Objective
We sought to compare weight loss in the first 6 weeks postpartum among women with gestational diabetes mellitus (GDM) treated with metformin or placebo, a promising therapy to reduce later risk of progression to diabetes mellitus.
Study Design
We conducted a pilot, randomized trial of metformin vs placebo in postpartum women with GDM. Women with pre-GDM, unable to tolerate metformin, resumed on insulin or oral hypoglycemic agent, delivered <34 weeks’ gestation, or with a body mass index <20 kg/m 2 were excluded. Women were randomized to either metformin 850 mg daily for 7 days, then metformin 850 mg twice a day for the next 5 weeks or placebo prescribed in a similar frequency. The subject, health care provider, and research staff were blinded to the treatment. The primary outcome was weight change from delivery to 6 weeks postpartum. Secondary outcomes included the percentage of women achieving their self-reported prepregnancy weight, reported medication adherence, adverse effects, and satisfaction. Differences in weight change between groups were determined by Wilcoxon rank sum test and in achieving prepregnancy weight by χ 2 test.
Results
Of 114 women randomized, 79 (69.3%) completed the 6 weeks; 36 (45.6%) were randomized to metformin and 43 (54.4%) to placebo. Metformin and placebo groups were similar in median weight loss (6.3 kg [range, –0.3 to 19.8] vs 6.5 kg [range, –0.3 to 12.1], P = .988) and percentage of women achieving reported prepregnancy weight (41.7 vs 37.2%, P = .69). Self-reported adherence in taking >50% of medication was 75% at 3 weeks and 97% at 6 weeks. Nausea, diarrhea, and hypoglycemia were reported in approximately 11-17% of women and 56-63% reported dissatisfaction with the medication.
Conclusion
Women with GDM lost approximately 6 kg by 6 weeks’ postpartum. This was similar in both groups and resulted in <50% of women achieving their prepregnancy weight. Although the reported adherence and satisfaction with the medication was high, adverse effects were reported with nearly 1 in 5 women including nausea, diarrhea, and hypoglycemia. Contrary to expectation, we found no evidence of benefit from metformin. However, longer treatment periods and larger studies with minimal attrition may be warranted.
Gestational diabetes mellitus (GDM) is carbohydrate intolerance first recognized in pregnancy. After delivery, carbohydrate intolerance is expected to resolve gradually. Once a woman develops GDM, she remains at risk of recurrence in future pregnancies, and has a 7-fold risk of developing type 2 diabetes mellitus (DM) later in life. Additionally, GDM develops in 14% of obese women. Excessive gestational weight gain (EGWG) occurs in approximately 45% of obese pregnant women, many of whom have GDM, and has been linked to postpartum weight retention. This fuels the progression of obesity even after pregnancy ends and, as a result, places these women at further risk of developing type 2 DM.
Weight loss is a key element associated with preventing the onset of diabetes. Every 1 kg lost is associated with a 16% reduction in diabetes risk. Lifestyle modifications focused on nutrition and exercise are first-line therapies for prevention and control of type 2 DM in obese, nonpregnant women. Though weight loss is recommended after delivery, behavioral alterations are difficult postpartum when mothers have the onset of responsibilities to a newborn. Though good intentions are present, realistic circumstances may fall short of expectations and many cease these behavioral modifications over time. With the failure to reduce postpartum weight gain in multiple randomized trials, the evaluation of new strategies, including medications, is needed.
Metformin is an insulin-sensitizing medication. It functions to improve insulin sensitivity by reducing fasting plasma glucose and insulin concentrations. Importantly, it has been shown to be beneficial in reducing weight. However, the true mechanism by which this insulin sensitizer results in weight loss is not fully known. Weight loss is a natural physiologic occurrence in the postpartum period including water loss and adipose tissue. Enhancement of this natural weight loss represents an opportunity for obstetricians to intervene and halt the progression towards persistent obesity. We considered that metformin could act in conjunction with the physiologic weight loss unique to the postpartum period to accentuate the inherent descending slope of weight. Moreover, postpartum women accustomed to taking oral medications, such as prenatal vitamins, would find the concept of an oral daily medication for the purpose of weight loss appealing and be compliant with this medical treatment. If weight loss were enhanced, this could blunt the progression of obesity and potentially avoid the development of type 2 DM later in life in this high-risk group. Our objective was to conduct a pilot study to assess whether metformin increased postpartum weight loss compared to placebo among women with GDM.
