first decade after pregnancy. Our results suggest that the shift towards a more atherogenic lipid profile in the metformin‐ compared with the insulin‐treated patients in late pregnancy is unlikely to lead to inferior metabolic health by 9 years postpartum.
Markers of low‐grade inflammation such as CRP and interleukin 6 14 have been associated with increased risk of developing T2DM, whereas adiponectin is protective. 24 We evaluated the markers of inflammation (hs‐CRP and ferritin) and adiponectin 9 years after the index pregnancy and observed that metformin treatment may be related to lower hs‐CRP. This observation was, however, among the exploratory secondary outcomes and hence there is a high risk of a false‐positive finding.
The rate of T2DM (15.2%) was similar to that reported previously 4 and it is expected to increase further after 10 years postpartum. 25 Strikingly, we found roughly as many new cases of T2DM (incident T2DM) as preexisting ones in the study population. This result probably reflects insufficient regular assessment of glucose metabolism and cardiovascular risk factors postpartum.
To our knowledge this is the second‐largest follow‐up study of women randomized to metformin or insulin treatment of GDM, and the only one to include subsequent detailed evaluation of maternal glucose metabolism. Other strengths of this study are use of data from two study centers, longitudinal design and treatment allocation by randomization. The diagnostic criteria 17 , 18 and management protocols between the study centers were slightly different, thus improving the generalizability of the results.
There are some limitations in our study. First, our follow‐up study population may not be fully representative of all women with previous medically treated GDM due to possible selection at the time of recruitment (more health‐conscious individuals participate more often) and to loss to follow‐up. The non‐participating subjects differed at baseline slightly from those who participated the 9‐year follow‐up. Thus the participating and non‐participating subjects may have responded differently to metformin or insulin, causing the observed results to differ from a real‐life scenario. Therefore, observational studies are required to complement our data.
Secondly, our study was underpowered to demonstrate small differences in the rates of the primary composite outcome of pre‐diabetes and T2DM, or the main secondary outcomes: pre‐diabetes, T2DM and MetS.
Thirdly, the diagnostic criteria for GDM were slightly different in the two study centers and, in addition, actually changed in Turku during the recruitment period. However, due to limited numbers of participants it is not possible to judge whether these issues affected the results.
Finally, a core outcome set for follow‐up studies comparing insulin and/or oral glucose‐lowering agents in GDM was published after the initiation of this study;26 therefore, the number of pregnancies complicated by GDM after the index pregnancy was not recorded in our study.
5. CONCLUSION
The available data regarding long‐term metabolic effects of the main pharmacological treatment options for GDM have been scarce. Based on our study, neither metformin nor insulin seems to be superior to the other regarding the long‐term effects on maternal health. From this perspective, both treatments may be considered in GDM.
CONFLICT OF INTEREST
None.
AUTHOR CONTRIBUTIONS
KT, TR, HiN and MV were the authors of the original randomized trials and provided the baseline data of the study participants. KT, TR and HaN designed the follow‐up study. HiN and EP collected the follow‐up data. BML planned and supervised the collection and analysis of blood samples. MH combined the datasets, performed the statistical analyses and wrote the first draft of the manuscript. All the authors approved the final version of the manuscript.
Supporting information
Appendix S1
ACKNOWLEDGMENTS
We thank all the women who participated in the original studies and the 9‐year follow‐up study. We extend our gratitude to the midwives and doctors at the Turku University Hospital, Oulu University Hospital, and Kainuu Central Hospital, who contributed to the recruitment and treatment of these women. Especially we thank the study nurses Ulla Torkko and Tiina Latva‐Aho and laboratory technicians Minna Romo and Elina Åkerblom.
Huhtala MP, Nikkinen H, Paavilainen E, et al. Comparison of glucose metabolism and anthropometry in women with previous gestational diabetes treated with metformin vs. insulin: 9‐year follow‐up of two randomized trials. Acta Obstet Gynecol Scand. 2022;101:514–523. doi: 10.1111/aogs.14343
Funding informationThis study was funded by the Finnish Foundation for Pediatric Research (Finland), the Diabetes Research Foundation (Finland), the State funding for university‐level health research (Finland), and the Turku University Hospital Foundation (Finland). The funding sources had no involvement in the collection, analysis or interpretation of the data.