Materials and Methods
Study design
We conducted a pilot randomized, placebo-controlled trial of metformin vs placebo from January 2011 through January 2014, at the Memorial Hermann Hospital-Texas Medical Center and Lyndon B. Johnson Hospital in Houston, TX. On the postpartum ward, women with GDM were invited to participate within 24 hours of delivery and prior to discharge. Women were excluded if they had pre-GDM (either type 1 or 2 DM), reported inability to tolerate metformin, discharged home on insulin or oral hypoglycemic agent, delivered <34 weeks of pregnancy, were <18 or >49 years old, or had a body mass index (BMI) <20 kg/m 2 . The diagnosis of GDM (treated with insulin, oral hypoglycemic agent, or diet control) was made >24 weeks based on a documented 1-hour glucola screen >200 mg/dL or by a confirmatory 3-hour glucola test (based on either the Carpenter and Coustan or the Diabetes Task Force criteria).
Informed consent and randomization
This study was approved by the institutional review board at University of Texas Health, Houston, TX (no. HSC-MS-10-0426, approved October 2010). After reviewing the potential benefits, risks, and adverse effects of the medication and placebo, written informed consent was obtained. Subjects were randomized to either metformin or placebo via central randomization conducted by the Investigational Drug Service (IDS) pharmacy at Memorial Hermann Hospital-Texas Medical Center and stratified by site. Permuted block randomization with a random fashion was used to prevent imbalances between groups. The subject, health care provider, research staff, and statistician were blinded to the treatment group. Only the IDS pharmacy knew the treatment group. The randomization scheme was unmasked after completion of the analysis.
Study procedures
Prior to discharge, the subject was started on either metformin or placebo. The metformin dose was 850 mg daily for 7 days, then 850 mg twice a day for the next 5 weeks. The placebo was similar in size, color, and taste, and prescribed in a similar frequency: once daily for 7 days, then twice daily for the next 5 weeks for a total of 6 weeks. The metformin and placebo were compounded by a licensed compounding pharmacy and monitored on a routine schedule for quality assurance and potency of the drugs by the IDS. The subject received the medication or placebo in prefilled push cards designating the regimen for subject convenience. There were no drop-in, drop-out, or crossover of subjects.
Within 24 hours of delivery, maternal weight was measured using a single specified digital weight scale (did not require calibration) on a hard surface at each site. A research nurse counseled all participants regarding their diet and provided a simple exercise plan of walking a minimum of 30 minutes, 3-5 times a week. Maternal demographics, clinical characteristics, and neonatal outcomes were collected. At 3 weeks postpartum (range, 2–4 weeks), a research nurse contacted the subject via telephone to inquire about adverse effects and ability to take the prescribed medication (metformin or placebo). At 6 weeks postpartum (range, 5–8 weeks), maternal weight was measured with the same digital weight scale used for the initial maternal postpartum weight. The research nurse again inquired about adverse effects and conducted a satisfaction survey.
Study outcomes
The primary outcome was weight change in kilograms defined as: weight change = Weight postpartum(PP) -Weight 6wk . Secondary outcomes included the rate of retained gestational weight represented as the percentage of women achieving their self-reported prepregnancy weight and percentage of women achieving their ideal body weight. Demographic, pregnancy characteristics, self-reported medication adherence, and adverse effects of medications were collected. A satisfaction survey was performed assessing difficulty and satisfaction with diet, exercise, and medication according to a 5-point Likert scale.
Analysis
We conducted an intent-to-treat analysis of patients as randomized. We also analyzed patients as treated in exploring the potential impact on subject adherence to treatment on weight change. Differences in weight change (primary outcome) between groups were determined by Wilcoxon rank sum (Mann-Whitney) test. Differences between groups in achieving prepregnancy weight and achieving ideal body weight were compared using χ 2 and relative risks (RR) with 95% confidence intervals (CI) were reported. A P value < .05 was considered statistically significant. Statistical analysis was conducted using STATA version 21 (StataCorp, College Station, TX).
Sample size calculation was determined based on data from Scholl et al, who demonstrated that maternal weight showed little change between 4-6 weeks and 6 months’ postpartum. Because the majority of women with GDM have EGWG, we used weight change data in this population to calculate a sample size. Based on a maternal weight change of 10.4 ± 6.1 kg, effect size of 50%, alpha of 0.05, power of 0.8, and attrition rate of 10%, a sample size was calculated to be 40 subjects per group.
Results
Population
Of 239 women screened for eligibility, 114 (47.7%) women were randomized as described in Figure 1 following Consolidated Standards of Reporting Trials (CONSORT) guidelines. In all, 35 randomized women did not complete the primary outcome assessment at 6 weeks. At the 3-week telephone call, the metformin group had 3 subjects choose to stop the study, 1 subject was instructed to discontinue the study by her physician, and 9 subjects were lost to follow-up and did not answer their telephone. In the placebo group, 4 subjects chose to stop the study, 1 subject was instructed to discontinue the study by her physician, and 3 subjects were lost to follow-up and did not answer their telephone. At the 6-week visit, 6 subjects were lost to follow-up in the metformin group, and 8 subjects in the placebo group.
A total of 79 (69%) of the randomized subjects completed the assessment of primary outcome; 36 (45.6%) received metformin and 43 (54.4%) received placebo. Maternal characteristics, pregnancy outcomes, and neonatal outcomes of those who completed the primary evaluation are described in Table 1 . Women randomized to metformin but who did not complete the primary outcome had a higher rate of prior preterm birth and delivered at an earlier gestational age compared to placebo ( Table 2 ). They also had a higher BMI at enrollment compared to other groups. Thus, the subjects with the highest BMI (potentially the highest risk group) did not complete the study and were not part of the final analysis of weight loss.
| Variable | Metformin, n = 36 | Placebo, n = 43 |
|---|---|---|
| Maternal age, y | 31.6 ± 5.3 | 28.8 ± 6.4 |
| Gravidity | 2 (1–7) | 3 (1–12) |
| Parity | 2 (0–5) | 3 (0–10) |
| Race | ||
| African American | 4 (11.1%) | 12 (27.9%) |
| Caucasian | 1 (2.8%) | 5 (11.6%) |
| Hispanic | 30 (83.3%) | 25 (58.1%) |
| Asian | 1 (2.8%) | 1 (2.3%) |
| Insurance | ||
| Public | 30 (83.3%) | 41 (95.4%) |
| Private | 5 (13.9%) | 2 (4.6%) |
| Self-pay | 1 (2.8%) | 0 (0%) |
| Prior preterm birth | 2 (5.6%) | 3 (7%) |
| Prior cesarean delivery | 15 (41.7%) | 12 (27.9%) |
| Smoking | 0 (0%) | 1 (2.3%) |
| Gestational age at delivery, wk | 37.8 ± 1.0 | 38.1 ± 1.3 |
| GDM, 50-g glucola >200 mg/dL | 7 (19.4%) | 10 (23.8%) |
| Prepregnancy weight, kg | 76.1 (54.1–131.8) | 77.3 (45.4–131.8) |
| Postpartum weight at enrollment, kg | 79.8 (59.4–165.6) | 88.2 (59.6–143.6) |
| Gestational weight gain, kg | 8.1 (–12.9 to 65.6) | 9.6 (–11.0 to 37.1) |