REFERENCES
- 1. Yuen L, Saeedi P, Riaz M, et al. Projections of the prevalence of hyperglycaemia in pregnancy in 2019 and beyond: results from the International Diabetes Federation Diabetes Atlas, 9th edition. Diabetes Res Clin Pract. 2019;157:107841. [DOI] [PubMed] [Google Scholar]
- 2. Langer O, Yogev Y, Most O, Xenakis EMJ. Gestational diabetes: the consequences of not treating. Am J Obstet Gynecol. 2005;192:989‐997. [DOI] [PubMed] [Google Scholar]
- 3. HAPO Study Cooperative Research Group , Metzger BE, Lowe LP, et al. Hyperglycemia and adverse pregnancy outcomes. N Engl J Med. 2008;358:1991‐2002. [DOI] [PubMed] [Google Scholar]
- 4. Vounzoulaki E, Khunti K, Abner SC, Tan BK, Davies MJ, Gillies CL. Progression to type 2 diabetes in women with a known history of gestational diabetes: systematic review and meta‐analysis. BMJ. 2020;369:m1361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Lauenborg J, Mathiesen E, Hansen T, et al. The prevalence of the metabolic syndrome in a Danish population of women with previous gestational diabetes mellitus is three‐fold higher than in the general population. J Clin Endocrinol Metab. 2005;90:4004‐4010. [DOI] [PubMed] [Google Scholar]
- 6. Catalano PM, Tyzbir ED, Wolfe RR, et al. Carbohydrate metabolism during pregnancy in control subjects and women with gestational diabetes. Am J Physiol Metab. 1993;264:E60‐E67. [DOI] [PubMed] [Google Scholar]
- 7. Kautzky‐Willer A, Prager R, Waldhäusl W, et al. Pronounced insulin resistance and inadequate β‐cell secretion characterize lean gestational diabetes during and after pregnancy. Diabetes Care. 1997;20:1717‐1723. [DOI] [PubMed] [Google Scholar]
- 8. Casey BM, Rice MM, Landon MB, et al. Effect of treatment of mild gestational diabetes on long‐term maternal outcomes. Am J Perinatol. 2020;37:475‐482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Landon MB, Spong CY, Thom E, et al. A multicenter, randomized trial of treatment for mild gestational diabetes. N Engl J Med. 2009;361:1339‐1348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Farrar D, Simmonds M, Bryant M, et al. Treatments for gestational diabetes: a systematic review and meta‐analysis. BMJ Open. 2017;7:e015557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Barrett HL, Gatford KL, Houda CM, et al. Maternal and neonatal circulating markers of metabolic and cardiovascular risk in the metformin in gestational diabetes (MiG) trial: responses to maternal metformin versus insulin treatment. Diabetes Care. 2013;36:529‐536. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Huhtala MS, Tertti K, Rönnemaa T. Serum lipids and their association with birth weight in metformin and insulin treated patients with gestational diabetes. Diabetes Res Clin Pract. 2020;170:108456. [DOI] [PubMed] [Google Scholar]
- 13. Rowan JA, Rush EC, Plank LD, et al. Metformin in gestational diabetes: the offspring follow‐up (MiG TOFU): body composition and metabolic outcomes at 7–9 years of age. BMJ Open Diabetes Res Care. 2018;6:e000456. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Pradhan AD, Manson JE, Rifai N, Buring JE, Ridker PM. C‐reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA. 2001;286:327‐334. [DOI] [PubMed] [Google Scholar]
- 15. Lappas M, Georgiou HM, Velagic A, Willcox JC, Permezel M, Shub A. Do postpartum levels of apolipoproteins prospectively predict the development of type 2 diabetes in women with previous gestational diabetes mellitus? Exp Clin Endocrinol Diabetes. 2019;127:353‐358. [DOI] [PubMed] [Google Scholar]
- 16. Lai M, Liu Y, Ronnett GV, et al. Amino acid and lipid metabolism in post‐gestational diabetes and progression to type 2 diabetes: A metabolic profiling study. PLOS Medicine. 2020;17:e1003112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Ijäs H, Vääräsmäki M, Morin‐Papunen L, et al. Metformin should be considered in the treatment of gestational diabetes: a prospective randomised study. BJOG. 2011;118:880‐885. [DOI] [PubMed] [Google Scholar]
- 18. Tertti K, Ekblad U, Koskinen P, Vahlberg T, Rönnemaa T. Metformin vs. insulin in gestational diabetes. A randomized study characterizing metformin patients needing additional insulin. Diabetes Obes Metab. 2013;15:246‐251. [DOI] [PubMed] [Google Scholar]
- 19. Sankilampi U, Hannila ML, Saari A, Gissler M, Dunkel L. New population‐based references for birth weight, length, and head circumference in singletons and twins from 23 to 43 gestation weeks. Ann Med. 2013;45:446‐454. [DOI] [PubMed] [Google Scholar]
- 20. Alberti KGMM, Zimmet P, Shaw J. Metabolic syndrome ‐ a new world‐wide definition. A consensus statement from the international diabetes federation. Diabet Med. 2006;23:469‐480. [DOI] [PubMed] [Google Scholar]
- 21. Pellonperä O, Rönnemaa T, Ekblad U, Vahlberg T, Tertti K. The effects of metformin treatment of gestational diabetes on maternal weight and glucose tolerance postpartum ‐ a prospective follow‐up study. Acta Obstet Gynecol Scand. 2016;95:79‐87. [DOI] [PubMed] [Google Scholar]
- 22. Paavilainen E, Tertti K, Nikkinen H, et al. Metformin versus insulin therapy for gestational diabetes: effects on offspring anthropometrics and metabolism at the age of 9 years: a follow‐up study of two open‐label, randomized controlled trials. Diabetes Obes Metab. 2022;24:402‐410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Rena G, Hardie DG, Pearson ER. The mechanisms of action of metformin. Diabetologia. 2017;60:1577‐1585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Spranger J, Kroke A, Möhlig M, et al. Adiponectin and protection against type 2 diabetes mellitus. Lancet. 2003;361:226‐228. [DOI] [PubMed] [Google Scholar]
- 25. Pirkola J, Pouta A, Bloigu A, et al. Prepregnancy overweight and gestational diabetes as determinants of subsequent diabetes and hypertension after 20‐year follow‐up. J Clin Endocrinol Metab. 2010;95:772‐778. [DOI] [PubMed] [Google Scholar]
- 26. Bogdanet D, Reddin C, Macken E, et al. Follow‐up at 1 year and beyond of women with gestational diabetes treated with insulin and/or oral glucose‐lowering agents: a core outcome set using a Delphi survey. Diabetologia. 2019;62:2007‐2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix S1