| Excessive gestational weight gain | 11 (33.3%) | 20 (48.8%) |
| BMI at enrollment, kg/m 2 | 31.3 (21.1–42.8) | 31.9 (17.2–46.3) |
| Obese at enrollment | 26 (72.2%) | 35 (81.4%) |
| Class III obese at enrollment | 7 (19.4%) | 13 (30.2%) |
| Rate of preterm birth <37 wk | 3 (8.3%) | 4 (9.3%) |
| Preeclampsia | 5 (3.1%) | 5 (18.4%) |
| Cesarean delivery rate | 4 (50%) | 0 (42.1%) |
| Male sex | 6 (53.1%) | 4 (42.1%) |
| Birthweight, g | 3372 (2370–4745) | 3315 (2409–4260) |
| Birth length, cm | 50.3 (46.5–54.7) | 50.0 (45.4–55.0) |
| Birth head circumference, cm | 34.5 (31.0–36.0) | 33.5 (33.6–37.0) |
| Macrosomia rate (>4000 g) | 5 (13.9%) | 1 (2.3%) |
| Breast-feeding immediately postpartum | 29 (84.3%) | 31 (79.0%) |
| Demographic | Women who did not complete 6-wk assessment, n = 35 | |
|---|---|---|
| Metformin, n = 20 | Placebo, n = 15 | |
| Maternal age, y | 30.7 ± 6.7 | 31.1 ± 7.2 |
| Gravidity | 4 (1–9) | 3.5 (1–7) |
| Parity | 3 (0–9) | 2.5 (0–6) |
| Race | ||
| African American | 5 (25%) | 4 (26.7%) |
| Caucasian | 0 (0%) | 1 (6.7%) |
| Hispanic | 14 (70%) | 8 (53.3%) |
| Asian | 1 (5%) | 1 (6.7%) |
| Insurance | ||
| Public | 18 (90%) | 13 (86.7%) |
| Private | 2 (10%) | 2 (13.3%) |
| Self-pay | 0 (0%) | 0 (0%) |
| Prior preterm birth | 9 (45%) a | 1 (6.7%) |
| Prior cesarean delivery | 8 (40%) | 7 (46.7%) |
| Smoking | 0 (0%) | 4 (26.7%) |
| Gestational age at delivery, wk | 36.8 ± 1.8 a | 38.3 ± 0.9 |
| GDM, 50-g glucola >200 mg/dL | 2 (10%) | 3 (20%) |
| Prepregnancy weight, kg | 72.7 (51.4–131.4) | 76.2 (49.1–111.5) |
| Postpartum weight at enrollment, kg | 87.4 (63.0–122.5) | 89.9 (63.7–132.9) |
| Gestational weight gain, kg | 11.5 (–26.6 to 24.1) | 8.0 (–2.1 to 19.5) |
| Excessive gestational weight gain | 12 (60%) | 6 (40%) |
| BMI at enrollment, kg/m 2 | 36.4 (25.4–45.2) a | 35.1 (25.5–51.8) |
| Obese at enrollment | 14 (70%) | 10 (66.7%) |
| Morbidly obese at enrollment | 4 (20%) | 5 (33.3%) |
| Rate of preterm birth <37 wk | 9 (45%) a | 0 (0%) |
| Preeclampsia | 5 (25%) | 0 (0%) |
| Cesarean delivery rate | 10 (50%) | 7 (46.7%) |
| Male sex | 8 (40%) | 7 (46.7%) |
| Birthweight, g | 3060 (2245–4050) | 3239 (2631–5135) |
| Birth length, cm | 50.0 (44.0–56.0) | 49.5 (47.0–53.5) |
| Birth head circumference, cm | 34.0 (32.0–36.2) | 34.0 (30.5–38.5) |
| Macrosomia rate (>4000 g) | 1 (5%) | 3 (20%) |
| Breast-feeding immediately postpartum | 16 (80%) | 10 (66.7%) |
Weight outcomes
Contrary to our hypothesis, median weight loss among all for whom the data were obtained was similar between groups (metformin, 6.3 kg [range, –0.3 to 19.8] vs placebo, 6.5 kg [range, –0.3 to 12.1], P = .99) ( Table 3 ). The groups were also similar in the percent of women achieving their prepregnancy weight (metformin, 41.7% vs placebo, 37.2%; RR, 0.80; 95% CI, 0.5–2.9; P = .686) and achieving their ideal body weight (metformin, 2.8% vs placebo, 2.3%; RR, 0.80; 95% CI, 0.7–19.9; P = .899).
